diff --git a/alembic/versions/030_add_tags_and_asset_tags.py b/alembic/versions/030_add_tags_and_asset_tags.py
new file mode 100644
index 000000000..338750833
--- /dev/null
+++ b/alembic/versions/030_add_tags_and_asset_tags.py
@@ -0,0 +1,68 @@
+"""Add tags and asset_tags tables
+
+Revision ID: 030
+Revises: 029
+Create Date: 2026-07-07
+
+新增标签表和素材-标签关联表，支持规范化多对多标签管理。
+"""
+
+import sqlalchemy as sa
+
+from alembic import op
+
+revision = "030"
+down_revision = "029"
+branch_labels = None
+depends_on = None
+
+
+def _table_exists(table: str) -> bool:
+    ctx = op.get_context()
+    if ctx.as_sql:
+        return False
+    conn = op.get_bind()
+    result = conn.execute(
+        sa.text("SELECT COUNT(*) FROM information_schema.tables WHERE table_name = :table"),
+        {"table": table},
+    )
+    return (result.scalar() or 0) > 0
+
+
+def upgrade() -> None:
+    if not _table_exists("tags"):
+        op.create_table(
+            "tags",
+            sa.Column("id", sa.String(36), primary_key=True),
+            sa.Column("user_id", sa.String(36), nullable=False),
+            sa.Column("name", sa.String(100), nullable=False),
+            sa.Column(
+                "created_at",
+                sa.DateTime(),
+                nullable=False,
+                server_default=sa.func.now(),
+            ),
+            sa.UniqueConstraint("user_id", "name", name="uq_tags_user_name"),
+        )
+        op.create_index("ix_tags_user_id", "tags", ["user_id"])
+
+    if not _table_exists("asset_tags"):
+        op.create_table(
+            "asset_tags",
+            sa.Column("asset_id", sa.String(36), primary_key=True),
+            sa.Column("tag_id", sa.String(36), primary_key=True),
+            sa.Column(
+                "created_at",
+                sa.DateTime(),
+                nullable=False,
+                server_default=sa.func.now(),
+            ),
+        )
+        op.create_index("ix_asset_tags_tag_id", "asset_tags", ["tag_id"])
+
+
+def downgrade() -> None:
+    op.drop_index("ix_asset_tags_tag_id", table_name="asset_tags")
+    op.drop_table("asset_tags")
+    op.drop_index("ix_tags_user_id", table_name="tags")
+    op.drop_table("tags")
diff --git a/apps/api/app/api/router.py b/apps/api/app/api/router.py
index 714181055..875833cda 100755
--- a/apps/api/app/api/router.py
+++ b/apps/api/app/api/router.py
@@ -16,6 +16,7 @@ from app.api.routes.jobs import router as jobs_router
 from app.api.routes.projects import router as projects_router
 from app.api.routes.recipes import router as recipes_router
 from app.api.routes.subscription import router as subscription_router
+from app.api.routes.tags import router as tags_router
 from app.api.routes.task_center import router as task_center_router
 from app.api.routes.templates import router as templates_router
 from app.api.routes.titles import router as titles_router
@@ -38,6 +39,11 @@ api_router.include_router(
     prefix="/projects",
     tags=["Project"],
 )
+api_router.include_router(
+    tags_router,
+    prefix="/tags",
+    tags=["Tag"],
+)
 api_router.include_router(
     task_center_router,
     tags=["TaskCenter"],
diff --git a/apps/api/app/api/routes/assets.py b/apps/api/app/api/routes/assets.py
index b7a4c2aef..f351699f7 100644
--- a/apps/api/app/api/routes/assets.py
+++ b/apps/api/app/api/routes/assets.py
@@ -7,14 +7,18 @@ from app.dependencies import (
     get_asset_library_repository,
     get_asset_repository,
     get_project_repository,
+    get_tag_repository,
 )
 from app.schemas.asset import (
     AssetResponse,
+    BatchDeleteRequest,
+    BatchDeleteResponse,
     CreateAssetRequest,
     ListAssetsResponse,
     UpdateAssetRequest,
     UpdateAssetReviewRequest,
 )
+from app.schemas.tag import TagAssetsRequest
 from fastapi import APIRouter, Depends, HTTPException, Query
 
 from packages.application import (
@@ -64,6 +68,7 @@ def _to_asset_response(item, storage_service=None) -> AssetResponse:
         classification_status=item.classification_status.value,
         quality_score=item.quality_score,
         uploaded_by_user_id=item.uploaded_by_user_id,
+        tag_ids=getattr(item, "tag_ids", []),
     )
 
 
@@ -84,6 +89,7 @@ def list_assets(
     keyword: Optional[str] = Query(None, description="按名称模糊匹配"),
     gender: Optional[str] = Query(None, description="按 metadata.gender 筛选"),
     style: Optional[str] = Query(None, description="按 metadata.style 筛选"),
+    tag_ids: Optional[str] = Query(None, description="按标签 ID 筛选（逗号分隔，取交集）"),
     skip: int = Query(0, ge=0),
     limit: int = Query(100, ge=1, le=500),
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -96,12 +102,19 @@ def list_assets(
     # kind → file_type 映射（voice 对应 audio）
     kind_to_file_type = {"video": "video", "voice": "audio", "image": "image"}
 
-    def _apply_filters(items):
-        """依次应用 kind / keyword / gender / style 过滤。"""
+    # 解析 tag_ids 参数（逗号分隔）
+    filter_tag_ids: list[str] | None = None
+    if tag_ids:
+        filter_tag_ids = [t.strip() for t in tag_ids.split(",") if t.strip()]
+        if not filter_tag_ids:
+            filter_tag_ids = None
+
+    # 需要内存过滤的标志（keyword/gender/style/tag_ids 无法在 DB 层过滤）
+    needs_memory_filter = bool(keyword or gender or style or filter_tag_ids)
+
+    def _apply_memory_filters(items):
+        """应用 keyword / gender / style / tag_ids 内存过滤。"""
         result = items
-        if kind:
-            ft = kind_to_file_type.get(kind)
-            result = [i for i in result if i.mime_type and i.mime_type.startswith(ft or "")]
         if keyword:
             kw = keyword.lower()
             result = [i for i in result if kw in (i.name or "").lower()]
@@ -109,9 +122,90 @@ def list_assets(
             result = [i for i in result if (i.metadata or {}).get("gender") == gender]
         if style:
             result = [i for i in result if (i.metadata or {}).get("style") == style]
+        if filter_tag_ids:
+            tag_set = set(filter_tag_ids)
+            result = [i for i in result if tag_set.issubset(set(getattr(i, "tag_ids", [])))]
         return result
 
-    # 模式1：指定 library_id → 返回该库的素材
+    # ── 优化路径：无内存过滤时，使用 DB 级分页 ──
+    if not needs_memory_filter:
+        ft = kind_to_file_type.get(kind) if kind else None
+
+        # 模式1：指定 library_id
+        if library_id:
+            library = asset_library_repository.get(library_id)
+            if library is None:
+                raise HTTPException(status_code=404, detail=f"AssetLibrary {library_id} not found")
+            _check_project_access(library.project_id, user_id, project_repository)
+            if ft:
+                items = asset_repository.find_by_library_and_file_type(library_id, ft, skip=skip, limit=limit)
+                total = asset_repository.count_by_project(library.project_id) if not kind else len(items)
+            else:
+                items = asset_repository.find_by_library(library_id, skip=skip, limit=limit)
+                total = asset_repository.count_by_project(library.project_id)
+            return ListAssetsResponse(
+                items=[_to_asset_response(item) for item in items],
+                total=total,
+                skip=skip,
+                limit=limit,
+            )
+
+        # 模式2：指定 project_id
+        if project_id:
+            _check_project_access(project_id, user_id, project_repository)
+            if ft:
+                # 无直接方法，加载后按 file_type 过滤（仍比全量加载好）
+                all_items = asset_repository.find_by_project(project_id)
+                items = [i for i in all_items if i.mime_type and i.mime_type.startswith(ft)]
+                total = len(items)
+                paged = items[skip : skip + limit]
+            else:
+                items = asset_repository.find_by_project(project_id, skip=skip, limit=limit)
+                total = asset_repository.count_by_project(project_id)
+                paged = items
+            return ListAssetsResponse(
+                items=[_to_asset_response(item) for item in paged],
+                total=total,
+                skip=skip,
+                limit=limit,
+            )
+
+        # 模式3：跨项目（无 library_id/project_id）
+        try:
+            projects = project_repository.find_accessible_projects(user_id)
+        except Exception:
+            logger.exception("查询用户可访问项目失败: user_id=%s", user_id)
+            return ListAssetsResponse(items=[], total=0, skip=skip, limit=limit)
+
+        project_ids = [p.id for p in projects]
+        if not project_ids:
+            return ListAssetsResponse(items=[], total=0, skip=skip, limit=limit)
+
+        total = asset_repository.count_by_project_ids(project_ids)
+        # 跨项目分页：逐项目累积直到凑够一页
+        paged_items: list = []
+        offset = skip
+        remaining = limit
+        for pid in project_ids:
+            proj_total = asset_repository.count_by_project(pid)
+            if offset >= proj_total:
+                offset -= proj_total
+                continue
+            proj_items = asset_repository.find_by_project(pid, skip=offset, limit=remaining)
+            paged_items.extend(proj_items)
+            remaining -= len(proj_items)
+            offset = 0
+            if remaining <= 0:
+                break
+
+        return ListAssetsResponse(
+            items=[_to_asset_response(item) for item in paged_items],
+            total=total,
+            skip=skip,
+            limit=limit,
+        )
+
+    # ── 内存过滤路径：有 keyword/gender/style 时，加载全量后内存过滤 ──
     if library_id:
         library = asset_library_repository.get(library_id)
         if library is None:
@@ -121,41 +215,24 @@ def list_assets(
             all_items = asset_repository.find_by_library_and_file_type(library_id, kind_to_file_type[kind])
         else:
             all_items = asset_repository.find_by_library(library_id)
-        filtered = _apply_filters(all_items)
-        total = len(filtered)
-        paged = filtered[skip : skip + limit]
-        return ListAssetsResponse(
-            items=[_to_asset_response(item) for item in paged],
-            total=total,
-            skip=skip,
-            limit=limit,
-        )
-
-    # 模式2：指定 project_id → 返回该项目所有素材
-    if project_id:
+    elif project_id:
         _check_project_access(project_id, user_id, project_repository)
         all_items = asset_repository.find_by_project(project_id)
-        filtered = _apply_filters(all_items)
-        total = len(filtered)
-        paged = filtered[skip : skip + limit]
-        return ListAssetsResponse(
-            items=[_to_asset_response(item) for item in paged],
-            total=total,
-            skip=skip,
-            limit=limit,
-        )
+    else:
+        try:
+            projects = project_repository.find_accessible_projects(user_id)
+        except Exception:
+            logger.exception("查询用户可访问项目失败: user_id=%s", user_id)
+            return ListAssetsResponse(items=[], total=0, skip=skip, limit=limit)
+        all_items = []
+        for proj in projects:
+            all_items.extend(asset_repository.find_by_project(proj.id))
 
-    # 模式3：都不传 → 返回用户可访问的所有项目的所有素材
-    try:
-        projects = project_repository.find_accessible_projects(user_id)
-    except Exception:
-        logger.exception("查询用户可访问项目失败: user_id=%s", user_id)
-        return ListAssetsResponse(items=[], total=0, skip=skip, limit=limit)
-
-    all_items = []
-    for proj in projects:
-        all_items.extend(asset_repository.find_by_project(proj.id))
-    filtered = _apply_filters(all_items)
+    # 应用 kind 过滤（如果有）+ keyword/gender/style
+    if kind:
+        ft = kind_to_file_type.get(kind)
+        all_items = [i for i in all_items if i.mime_type and i.mime_type.startswith(ft or "")]
+    filtered = _apply_memory_filters(all_items)
     total = len(filtered)
     paged = filtered[skip : skip + limit]
     return ListAssetsResponse(
@@ -191,6 +268,35 @@ def update_asset_review_status(
     return _to_asset_response(updated)
 
 
+@router.post("/batch-delete", response_model=BatchDeleteResponse)
+def batch_delete_assets(
+    request: BatchDeleteRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    asset_repository: Any = Depends(get_asset_repository),
+    project_repository: Any = Depends(get_project_repository),
+) -> BatchDeleteResponse:
+    """批量删除素材（配音素材等），需逐项校验项目权限。"""
+    user_id = authenticated_user.user.id
+    deleted_ids: list[str] = []
+    failed_ids: list[str] = []
+
+    for asset_id in request.ids:
+        item = asset_repository.find_by_id(asset_id)
+        if item is None:
+            failed_ids.append(asset_id)
+            continue
+        try:
+            _check_project_access(item.project_id, user_id, project_repository)
+            deleted_ids.append(asset_id)
+        except HTTPException:
+            failed_ids.append(asset_id)
+
+    if deleted_ids:
+        asset_repository.batch_delete(deleted_ids)
+
+    return BatchDeleteResponse(deleted_count=len(deleted_ids), failed_ids=failed_ids)
+
+
 @router.get("/{asset_id}", response_model=AssetResponse)
 def get_asset(
     asset_id: str,
@@ -244,6 +350,48 @@ def delete_asset(
     asset_repository.delete(asset_id)
 
 
+@router.post("/{asset_id}/tags", response_model=AssetResponse)
+def tag_asset(
+    asset_id: str,
+    request: TagAssetsRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    asset_repository: Any = Depends(get_asset_repository),
+    project_repository: Any = Depends(get_project_repository),
+    tag_repository: Any = Depends(get_tag_repository),
+) -> AssetResponse:
+    """给素材打标签。"""
+    item = asset_repository.find_by_id(asset_id)
+    if item is None:
+        raise HTTPException(status_code=404, detail=f"Asset {asset_id} not found")
+    _check_project_access(item.project_id, authenticated_user.user.id, project_repository)
+    for tag_id in request.tag_ids:
+        tag = tag_repository.get(tag_id)
+        if tag is None:
+            raise HTTPException(status_code=404, detail=f"Tag {tag_id} not found")
+        if tag.user_id != authenticated_user.user.id:
+            raise HTTPException(status_code=403, detail=f"无权使用标签 {tag_id}")
+        item.add_tag(tag_id)
+    updated = asset_repository.update(item)
+    return _to_asset_response(updated)
+
+
+@router.delete("/{asset_id}/tags/{tag_id}", status_code=204)
+def untag_asset(
+    asset_id: str,
+    tag_id: str,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    asset_repository: Any = Depends(get_asset_repository),
+    project_repository: Any = Depends(get_project_repository),
+) -> None:
+    """取消素材的标签。"""
+    item = asset_repository.find_by_id(asset_id)
+    if item is None:
+        raise HTTPException(status_code=404, detail=f"Asset {asset_id} not found")
+    _check_project_access(item.project_id, authenticated_user.user.id, project_repository)
+    item.remove_tag(tag_id)
+    asset_repository.update(item)
+
+
 @router.post("", response_model=AssetResponse)
 def create_asset(
     request: CreateAssetRequest,
diff --git a/apps/api/app/api/routes/tags.py b/apps/api/app/api/routes/tags.py
new file mode 100644
index 000000000..7de3e278b
--- /dev/null
+++ b/apps/api/app/api/routes/tags.py
@@ -0,0 +1,67 @@
+"""标签 CRUD 路由。"""
+
+import logging
+from typing import Any
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_tag_repository
+from app.schemas.tag import (
+    CreateTagRequest,
+    ListTagsResponse,
+    TagResponse,
+)
+from fastapi import APIRouter, Depends, HTTPException
+
+from packages.domain import Tag
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+@router.get("", response_model=ListTagsResponse)
+def list_tags(
+    skip: int = 0,
+    limit: int = 100,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    tag_repository: Any = Depends(get_tag_repository),
+) -> ListTagsResponse:
+    """列出当前用户的标签。"""
+    user_id = authenticated_user.user.id
+    items = tag_repository.list_by_user(user_id, skip=skip, limit=limit)
+    total = tag_repository.count_by_user(user_id)
+    return ListTagsResponse(
+        items=[TagResponse(id=t.id, name=t.name, created_at=t.created_at) for t in items],
+        total=total,
+    )
+
+
+@router.post("", response_model=TagResponse, status_code=201)
+def create_tag(
+    request: CreateTagRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    tag_repository: Any = Depends(get_tag_repository),
+) -> TagResponse:
+    """创建标签（同用户同名去重，返回 409）。"""
+    user_id = authenticated_user.user.id
+    existing = tag_repository.find_by_name(user_id, request.name)
+    if existing:
+        raise HTTPException(status_code=409, detail="标签名称已存在")
+    tag = Tag.create(user_id=user_id, name=request.name)
+    created = tag_repository.create(tag)
+    return TagResponse(id=created.id, name=created.name, created_at=created.created_at)
+
+
+@router.delete("/{tag_id}", status_code=204)
+def delete_tag(
+    tag_id: str,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    tag_repository: Any = Depends(get_tag_repository),
+) -> None:
+    """删除标签（同时清理素材关联）。"""
+    tag = tag_repository.get(tag_id)
+    if tag is None:
+        raise HTTPException(status_code=404, detail="标签不存在")
+    if tag.user_id != authenticated_user.user.id:
+        raise HTTPException(status_code=403, detail="无权删除该标签")
+    tag_repository.delete(tag_id)
diff --git a/apps/api/app/api/routes/tts.py b/apps/api/app/api/routes/tts.py
index 38ba5ccd1..7bb55d80b 100644
--- a/apps/api/app/api/routes/tts.py
+++ b/apps/api/app/api/routes/tts.py
@@ -9,22 +9,28 @@ from app.auth import AuthenticatedUser, get_current_user
 from app.dependencies import (
     get_cosyvoice_service,
     get_db_session,
+    get_user_repository,
     get_voice_clone_profile_repository,
+    get_voice_library_repository,
 )
 from app.schemas.tts import (
     ListTTSJobResponse,
+    SaveToLibraryRequest,
+    SaveToLibraryResponse,
     TTSJobResponse,
     TTSStatusResponse,
     TTSSynthesizeRequest,
     TTSSynthesizeResponse,
 )
-from fastapi import APIRouter, Depends, HTTPException, Query, Response, status
+from fastapi import APIRouter, Depends, HTTPException, Query, Response, WebSocket, WebSocketDisconnect, status
 from sqlalchemy.orm import Session
 
 from packages.adapters.sqlalchemy_impl.tts_job_repository import (
     SQLAlchemyTTSJobRepository,
 )
+from packages.adapters.sqlalchemy_impl.voice_library_repository import SQLAlchemyVoiceLibraryRepository
 from packages.application.cosyvoice_service import CosyVoiceService
+from packages.application.tts_job.streaming_service import TTSStreamingService
 from packages.application.tts_job.use_cases import (
     CreateTTSJobUseCase,
     DeleteTTSJobUseCase,
@@ -34,6 +40,12 @@ from packages.application.tts_job.use_cases import (
     TTSJobNotFoundError,
 )
 from packages.application.tts_job.workflow import TTSWorkflowService
+from packages.application.voice_library.commands import CreateVoiceLibraryCommand
+from packages.application.voice_library.use_cases import (
+    CreateVoiceLibraryUseCase,
+    QuotaExceededError,
+)
+from packages.ports.user_repository import UserRepository
 
 logger = logging.getLogger(__name__)
 
@@ -130,18 +142,25 @@ def synthesize(
 
     # 若任务处于 processing 状态（异步模式），触发 Celery 后台轮询
     if job.status.value == "processing":
-        task_id = (job.metadata or {}).get("cosyvoice_task_id", "")
-        if task_id:
-            try:
+        # 分段合成任务 vs 普通单段任务
+        segment_task_ids = (job.metadata or {}).get("segment_task_ids", [])
+        is_segment = len(segment_task_ids) > 0
+
+        try:
+            if is_segment:
+                from worker_app.tasks import process_tts_segment_synthesis
+
+                process_tts_segment_synthesis.delay(job.id)
+            else:
                 from worker_app.tasks import process_tts_synthesis
 
                 process_tts_synthesis.delay(job.id)
-            except Exception as e:
-                # Celery 调度失败，标记 job 为 failed
-                try:
-                    workflow.process_synthesis_failure(job.id, f"Celery 任务调度失败: {e}")
-                except Exception as inner_e:
-                    logger.error(f"Celery 调度后标记失败时出错: job_id={job.id}, error={inner_e}")
+        except Exception as e:
+            # Celery 调度失败，标记 job 为 failed
+            try:
+                workflow.process_synthesis_failure(job.id, f"Celery 任务调度失败: {e}")
+            except Exception as inner_e:
+                logger.error(f"Celery 调度后标记失败时出错: job_id={job.id}, error={inner_e}")
 
     return TTSSynthesizeResponse(
         job_id=job.id,
@@ -225,3 +244,127 @@ def delete_tts_job(
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found")
     return Response(status_code=204)
+
+
+@router.post(
+    "/jobs/{job_id}/save-to-library",
+    response_model=SaveToLibraryResponse,
+    status_code=status.HTTP_201_CREATED,
+)
+def save_tts_job_to_library(
+    job_id: str,
+    request: SaveToLibraryRequest = SaveToLibraryRequest(),
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    tts_repository: SQLAlchemyTTSJobRepository = Depends(_get_repository),
+    voice_library_repository: SQLAlchemyVoiceLibraryRepository = Depends(get_voice_library_repository),
+    user_repository: UserRepository = Depends(get_user_repository),
+) -> SaveToLibraryResponse:
+    """将已完成的 TTS 合成结果保存到配音库。
+
+    自动携带音色名、时长、语速等元信息。
+    """
+    user_id = authenticated_user.user.id
+
+    # 获取 TTS job
+    get_use_case = GetTTSJobUseCase(tts_repository)
+    try:
+        job = get_use_case.execute(job_id, user_id)
+    except TTSJobNotFoundError:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found")
+
+    # 校验已完成
+    if not job.is_completed:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="TTS job is not completed yet",
+        )
+
+    # 构建配音素材名称
+    name = request.name or f"TTS-{job.id[:8]}"
+
+    # 构建元信息
+    metadata_ = {
+        "source": "tts_job",
+        "tts_job_id": job.id,
+        "format": job.format,
+        "sample_rate": job.sample_rate,
+    }
+    if job.metadata:
+        # 保留原始 job 的有用元信息
+        for key in ("speed", "language"):
+            if key in job.metadata:
+                metadata_[key] = job.metadata[key]
+
+    # 获取用户套餐（用于配额检查）
+    user = user_repository.find_by_id(user_id)
+    plan_name = getattr(user, "subscription_plan", "free") if user else "free"
+
+    # 构建命令并执行
+    command = CreateVoiceLibraryCommand(
+        user_id=user_id,
+        name=name,
+        text=job.input_text,
+        voice_provider="cosyvoice",
+        voice_id=job.voice_id,
+        voice_name=job.voice_model or "",
+        audio_url=job.output_audio_url,
+        duration=job.duration,
+        file_size=job.file_size,
+        status="completed",
+        project_id=job.project_id or "",
+        tags=[],
+        metadata_=metadata_,
+    )
+
+    use_case = CreateVoiceLibraryUseCase(voice_library_repository)
+    try:
+        item = use_case.execute(command, plan_name=plan_name or "free")
+    except QuotaExceededError as exc:
+        raise HTTPException(
+            status_code=status.HTTP_429_TOO_MANY_REQUESTS,
+            detail=f"配音库配额已满（{exc.used}/{exc.limit}），请升级套餐",
+        )
+
+    return SaveToLibraryResponse(
+        id=item.id,
+        name=item.name,
+        audio_url=item.audio_url,
+        duration=item.duration,
+        voice_id=item.voice_id,
+        voice_name=item.voice_name,
+        status=item.status,
+    )
+
+
+@router.websocket("/ws/tts/stream")
+async def tts_websocket_stream(
+    websocket: WebSocket,
+    cosyvoice_service: CosyVoiceService = Depends(get_cosyvoice_service),
+) -> None:
+    """WebSocket 流式 TTS 合成。
+
+    协议：
+    1. 客户端发送 JSON 文本帧: {"text": "...", "voice_id": "...", ...}
+    2. 服务端发送 JSON 状态帧 + 二进制音频帧
+    3. 完成时发送 JSON 结束帧
+    """
+    await websocket.accept()
+    try:
+        message = await websocket.receive_json()
+        params = {
+            "text": message.get("text", ""),
+            "voice_id": message.get("voice_id", ""),
+            "sample_rate": message.get("sample_rate", 0),
+            "format": message.get("format", "mp3"),
+            "speed": message.get("speed", 1.0),
+        }
+        streaming_service = TTSStreamingService(cosyvoice_service)
+        await streaming_service.synthesize_and_stream(websocket, params)
+    except WebSocketDisconnect:
+        logger.info("WebSocket 客户端断开连接")
+    except Exception as e:
+        logger.error(f"WebSocket 流式合成异常: {e}", exc_info=True)
+        try:
+            await websocket.send_json({"type": "error", "message": f"服务异常: {e}"})
+        except Exception:
+            pass
diff --git a/apps/api/app/dependencies.py b/apps/api/app/dependencies.py
index 5d38872f7..9c42a59b3 100755
--- a/apps/api/app/dependencies.py
+++ b/apps/api/app/dependencies.py
@@ -39,6 +39,7 @@ from packages.adapters.sqlalchemy_impl.project_repository import (
     SQLAlchemyProjectRepository,
 )
 from packages.adapters.sqlalchemy_impl.session import build_session_factory
+from packages.adapters.sqlalchemy_impl.tag_repository import SQLAlchemyTagRepository
 from packages.adapters.sqlalchemy_impl.title_library_repository import (
     SQLAlchemyTitleLibraryRepository,
 )
@@ -58,6 +59,7 @@ from packages.ports.generation_task_repository import GenerationTaskRepository
 from packages.ports.ingest_job_repository import IngestJobRepository
 from packages.ports.job_repository import JobRepository
 from packages.ports.project_repository import ProjectRepository
+from packages.ports.tag_repository import TagRepository
 from packages.ports.title_library_repository import TitleLibraryRepository
 from packages.ports.user_repository import UserRepository
 from packages.ports.voice_clone_profile_repository import VoiceCloneProfileRepository
@@ -138,6 +140,13 @@ def get_project_repository(
     return SQLAlchemyProjectRepository(session)
 
 
+def get_tag_repository(
+    session: Session = Depends(get_db_session),
+) -> TagRepository:
+    """Provide the SQLAlchemy tag repository implementation."""
+    return SQLAlchemyTagRepository(session)
+
+
 def get_user_repository(
     session: Session = Depends(get_db_session),
 ) -> UserRepository:
diff --git a/apps/api/app/schemas/asset.py b/apps/api/app/schemas/asset.py
index 8b783b539..1c686772f 100644
--- a/apps/api/app/schemas/asset.py
+++ b/apps/api/app/schemas/asset.py
@@ -51,6 +51,20 @@ class AssetResponse(BaseModel):
     classification_status: str
     quality_score: float | None = None
     uploaded_by_user_id: str
+    tag_ids: list[str] = Field(default_factory=list)
+
+
+class BatchDeleteRequest(BaseModel):
+    """批量删除请求。"""
+
+    ids: list[str] = Field(..., min_length=1, max_length=100, description="要删除的素材 ID 列表")
+
+
+class BatchDeleteResponse(BaseModel):
+    """批量删除响应。"""
+
+    deleted_count: int = Field(..., ge=0, description="实际删除数量")
+    failed_ids: list[str] = Field(default_factory=list, description="删除失败的 ID 列表")
 
 
 class ListAssetsResponse(BaseModel):
diff --git a/apps/api/app/schemas/tag.py b/apps/api/app/schemas/tag.py
new file mode 100644
index 000000000..70c8e2eb5
--- /dev/null
+++ b/apps/api/app/schemas/tag.py
@@ -0,0 +1,24 @@
+"""标签相关 Schema。"""
+
+from datetime import datetime
+
+from pydantic import BaseModel, Field
+
+
+class CreateTagRequest(BaseModel):
+    name: str = Field(..., min_length=1, max_length=100)
+
+
+class TagResponse(BaseModel):
+    id: str
+    name: str
+    created_at: datetime
+
+
+class ListTagsResponse(BaseModel):
+    items: list[TagResponse]
+    total: int = Field(default=0, ge=0)
+
+
+class TagAssetsRequest(BaseModel):
+    tag_ids: list[str] = Field(..., min_length=1, max_length=50)
diff --git a/apps/api/app/schemas/tts.py b/apps/api/app/schemas/tts.py
index bfad312e1..d2d098be7 100644
--- a/apps/api/app/schemas/tts.py
+++ b/apps/api/app/schemas/tts.py
@@ -83,3 +83,21 @@ class ListTTSJobResponse(BaseModel):
     total: int
     page: int
     page_size: int
+
+
+class SaveToLibraryRequest(BaseModel):
+    """保存到配音库请求。"""
+
+    name: Optional[str] = Field(None, description="配音素材名称，留空则自动生成")
+
+
+class SaveToLibraryResponse(BaseModel):
+    """保存到配音库响应。"""
+
+    id: str
+    name: str
+    audio_url: str
+    duration: float
+    voice_id: str
+    voice_name: str
+    status: str
diff --git a/apps/web/src/api/assets.ts b/apps/web/src/api/assets.ts
index 964f82bd7..856652f85 100644
--- a/apps/web/src/api/assets.ts
+++ b/apps/web/src/api/assets.ts
@@ -19,6 +19,7 @@ export interface AssetItem {
   status?: string;
   classification_status?: string | null;
   quality_score?: number | null;
+  tag_ids?: string[];
   created_at?: string;
 }
 
@@ -144,12 +145,18 @@ export const getAssets = async (libraryId: string): Promise<AssetItem[]> => {
 /** 按类型获取素材（如 voice/video/image），支持可选筛选 */
 export const getAssetsByKind = async (
   kind: string,
-  filters?: { keyword?: string; gender?: string; style?: string },
+  filters?: {
+    keyword?: string;
+    gender?: string;
+    style?: string;
+    tag_ids?: string[];
+  },
 ): Promise<AssetItem[]> => {
   const params: Record<string, string> = { kind };
   if (filters?.keyword) params.keyword = filters.keyword;
   if (filters?.gender) params.gender = filters.gender;
   if (filters?.style) params.style = filters.style;
+  if (filters?.tag_ids?.length) params.tag_ids = filters.tag_ids.join(",");
   const response = await apiClient.get("/assets", { params });
   return response.data.items || [];
 };
@@ -233,10 +240,11 @@ export const completeDirectUpload = async (data: {
   return response.data;
 };
 
-/** 直传上传（大文件推荐） */
+/** 直传上传（大文件推荐），支持可选进度回调 */
 export const uploadAssetDirect = async (data: {
   file: File;
   library_id: string;
+  onProgress?: (percent: number) => void;
 }): Promise<{ storage_key: string; ingest_job_id: string }> => {
   // 后端要求 project_id，前端自动获取默认项目
   const project = await getOrCreateDefaultProject();
@@ -255,13 +263,25 @@ export const uploadAssetDirect = async (data: {
   );
   directForm.append("file", data.file);
 
-  const uploadResponse = await fetch(prepared.upload_url, {
-    method: prepared.method,
-    body: directForm,
+  // 使用 XMLHttpRequest 以获取上传进度（fetch 不支持）
+  await new Promise<void>((resolve, reject) => {
+    const xhr = new XMLHttpRequest();
+    xhr.open(prepared.method, prepared.upload_url);
+    xhr.upload.onprogress = (e) => {
+      if (e.lengthComputable && data.onProgress) {
+        data.onProgress(Math.round((e.loaded / e.total) * 100));
+      }
+    };
+    xhr.onload = () => {
+      if (xhr.status >= 200 && xhr.status < 300) {
+        resolve();
+      } else {
+        reject(new Error(`OSS direct upload failed: ${xhr.status}`));
+      }
+    };
+    xhr.onerror = () => reject(new Error("OSS direct upload failed"));
+    xhr.send(directForm);
   });
-  if (!uploadResponse.ok) {
-    throw new Error(`OSS direct upload failed: ${uploadResponse.status}`);
-  }
 
   return completeDirectUpload({
     project_id: project.id,
diff --git a/apps/web/src/api/tags.ts b/apps/web/src/api/tags.ts
new file mode 100644
index 000000000..7a825ef12
--- /dev/null
+++ b/apps/web/src/api/tags.ts
@@ -0,0 +1,46 @@
+/**
+ * 标签 CRUD API
+ * P3 标签体系：对接后端标签表
+ */
+import apiClient from "./client";
+
+export interface TagItem {
+  id: string;
+  name: string;
+  created_at?: string;
+  usage_count?: number;
+}
+
+/** 获取当前用户所有标签 */
+export const getTags = async (): Promise<TagItem[]> => {
+  const response = await apiClient.get("/tags");
+  return response.data.items || [];
+};
+
+/** 创建标签（同名返回 409） */
+export const createTag = async (name: string): Promise<TagItem> => {
+  const response = await apiClient.post("/tags", { name });
+  return response.data;
+};
+
+/** 删除标签（同时清理素材关联） */
+export const deleteTag = async (tagId: string): Promise<void> => {
+  await apiClient.delete(`/tags/${tagId}`);
+};
+
+/** 为素材添加标签（最多 50 个） */
+export const tagAsset = async (
+  assetId: string,
+  tagIds: string[],
+): Promise<void> => {
+  if (tagIds.length === 0) return;
+  await apiClient.post(`/assets/${assetId}/tags`, { tag_ids: tagIds });
+};
+
+/** 移除素材的某个标签 */
+export const untagAsset = async (
+  assetId: string,
+  tagId: string,
+): Promise<void> => {
+  await apiClient.delete(`/assets/${assetId}/tags/${tagId}`);
+};
diff --git a/apps/web/src/api/tts.ts b/apps/web/src/api/tts.ts
index d060e115e..05b028def 100644
--- a/apps/web/src/api/tts.ts
+++ b/apps/web/src/api/tts.ts
@@ -122,6 +122,20 @@ export const getTTSJobs = async (
   return response.data;
 };
 
+/** 存为素材请求参数 */
+export interface SaveTtsToLibraryRequest {
+  name?: string;
+  tag_ids?: string[];
+}
+
+/** 将 TTS 合成结果保存到配音素材库 */
+export const saveTtsToLibrary = async (
+  jobId: string,
+  data?: SaveTtsToLibraryRequest,
+): Promise<void> => {
+  await apiClient.post(`/tts/jobs/${jobId}/save-to-library`, data ?? {});
+};
+
 /** 删除 TTS 任务 */
 export const deleteTTSJob = async (jobId: string): Promise<void> => {
   await apiClient.delete(`/tts/jobs/${jobId}`);
diff --git a/apps/web/src/pages/editing-planner/EditingPlanner.css b/apps/web/src/pages/editing-planner/EditingPlanner.css
index 2df388f17..0dbb3ee97 100644
--- a/apps/web/src/pages/editing-planner/EditingPlanner.css
+++ b/apps/web/src/pages/editing-planner/EditingPlanner.css
@@ -1968,7 +1968,9 @@
   border: 1px dashed var(--primary, #6366f1);
   border-radius: 6px;
   cursor: pointer;
-  transition: background 0.15s, color 0.15s;
+  transition:
+    background 0.15s,
+    color 0.15s;
 }
 
 .ep-voice-upload-btn:hover {
@@ -1996,7 +1998,9 @@
   border: none;
   border-radius: 4px;
   cursor: pointer;
-  transition: background 0.15s, color 0.15s;
+  transition:
+    background 0.15s,
+    color 0.15s;
 }
 
 .ep-voice-refresh-btn:hover {
@@ -2035,7 +2039,9 @@
   border: 1px solid var(--border-color, #334155);
   border-radius: 6px;
   cursor: pointer;
-  transition: background 0.15s, border-color 0.15s;
+  transition:
+    background 0.15s,
+    border-color 0.15s;
 }
 
 .ep-voice-preview-btn:hover {
diff --git a/apps/web/src/pages/generate/GeneratePage.tsx b/apps/web/src/pages/generate/GeneratePage.tsx
index 90c06365a..6e9dc0342 100644
--- a/apps/web/src/pages/generate/GeneratePage.tsx
+++ b/apps/web/src/pages/generate/GeneratePage.tsx
@@ -15,6 +15,9 @@ import {
   PauseCircleOutlined,
   DownloadOutlined,
   ShareAltOutlined,
+  SaveOutlined,
+  PlusOutlined,
+  CloseOutlined,
 } from "@ant-design/icons";
 import type { AssetItem } from "@/api/assets";
 import { getAssets, getAssetLibraries } from "@/api/assets";
@@ -28,9 +31,10 @@ import type { PresetVoiceItem } from "@/api/voices";
 import { formatDuration } from "@/api/voiceClone";
 import type { VoiceClone } from "@/api/voiceClone";
 import CloneModal from "@/components/voice/CloneModal";
-import { synthesizeSpeech, getTTSJobStatus } from "@/api/tts";
+import { synthesizeSpeech, getTTSJobStatus, saveTtsToLibrary } from "@/api/tts";
+import { getTags, createTag } from "@/api/tags";
 import { useCloneProgress } from "@/hooks/useCloneProgress";
-import { useSearchParams } from "react-router-dom";
+import { useSearchParams, useNavigate } from "react-router-dom";
 import { getEditPlan } from "@/api/editPlans";
 import "./generate.css";
 
@@ -97,6 +101,8 @@ const STEPS = [
    ================================================================ */
 
 const GeneratePage: React.FC = () => {
+  const navigate = useNavigate();
+
   /* ── 步骤状态 ── */
   const [currentStep, setCurrentStep] = useState(1);
 
@@ -225,6 +231,23 @@ const GeneratePage: React.FC = () => {
   const [customAudioUrl, setCustomAudioUrl] = useState<string | null>(null);
   const [ttsError, setTtsError] = useState<string | null>(null);
   const [ttsJobId, setTtsJobId] = useState<string | null>(null);
+  /** 合成完成后保留的 job ID，用于"存为素材" */
+  const [completedTtsJobId, setCompletedTtsJobId] = useState<string | null>(
+    null,
+  );
+
+  /* ── 存为素材弹窗状态 ── */
+  const [saveModalOpen, setSaveModalOpen] = useState(false);
+  const [saveName, setSaveName] = useState("");
+  const [saveTagIds, setSaveTagIds] = useState<string[]>([]);
+  const [saveNewTag, setSaveNewTag] = useState("");
+
+  /* ── 标签列表（用于存为素材弹窗） ── */
+  const { data: allTags = [] } = useQuery({
+    queryKey: ["generate-save-tags"],
+    queryFn: getTags,
+    staleTime: 30_000,
+  });
 
   /* ── 素材数据 API ── */
   const { data: libraries = [] } = useQuery({
@@ -310,6 +333,7 @@ const GeneratePage: React.FC = () => {
         if (cancelled) return;
         if (status.status === "completed") {
           setCustomAudioUrl(status.output_audio_url);
+          setCompletedTtsJobId(ttsJobId);
           setTtsJobId(null);
           setTtsError(null);
           message.success("语音合成完成！");
@@ -351,6 +375,86 @@ const GeneratePage: React.FC = () => {
     });
   }, [customVoiceText, selectedVoice, synthesizeMutation]);
 
+  /* ── 存为素材 mutation ── */
+  const saveToLibraryMutation = useMutation({
+    mutationFn: (params: { name?: string; tag_ids?: string[] }) =>
+      saveTtsToLibrary(completedTtsJobId!, params),
+    onSuccess: () => {
+      message.success({
+        content: (
+          <span>
+            已保存到配音素材库！{" "}
+            <a
+              onClick={handleGoToLibrary}
+              style={{
+                color: "var(--primary-500, #6366f1)",
+                cursor: "pointer",
+              }}
+            >
+              去素材库查看
+            </a>
+          </span>
+        ),
+        duration: 5,
+      });
+      setSaveModalOpen(false);
+      setSaveName("");
+      setSaveTagIds([]);
+      setSaveNewTag("");
+      setCompletedTtsJobId(null);
+      setCustomAudioUrl(null);
+    },
+    onError: (err: Error) => {
+      message.error(`保存失败：${err.message || "请重试"}`);
+    },
+  });
+
+  /** 打开存为素材弹窗 */
+  const handleOpenSaveModal = useCallback(() => {
+    setSaveName("");
+    setSaveTagIds([]);
+    setSaveNewTag("");
+    setSaveModalOpen(true);
+  }, []);
+
+  /** 确认保存 */
+  const handleConfirmSave = useCallback(() => {
+    if (!completedTtsJobId) return;
+    saveToLibraryMutation.mutate({
+      name: saveName.trim() || undefined,
+      tag_ids: saveTagIds.length > 0 ? saveTagIds : undefined,
+    });
+  }, [completedTtsJobId, saveName, saveTagIds, saveToLibraryMutation]);
+
+  /** 在弹窗中新增标签（先创建再选中） */
+  const handleAddTagInModal = useCallback(
+    async (tagName: string) => {
+      const trimmed = tagName.trim();
+      if (!trimmed) return;
+      /* 已在选中列表则跳过 */
+      const existing = allTags.find((t) => t.name === trimmed);
+      if (existing) {
+        if (!saveTagIds.includes(existing.id)) {
+          setSaveTagIds((prev) => [...prev, existing.id]);
+        }
+        return;
+      }
+      try {
+        const created = await createTag(trimmed);
+        setSaveTagIds((prev) => [...prev, created.id]);
+        setSaveNewTag("");
+      } catch {
+        message.error(`创建标签"${trimmed}"失败`);
+      }
+    },
+    [allTags, saveTagIds],
+  );
+
+  /** 保存成功后跳转到素材库 */
+  const handleGoToLibrary = useCallback(() => {
+    navigate("/app/voice-materials");
+  }, [navigate]);
+
   const handleGenerate = useCallback(async () => {
     if (!title.trim()) {
       message.warning("请先选择或输入标题");
@@ -910,16 +1014,136 @@ const GeneratePage: React.FC = () => {
               {ttsError}
             </Text>
           )}
-          {customAudioUrl && (
-            <Text
+          {customAudioUrl && completedTtsJobId && (
+            <div
               style={{
-                color: "var(--success, #10b981)",
                 marginTop: 8,
-                display: "block",
+                display: "flex",
+                alignItems: "center",
+                gap: 12,
               }}
             >
-              ✓ 语音合成完成
-            </Text>
+              <Text style={{ color: "var(--success, #10b981)" }}>
+                ✓ 语音合成完成
+              </Text>
+              <button
+                className="xx-btn xx-btn-primary"
+                style={{ height: 30, padding: "0 14px", fontSize: 12 }}
+                onClick={handleOpenSaveModal}
+              >
+                <SaveOutlined /> 存为素材
+              </button>
+            </div>
+          )}
+
+          {/* ── 存为素材弹窗 ── */}
+          {saveModalOpen && (
+            <div
+              className="xx-save-modal-overlay"
+              onClick={() => setSaveModalOpen(false)}
+            >
+              <div
+                className="xx-save-modal"
+                onClick={(e) => e.stopPropagation()}
+              >
+                <div className="xx-save-modal-header">
+                  <span>保存到配音素材库</span>
+                  <button
+                    className="xx-save-modal-close"
+                    onClick={() => setSaveModalOpen(false)}
+                  >
+                    <CloseOutlined />
+                  </button>
+                </div>
+                <div className="xx-save-modal-body">
+                  <label className="xx-save-modal-label">素材名称</label>
+                  <input
+                    className="xx-save-modal-input"
+                    placeholder="留空则自动生成名称"
+                    value={saveName}
+                    onChange={(e) => setSaveName(e.target.value)}
+                    maxLength={50}
+                  />
+                  <label className="xx-save-modal-label">
+                    标签
+                    <span
+                      style={{
+                        fontWeight: 400,
+                        color: "var(--text-tertiary, #94a3b8)",
+                      }}
+                    >
+                      （可选）
+                    </span>
+                  </label>
+                  <div className="xx-save-modal-tags">
+                    {saveTagIds.map((id) => {
+                      const tag = allTags.find((t) => t.id === id);
+                      return tag ? (
+                        <span key={id} className="xx-save-modal-tag active">
+                          {tag.name}
+                          <CloseOutlined
+                            className="xx-save-modal-tag-remove"
+                            onClick={() =>
+                              setSaveTagIds((prev) =>
+                                prev.filter((x) => x !== id),
+                              )
+                            }
+                          />
+                        </span>
+                      ) : null;
+                    })}
+                    <input
+                      className="xx-save-modal-tag-input"
+                      placeholder="输入标签名回车添加"
+                      value={saveNewTag}
+                      onChange={(e) => setSaveNewTag(e.target.value)}
+                      onKeyDown={(e) => {
+                        if (e.key === "Enter") {
+                          e.preventDefault();
+                          handleAddTagInModal(saveNewTag);
+                        }
+                      }}
+                    />
+                  </div>
+                  {allTags.length > 0 && (
+                    <div className="xx-save-modal-tag-presets">
+                      {allTags
+                        .filter((t) => !saveTagIds.includes(t.id))
+                        .slice(0, 12)
+                        .map((t) => (
+                          <button
+                            key={t.id}
+                            className="xx-save-modal-tag-preset"
+                            onClick={() =>
+                              setSaveTagIds((prev) => [...prev, t.id])
+                            }
+                          >
+                            {t.name}
+                            <PlusOutlined
+                              style={{ fontSize: 10, marginLeft: 4 }}
+                            />
+                          </button>
+                        ))}
+                    </div>
+                  )}
+                </div>
+                <div className="xx-save-modal-footer">
+                  <button
+                    className="xx-btn xx-btn-ghost"
+                    onClick={() => setSaveModalOpen(false)}
+                  >
+                    取消
+                  </button>
+                  <button
+                    className="xx-btn xx-btn-primary"
+                    disabled={saveToLibraryMutation.isPending}
+                    onClick={handleConfirmSave}
+                  >
+                    {saveToLibraryMutation.isPending ? "保存中…" : "保存"}
+                  </button>
+                </div>
+              </div>
+            </div>
           )}
         </div>
       )}
diff --git a/apps/web/src/pages/generate/generate.css b/apps/web/src/pages/generate/generate.css
index 76331e052..82052f16b 100644
--- a/apps/web/src/pages/generate/generate.css
+++ b/apps/web/src/pages/generate/generate.css
@@ -883,3 +883,194 @@
     flex-direction: column;
   }
 }
+
+/* ── 存为素材弹窗 ── */
+.xx-save-modal-overlay {
+  position: fixed;
+  inset: 0;
+  background: rgba(0, 0, 0, 0.45);
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  z-index: 1000;
+  animation: xxFadeIn 0.15s ease;
+}
+
+.xx-save-modal {
+  background: var(--bg-card, #fff);
+  border-radius: var(--radius-lg, 16px);
+  width: 420px;
+  max-width: 90vw;
+  box-shadow: 0 20px 60px rgba(0, 0, 0, 0.15);
+  animation: xxSlideUp 0.2s ease;
+}
+
+.xx-save-modal-header {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  padding: 16px 20px;
+  border-bottom: 1px solid var(--border-light, #f1f5f9);
+  font-weight: 600;
+  font-size: 15px;
+  color: var(--text-primary, #0f172a);
+}
+
+.xx-save-modal-close {
+  background: none;
+  border: none;
+  cursor: pointer;
+  color: var(--text-tertiary, #94a3b8);
+  font-size: 14px;
+  padding: 4px;
+  border-radius: var(--radius-sm, 6px);
+  transition: all 0.15s;
+}
+
+.xx-save-modal-close:hover {
+  background: var(--bg-hover, #f8fafc);
+  color: var(--text-primary, #0f172a);
+}
+
+.xx-save-modal-body {
+  padding: 20px;
+}
+
+.xx-save-modal-label {
+  display: block;
+  font-size: 13px;
+  font-weight: 500;
+  color: var(--text-secondary, #475569);
+  margin-bottom: 6px;
+  margin-top: 14px;
+}
+
+.xx-save-modal-label:first-child {
+  margin-top: 0;
+}
+
+.xx-save-modal-input {
+  width: 100%;
+  padding: 8px 12px;
+  border: 1px solid var(--border-color, #e2e8f0);
+  border-radius: var(--radius-sm, 10px);
+  font-size: 14px;
+  outline: none;
+  background: var(--bg-surface, #fff);
+  color: var(--text-primary, #0f172a);
+  transition: border-color 0.15s;
+}
+
+.xx-save-modal-input:focus {
+  border-color: var(--primary-500, #6366f1);
+  box-shadow: 0 0 0 2px var(--primary-100, #e0e7ff);
+}
+
+.xx-save-modal-tags {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 6px;
+  padding: 8px;
+  border: 1px solid var(--border-color, #e2e8f0);
+  border-radius: var(--radius-sm, 10px);
+  min-height: 40px;
+  align-items: center;
+  cursor: text;
+  transition: border-color 0.15s;
+}
+
+.xx-save-modal-tags:focus-within {
+  border-color: var(--primary-500, #6366f1);
+  box-shadow: 0 0 0 2px var(--primary-100, #e0e7ff);
+}
+
+.xx-save-modal-tag {
+  display: inline-flex;
+  align-items: center;
+  gap: 4px;
+  padding: 2px 8px;
+  font-size: 12px;
+  border-radius: 12px;
+  background: var(--primary-50, #eef2ff);
+  color: var(--primary-600, #4f46e5);
+  border: 1px solid var(--primary-200, #c7d2fe);
+}
+
+.xx-save-modal-tag-remove {
+  font-size: 10px;
+  cursor: pointer;
+  opacity: 0.6;
+  transition: opacity 0.15s;
+}
+
+.xx-save-modal-tag-remove:hover {
+  opacity: 1;
+}
+
+.xx-save-modal-tag-input {
+  border: none;
+  outline: none;
+  flex: 1;
+  min-width: 100px;
+  font-size: 13px;
+  background: transparent;
+  color: var(--text-primary, #0f172a);
+}
+
+.xx-save-modal-tag-input::placeholder {
+  color: var(--text-tertiary, #94a3b8);
+}
+
+.xx-save-modal-tag-presets {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 6px;
+  margin-top: 10px;
+  padding-top: 10px;
+  border-top: 1px solid var(--border-light, #f1f5f9);
+}
+
+.xx-save-modal-tag-preset {
+  padding: 3px 10px;
+  font-size: 12px;
+  border: 1px solid var(--border-color, #e2e8f0);
+  border-radius: 12px;
+  background: var(--bg-surface, #fff);
+  color: var(--text-secondary, #475569);
+  cursor: pointer;
+  transition: all 0.15s;
+}
+
+.xx-save-modal-tag-preset:hover {
+  border-color: var(--primary-300, #a5b4fc);
+  color: var(--primary-600, #4f46e5);
+  background: var(--primary-50, #eef2ff);
+}
+
+.xx-save-modal-footer {
+  display: flex;
+  justify-content: flex-end;
+  gap: 8px;
+  padding: 14px 20px;
+  border-top: 1px solid var(--border-light, #f1f5f9);
+}
+
+@keyframes xxFadeIn {
+  from {
+    opacity: 0;
+  }
+  to {
+    opacity: 1;
+  }
+}
+
+@keyframes xxSlideUp {
+  from {
+    transform: translateY(12px);
+    opacity: 0;
+  }
+  to {
+    transform: translateY(0);
+    opacity: 1;
+  }
+}
diff --git a/apps/web/src/pages/voice-materials/VoiceMaterialLibrary.tsx b/apps/web/src/pages/voice-materials/VoiceMaterialLibrary.tsx
index e232889e5..92ab18f08 100644
--- a/apps/web/src/pages/voice-materials/VoiceMaterialLibrary.tsx
+++ b/apps/web/src/pages/voice-materials/VoiceMaterialLibrary.tsx
@@ -32,8 +32,12 @@ import {
   UserOutlined,
   ManOutlined,
   WomanOutlined,
+  CheckOutlined,
+  TagsOutlined,
+  MutedOutlined,
 } from "@ant-design/icons";
 import { Button, Input, Select, Modal, Tag } from "@/components/ui";
+import { message, Popover, Popconfirm, Tooltip } from "antd";
 import PageHead from "@/components/layout/PageHead";
 import {
   getAssetsByKind,
@@ -45,6 +49,13 @@ import {
   createAssetLibrary,
   type AssetItem,
 } from "@/api/assets";
+import {
+  type TagItem,
+  getTags,
+  createTag,
+  tagAsset,
+  untagAsset,
+} from "@/api/tags";
 import "./voice-materials.css";
 
 /* ============================================================
@@ -54,19 +65,16 @@ import "./voice-materials.css";
 type VoiceGender = "male" | "female" | "child" | "neutral";
 type ViewMode = "card" | "list";
 
-/** 风格标签预设 */
-const STYLE_TAGS = [
-  "活力",
-  "沉稳",
-  "温柔",
-  "专业",
-  "甜美",
-  "磁性",
-  "清新",
-  "激昂",
-  "幽默",
-  "庄重",
-];
+/** 标签溢出限制 */
+const MAX_CARD_TAGS = 3;
+const MAX_ROW_TAGS = 2;
+const TAG_VARIANTS = [
+  "info",
+  "primary",
+  "success",
+  "warning",
+  "error",
+] as const;
 
 /** 性别选项 */
 const GENDER_OPTIONS: {
@@ -86,7 +94,7 @@ interface VoiceMaterial {
   name: string;
   description: string;
   gender: VoiceGender;
-  tags: string[];
+  tagIds: string[];
   fileName: string;
   fileSize: number;
   duration: number;
@@ -102,18 +110,13 @@ interface VoiceMaterial {
 /** 后端 AssetItem → 前端 VoiceMaterial */
 const mapAssetToMaterial = (asset: AssetItem): VoiceMaterial => {
   const meta = asset.metadata || {};
-  const tags = Array.isArray(meta.style)
-    ? (meta.style as string[])
-    : typeof meta.style === "string" && meta.style
-      ? (meta.style as string).split(",").map((s) => s.trim())
-      : [];
 
   return {
     id: asset.id,
     name: asset.name,
     description: (meta.description as string) || "",
     gender: (meta.gender as VoiceGender) || "neutral",
-    tags,
+    tagIds: Array.isArray(asset.tag_ids) ? asset.tag_ids : [],
     fileName: asset.storage_key?.split("/").pop() || asset.name,
     fileSize: asset.file_size || 0,
     duration: (meta.duration as number) || 0,
@@ -123,16 +126,14 @@ const mapAssetToMaterial = (asset: AssetItem): VoiceMaterial => {
   };
 };
 
-/** 前端表单数据 → 后端 metadata */
+/** 前端表单数据 → 后端 metadata（标签走独立 API，不再写 metadata.style） */
 const buildMetadata = (data: {
   gender: VoiceGender;
   description: string;
-  tags: string[];
   duration?: number;
 }): Record<string, unknown> => ({
   gender: data.gender,
   description: data.description,
-  style: data.tags,
   duration: data.duration || 0,
 });
 
@@ -184,6 +185,180 @@ const getAudioDuration = (file: File): Promise<number> => {
   });
 };
 
+/* ============================================================
+ * 标签选择器组件（支持自定义新增 + 移除）
+ * ============================================================ */
+
+interface TagSelectorProps {
+  /** 已选标签 ID 列表 */
+  value: string[];
+  onChange: (tagIds: string[]) => void;
+  /** 所有可用标签（来自 API） */
+  tags: TagItem[];
+  /** 标签 ID → TagItem 映射 */
+  tagMap: Map<string, TagItem>;
+  /** 创建新标签，返回带 ID 的 TagItem */
+  onCreateTag: (name: string) => Promise<TagItem>;
+  placeholder?: string;
+}
+
+const TagSelector: React.FC<TagSelectorProps> = ({
+  value,
+  onChange,
+  tags,
+  tagMap,
+  onCreateTag,
+  placeholder = "输入标签后回车添加",
+}) => {
+  const [inputVal, setInputVal] = useState("");
+  const [showSuggestions, setShowSuggestions] = useState(false);
+  const inputRef = useRef<HTMLInputElement>(null);
+
+  /** 按名称查找已有标签（大小写不敏感） */
+  const findTagByName = useCallback(
+    (name: string) =>
+      tags.find((t) => t.name.toLowerCase() === name.toLowerCase()),
+    [tags],
+  );
+
+  /** 去重添加标签（按 ID） */
+  const addTagId = useCallback(
+    (tagId: string) => {
+      if (value.includes(tagId)) return;
+      onChange([...value, tagId]);
+      setInputVal("");
+      setShowSuggestions(false);
+    },
+    [value, onChange],
+  );
+
+  /** 输入自定义标签名：若已存在则直接选，否则创建新标签 */
+  const addTagByName = useCallback(
+    async (name: string) => {
+      const trimmed = name.trim();
+      if (!trimmed) return;
+      const existing = findTagByName(trimmed);
+      if (existing) {
+        addTagId(existing.id);
+      } else {
+        try {
+          const created = await onCreateTag(trimmed);
+          addTagId(created.id);
+        } catch {
+          /* 创建失败静默忽略 */
+        }
+      }
+    },
+    [findTagByName, addTagId, onCreateTag],
+  );
+
+  const removeTagId = useCallback(
+    (tagId: string) => {
+      onChange(value.filter((t) => t !== tagId));
+    },
+    [value, onChange],
+  );
+
+  /** 输入补全建议（排除已选） */
+  const suggestions = useMemo(() => {
+    if (!inputVal.trim()) return [];
+    const lower = inputVal.toLowerCase();
+    return tags.filter(
+      (t) => t.name.toLowerCase().includes(lower) && !value.includes(t.id),
+    );
+  }, [inputVal, tags, value]);
+
+  const handleKeyDown = (e: React.KeyboardEvent) => {
+    if (e.key === "Enter") {
+      e.preventDefault();
+      if (suggestions.length > 0) {
+        addTagId(suggestions[0].id);
+      } else {
+        addTagByName(inputVal);
+      }
+    } else if (e.key === "Backspace" && !inputVal && value.length > 0) {
+      removeTagId(value[value.length - 1]);
+    }
+  };
+
+  return (
+    <div className="vmat-tag-selector-wrapper">
+      <div
+        className="vmat-tag-selector"
+        onClick={() => inputRef.current?.focus()}
+      >
+        {value.map((tagId) => (
+          <Tag
+            key={tagId}
+            variant="info"
+            closable
+            onClose={() => removeTagId(tagId)}
+          >
+            {tagMap.get(tagId)?.name ?? tagId}
+          </Tag>
+        ))}
+        <input
+          ref={inputRef}
+          className="vmat-tag-selector-input"
+          value={inputVal}
+          onChange={(e) => {
+            setInputVal(e.target.value);
+            setShowSuggestions(true);
+          }}
+          onFocus={() => setShowSuggestions(true)}
+          onBlur={() => setTimeout(() => setShowSuggestions(false), 150)}
+          onKeyDown={handleKeyDown}
+          placeholder={value.length === 0 ? placeholder : ""}
+        />
+      </div>
+
+      {/* 自动补全下拉 */}
+      {showSuggestions && suggestions.length > 0 && (
+        <div className="vmat-tag-suggestions">
+          {suggestions.slice(0, 6).map((tag) => (
+            <button
+              key={tag.id}
+              type="button"
+              className="vmat-tag-suggestion-item"
+              onMouseDown={(e) => {
+                e.preventDefault();
+                addTagId(tag.id);
+              }}
+            >
+              {tag.name}
+            </button>
+          ))}
+        </div>
+      )}
+
+      {/* 已有标签快捷选择 */}
+      {tags.length > 0 && (
+        <div className="vmat-tag-selector-presets">
+          {tags.map((tag) => {
+            const isSelected = value.includes(tag.id);
+            return (
+              <button
+                key={tag.id}
+                type="button"
+                className={`vmat-tag-selector-preset${isSelected ? " selected" : ""}`}
+                onClick={() => {
+                  if (isSelected) removeTagId(tag.id);
+                  else addTagId(tag.id);
+                }}
+              >
+                {isSelected && (
+                  <CheckOutlined style={{ fontSize: 10, marginRight: 2 }} />
+                )}
+                {tag.name}
+              </button>
+            );
+          })}
+        </div>
+      )}
+    </div>
+  );
+};
+
 /* ============================================================
  * 上传 / 编辑 表单
  * ============================================================ */
@@ -195,6 +370,10 @@ interface MaterialFormProps {
   ) => void;
   onCancel: () => void;
   loading?: boolean;
+  uploadProgress?: number | null;
+  tags?: TagItem[];
+  tagMap?: Map<string, TagItem>;
+  onCreateTag?: (name: string) => Promise<TagItem>;
 }
 
 const MaterialForm: React.FC<MaterialFormProps> = ({
@@ -202,24 +381,22 @@ const MaterialForm: React.FC<MaterialFormProps> = ({
   onSubmit,
   onCancel,
   loading,
+  uploadProgress,
+  tags = [],
+  tagMap = new Map(),
+  onCreateTag,
 }) => {
   const [name, setName] = useState(initial?.name ?? "");
   const [description, setDescription] = useState(initial?.description ?? "");
   const [gender, setGender] = useState<VoiceGender>(
     initial?.gender ?? "female",
   );
-  const [selectedTags, setSelectedTags] = useState<string[]>(
-    initial?.tags ?? [],
+  const [selectedTagIds, setSelectedTagIds] = useState<string[]>(
+    initial?.tagIds ?? [],
   );
   const [file, setFile] = useState<File | undefined>(undefined);
   const fileInputRef = useRef<HTMLInputElement>(null);
 
-  const handleTagToggle = (tag: string) => {
-    setSelectedTags((prev) =>
-      prev.includes(tag) ? prev.filter((t) => t !== tag) : [...prev, tag],
-    );
-  };
-
   const handleSubmit = () => {
     if (!name.trim()) return;
     if (!initial && !file) return;
@@ -227,7 +404,7 @@ const MaterialForm: React.FC<MaterialFormProps> = ({
       name: name.trim(),
       description: description.trim(),
       gender,
-      tags: selectedTags,
+      tagIds: selectedTagIds,
       fileName: file?.name ?? initial?.fileName ?? "",
       fileSize: file?.size ?? initial?.fileSize ?? 0,
       duration: initial?.duration ?? 0,
@@ -288,6 +465,18 @@ const MaterialForm: React.FC<MaterialFormProps> = ({
               </div>
             )}
           </div>
+          {/* 上传进度条 */}
+          {uploadProgress !== null && uploadProgress !== undefined && (
+            <div className="vmat-upload-progress">
+              <div
+                className="vmat-upload-progress-bar"
+                style={{ width: `${uploadProgress}%` }}
+              />
+              <span className="vmat-upload-progress-text">
+                {uploadProgress}%
+              </span>
+            </div>
+          )}
         </div>
       )}
 
@@ -335,18 +524,13 @@ const MaterialForm: React.FC<MaterialFormProps> = ({
       {/* 风格标签 */}
       <div className="vmat-form-field">
         <label className="vmat-form-label">风格标签</label>
-        <div className="vmat-tag-group">
-          {STYLE_TAGS.map((tag) => (
-            <button
-              key={tag}
-              type="button"
-              className={`vmat-tag-btn${selectedTags.includes(tag) ? " active" : ""}`}
-              onClick={() => handleTagToggle(tag)}
-            >
-              {tag}
-            </button>
-          ))}
-        </div>
+        <TagSelector
+          value={selectedTagIds}
+          onChange={setSelectedTagIds}
+          tags={tags}
+          tagMap={tagMap}
+          onCreateTag={onCreateTag ?? (async () => ({ id: "", name: "" }))}
+        />
       </div>
 
       {/* 操作按钮 */}
@@ -376,39 +560,88 @@ interface VoiceCardProps {
   material: VoiceMaterial;
   isPlaying: boolean;
   currentTime: number;
+  isSelected: boolean;
+  batchMode: boolean;
+  volume: number;
+  tagMap: Map<string, TagItem>;
   onPlay: () => void;
   onPause: () => void;
   onSeek: (time: number) => void;
   onEdit: () => void;
   onDelete: () => void;
+  onToggleSelect: (id: string) => void;
+  onVolumeChange: (e: React.ChangeEvent<HTMLInputElement>) => void;
+  onToggleMute: () => void;
 }
 
 const VoiceMaterialCard: React.FC<VoiceCardProps> = ({
   material,
   isPlaying,
   currentTime,
+  isSelected,
+  batchMode,
+  volume,
+  tagMap,
   onPlay,
   onPause,
   onSeek,
   onEdit,
   onDelete,
+  onToggleSelect,
+  onVolumeChange,
+  onToggleMute,
 }) => {
   const progressRef = useRef<HTMLDivElement>(null);
 
-  const handleProgressClick = (e: React.MouseEvent<HTMLDivElement>) => {
+  const handleProgressMouseDown = (e: React.MouseEvent<HTMLDivElement>) => {
     if (!progressRef.current) return;
-    const rect = progressRef.current.getBoundingClientRect();
-    const percent = (e.clientX - rect.left) / rect.width;
-    onSeek(Math.max(0, Math.min(1, percent)) * material.duration);
+    e.preventDefault();
+    const doSeek = (ev: MouseEvent) => {
+      if (!progressRef.current) return;
+      const rect = progressRef.current.getBoundingClientRect();
+      const percent = Math.max(
+        0,
+        Math.min(1, (ev.clientX - rect.left) / rect.width),
+      );
+      onSeek(percent * material.duration);
+    };
+    doSeek(e.nativeEvent);
+    const handleMove = (ev: MouseEvent) => doSeek(ev);
+    const handleUp = () => {
+      document.removeEventListener("mousemove", handleMove);
+      document.removeEventListener("mouseup", handleUp);
+    };
+    document.addEventListener("mousemove", handleMove);
+    document.addEventListener("mouseup", handleUp);
   };
 
   const progress =
     material.duration > 0 ? (currentTime / material.duration) * 100 : 0;
 
+  const handleCardClick = () => {
+    if (batchMode) {
+      onToggleSelect(material.id);
+    }
+  };
+
   return (
     <div
-      className={`vmat-card ${genderClass(material.gender)}${isPlaying ? " playing" : ""}`}
+      className={`vmat-card ${genderClass(material.gender)}${isPlaying ? " playing" : ""}${isSelected ? " selected" : ""}${batchMode ? " batch-mode" : ""}`}
+      onClick={handleCardClick}
     >
+      {/* 批量选择 checkbox */}
+      {(batchMode || isSelected) && (
+        <div
+          className={`vmat-card-checkbox vmat-checkbox${isSelected ? " checked" : ""}`}
+          onClick={(e) => {
+            e.stopPropagation();
+            onToggleSelect(material.id);
+          }}
+        >
+          {isSelected && <CheckOutlined />}
+        </div>
+      )}
+
       {/* 操作按钮 */}
       <div className="vmat-card-actions">
         <button
@@ -458,11 +691,37 @@ const VoiceMaterialCard: React.FC<VoiceCardProps> = ({
 
       {/* 标签 */}
       <div className="vmat-card-tags">
-        {material.tags.map((tag) => (
-          <Tag key={tag} variant="info">
-            {tag}
-          </Tag>
-        ))}
+        {material.tagIds.length === 0 ? (
+          <span
+            className="vmat-tag-empty"
+            onClick={(e) => {
+              e.stopPropagation();
+              onEdit();
+            }}
+          >
+            添加标签
+          </span>
+        ) : (
+          <>
+            {material.tagIds.slice(0, MAX_CARD_TAGS).map((tagId, i) => (
+              <Tag key={tagId} variant={TAG_VARIANTS[i % TAG_VARIANTS.length]}>
+                {tagMap.get(tagId)?.name ?? tagId}
+              </Tag>
+            ))}
+            {material.tagIds.length > MAX_CARD_TAGS && (
+              <Tooltip
+                title={material.tagIds
+                  .slice(MAX_CARD_TAGS)
+                  .map((id) => tagMap.get(id)?.name ?? id)
+                  .join("、")}
+              >
+                <Tag className="vmat-tag-overflow">
+                  +{material.tagIds.length - MAX_CARD_TAGS}
+                </Tag>
+              </Tooltip>
+            )}
+          </>
+        )}
       </div>
 
       {/* 元信息 */}
@@ -488,18 +747,51 @@ const VoiceMaterialCard: React.FC<VoiceCardProps> = ({
         <div
           ref={progressRef}
           className="vmat-progress"
-          onClick={handleProgressClick}
+          onMouseDown={handleProgressMouseDown}
         >
           <div
             className="vmat-progress-bar"
             style={{ width: `${progress}%` }}
           />
+          {isPlaying && (
+            <div
+              className="vmat-progress-thumb"
+              style={{ left: `${progress}%` }}
+            />
+          )}
         </div>
         <span className="vmat-time">
           {isPlaying
             ? formatDuration(currentTime)
             : formatDuration(material.duration)}
         </span>
+        {/* 音量控制 */}
+        <div className="vmat-volume">
+          <button
+            type="button"
+            className="vmat-volume-btn"
+            onClick={(e) => {
+              e.stopPropagation();
+              onToggleMute();
+            }}
+            title={volume === 0 ? "取消静音" : "静音"}
+          >
+            {volume === 0 ? <MutedOutlined /> : <SoundOutlined />}
+          </button>
+          <input
+            type="range"
+            className="vmat-volume-slider"
+            min={0}
+            max={1}
+            step={0.05}
+            value={volume}
+            onChange={(e) => {
+              e.stopPropagation();
+              onVolumeChange(e);
+            }}
+            onClick={(e) => e.stopPropagation()}
+          />
+        </div>
       </div>
     </div>
   );
@@ -513,30 +805,53 @@ interface VoiceRowProps {
   material: VoiceMaterial;
   isPlaying: boolean;
   currentTime: number;
+  isSelected: boolean;
+  batchMode: boolean;
+  tagMap: Map<string, TagItem>;
   onPlay: () => void;
   onPause: () => void;
   onSeek: (time: number) => void;
   onEdit: () => void;
   onDelete: () => void;
+  onToggleSelect: (id: string) => void;
 }
 
 const VoiceMaterialRow: React.FC<VoiceRowProps> = ({
   material,
   isPlaying,
   currentTime,
+  isSelected,
+  batchMode,
+  tagMap,
   onPlay,
   onPause,
   onSeek,
   onEdit,
   onDelete,
+  onToggleSelect,
 }) => {
   const progressRef = useRef<HTMLDivElement>(null);
 
-  const handleProgressClick = (e: React.MouseEvent<HTMLDivElement>) => {
+  const handleProgressMouseDown = (e: React.MouseEvent<HTMLDivElement>) => {
     if (!progressRef.current) return;
-    const rect = progressRef.current.getBoundingClientRect();
-    const percent = (e.clientX - rect.left) / rect.width;
-    onSeek(Math.max(0, Math.min(1, percent)) * material.duration);
+    e.preventDefault();
+    const doSeek = (ev: MouseEvent) => {
+      if (!progressRef.current) return;
+      const rect = progressRef.current.getBoundingClientRect();
+      const percent = Math.max(
+        0,
+        Math.min(1, (ev.clientX - rect.left) / rect.width),
+      );
+      onSeek(percent * material.duration);
+    };
+    doSeek(e.nativeEvent);
+    const handleMove = (ev: MouseEvent) => doSeek(ev);
+    const handleUp = () => {
+      document.removeEventListener("mousemove", handleMove);
+      document.removeEventListener("mouseup", handleUp);
+    };
+    document.addEventListener("mousemove", handleMove);
+    document.addEventListener("mouseup", handleUp);
   };
 
   const progress =
@@ -544,8 +859,21 @@ const VoiceMaterialRow: React.FC<VoiceRowProps> = ({
 
   return (
     <div
-      className={`vmat-row ${genderClass(material.gender)}${isPlaying ? " playing" : ""}`}
+      className={`vmat-row ${genderClass(material.gender)}${isPlaying ? " playing" : ""}${isSelected ? " selected" : ""}${batchMode ? " batch-mode" : ""}`}
     >
+      {/* 批量选择 checkbox */}
+      {(batchMode || isSelected) && (
+        <div
+          className={`vmat-row-checkbox vmat-checkbox${isSelected ? " checked" : ""}`}
+          onClick={(e) => {
+            e.stopPropagation();
+            onToggleSelect(material.id);
+          }}
+        >
+          {isSelected && <CheckOutlined />}
+        </div>
+      )}
+
       {/* 播放按钮 */}
       <button
         type="button"
@@ -575,23 +903,49 @@ const VoiceMaterialRow: React.FC<VoiceRowProps> = ({
 
       {/* 标签 */}
       <div className="vmat-row-tags">
-        {material.tags.map((tag) => (
-          <Tag key={tag} variant="info">
-            {tag}
-          </Tag>
-        ))}
+        {material.tagIds.length === 0 ? (
+          <span className="vmat-tag-empty" onClick={() => onEdit()}>
+            添加标签
+          </span>
+        ) : (
+          <>
+            {material.tagIds.slice(0, MAX_ROW_TAGS).map((tagId, i) => (
+              <Tag key={tagId} variant={TAG_VARIANTS[i % TAG_VARIANTS.length]}>
+                {tagMap.get(tagId)?.name ?? tagId}
+              </Tag>
+            ))}
+            {material.tagIds.length > MAX_ROW_TAGS && (
+              <Tooltip
+                title={material.tagIds
+                  .slice(MAX_ROW_TAGS)
+                  .map((id) => tagMap.get(id)?.name ?? id)
+                  .join("、")}
+              >
+                <Tag className="vmat-tag-overflow">
+                  +{material.tagIds.length - MAX_ROW_TAGS}
+                </Tag>
+              </Tooltip>
+            )}
+          </>
+        )}
       </div>
 
-      {/* 进度条 */}
+      {/* 进度条（可拖拽） */}
       <div
         ref={progressRef}
         className="vmat-row-progress"
-        onClick={handleProgressClick}
+        onMouseDown={handleProgressMouseDown}
       >
         <div
           className="vmat-row-progress-bar"
           style={{ width: `${progress}%` }}
         />
+        {isPlaying && (
+          <div
+            className="vmat-progress-thumb"
+            style={{ left: `${progress}%` }}
+          />
+        )}
       </div>
 
       {/* 时长 */}
@@ -667,16 +1021,46 @@ const VoiceMaterialLibrary: React.FC = () => {
     }
   }, [libraries, voiceLibrary, createLibMutation]);
 
+  // ── 获取标签列表 ───────────────────────────────────────────
+  const { data: tags = [] } = useQuery({
+    queryKey: ["tags"],
+    queryFn: getTags,
+    staleTime: 60_000,
+  });
+
+  /** 标签 ID → TagItem 映射（用于卡片/行渲染） */
+  const tagMap = useMemo(() => {
+    const m = new Map<string, TagItem>();
+    tags.forEach((t) => m.set(t.id, t));
+    return m;
+  }, [tags]);
+
+  /** 创建标签 mutation（供 TagSelector 调用） */
+  const createTagMutation = useMutation({
+    mutationFn: (name: string) => createTag(name),
+    onSuccess: () => {
+      queryClient.invalidateQueries({ queryKey: ["tags"] });
+    },
+  });
+
+  /** 创建标签并返回 TagItem（供 TagSelector 使用） */
+  const handleCreateTag = useCallback(
+    async (name: string): Promise<TagItem> => {
+      return createTagMutation.mutateAsync(name);
+    },
+    [createTagMutation],
+  );
+
   // ── 视图状态 ──────────────────────────────────────────────
   const [viewMode, setViewMode] = useState<ViewMode>("card");
   const [searchText, setSearchText] = useState("");
   const [filterGender, setFilterGender] = useState<string>("all");
-  const [filterTag, setFilterTag] = useState<string>("all");
+  const [filterTagId, setFilterTagId] = useState<string>("all");
 
   // ── 获取配音素材列表（筛选参数透传后端） ─────────────────
   const filterKeyword = searchText.trim() || undefined;
   const filterGenderParam = filterGender !== "all" ? filterGender : undefined;
-  const filterStyleParam = filterTag !== "all" ? filterTag : undefined;
+  const filterTagIdsParam = filterTagId !== "all" ? [filterTagId] : undefined;
 
   const { data: assets = [], isLoading } = useQuery({
     queryKey: [
@@ -685,14 +1069,14 @@ const VoiceMaterialLibrary: React.FC = () => {
       {
         keyword: filterKeyword,
         gender: filterGenderParam,
-        style: filterStyleParam,
+        tag_ids: filterTagIdsParam,
       },
     ],
     queryFn: () =>
       getAssetsByKind("voice", {
         keyword: filterKeyword,
         gender: filterGenderParam,
-        style: filterStyleParam,
+        tag_ids: filterTagIdsParam,
       }),
     staleTime: 30_000,
   });
@@ -709,69 +1093,100 @@ const VoiceMaterialLibrary: React.FC = () => {
       name: string;
       gender: VoiceGender;
       description: string;
-      tags: string[];
+      tagIds: string[];
     }) => {
-      // 1. 获取或等待 voice library
-      let lib = voiceLibrary;
-      if (!lib) {
-        if (createLibMutation.isPending) {
-          await createLibMutation.mutateAsync();
+      setUploadProgress(0);
+      try {
+        // 1. 获取或等待 voice library
+        let lib = voiceLibrary;
+        if (!lib) {
+          if (createLibMutation.isPending) {
+            await createLibMutation.mutateAsync();
+          }
+          const libs = await queryClient.fetchQuery({
+            queryKey: ["asset-libraries"],
+            queryFn: getAssetLibraries,
+          });
+          lib = libs.find((l) => l.kind === "voice");
+          if (!lib) throw new Error("无法创建配音素材库");
         }
-        const libs = await queryClient.fetchQuery({
-          queryKey: ["asset-libraries"],
-          queryFn: getAssetLibraries,
+
+        // 2. 上传文件（带进度）
+        const { storage_key } = await uploadAssetDirect({
+          file: data.file,
+          library_id: lib.id,
+          onProgress: (p) => setUploadProgress(p),
         });
-        lib = libs.find((l) => l.kind === "voice");
-        if (!lib) throw new Error("无法创建配音素材库");
+
+        // 3. 获取音频时长
+        const duration = await getAudioDuration(data.file);
+
+        // 4. 创建素材记录
+        const asset = await createAsset({
+          library_id: lib.id,
+          name: data.name,
+          storage_key,
+          mime_type: data.file.type || "audio/mpeg",
+          metadata: buildMetadata({
+            gender: data.gender,
+            description: data.description,
+            duration,
+          }),
+        });
+
+        // 5. 打标签（标签走独立 API）
+        if (data.tagIds.length > 0) {
+          await tagAsset(asset.id, data.tagIds);
+        }
+      } finally {
+        setUploadProgress(null);
       }
-
-      // 2. 上传文件
-      const { storage_key } = await uploadAssetDirect({
-        file: data.file,
-        library_id: lib.id,
-      });
-
-      // 3. 获取音频时长
-      const duration = await getAudioDuration(data.file);
-
-      // 4. 创建素材记录
-      await createAsset({
-        library_id: lib.id,
-        name: data.name,
-        storage_key,
-        mime_type: data.file.type || "audio/mpeg",
-        metadata: buildMetadata({
-          gender: data.gender,
-          description: data.description,
-          tags: data.tags,
-          duration,
-        }),
-      });
     },
     onSuccess: () => {
       queryClient.invalidateQueries({ queryKey: ["assets", "voice"] });
+      queryClient.invalidateQueries({ queryKey: ["tags"] });
+    },
+    onError: (err: Error) => {
+      message.error(err.message || "上传失败，请重试");
     },
   });
 
   // ── 编辑 mutation ─────────────────────────────────────────
   const editMutation = useMutation({
-    mutationFn: (data: {
+    mutationFn: async (data: {
       id: string;
       name: string;
       gender: VoiceGender;
       description: string;
-      tags: string[];
-    }) =>
-      updateAsset(data.id, {
+      tagIds: string[];
+    }) => {
+      // 1. 更新基础信息
+      await updateAsset(data.id, {
         name: data.name,
         metadata: buildMetadata({
           gender: data.gender,
           description: data.description,
-          tags: data.tags,
         }),
-      }),
+      });
+
+      // 2. 对比标签差异，调用 tag/untag API
+      const currentAsset = materials.find((m) => m.id === data.id);
+      const oldTagIds = currentAsset?.tagIds ?? [];
+      const newTagIds = data.tagIds;
+
+      const toAdd = newTagIds.filter((id) => !oldTagIds.includes(id));
+      const toRemove = oldTagIds.filter((id) => !newTagIds.includes(id));
+
+      if (toAdd.length > 0) {
+        await tagAsset(data.id, toAdd);
+      }
+      for (const tagId of toRemove) {
+        await untagAsset(data.id, tagId);
+      }
+    },
     onSuccess: () => {
       queryClient.invalidateQueries({ queryKey: ["assets", "voice"] });
+      queryClient.invalidateQueries({ queryKey: ["tags"] });
     },
   });
 
@@ -794,6 +1209,15 @@ const VoiceMaterialLibrary: React.FC = () => {
     null,
   );
 
+  // ── 批量操作 / 上传进度 / 音量 ────────────────────────────
+  const [selectedIds, setSelectedIds] = useState<Set<string>>(new Set());
+  const [uploadProgress, setUploadProgress] = useState<number | null>(null);
+  const [volume, setVolume] = useState(0.7);
+  const [pausedMaterial, setPausedMaterial] = useState<VoiceMaterial | null>(
+    null,
+  );
+  const [batchCustomTag, setBatchCustomTag] = useState("");
+
   // ── 播放控制 ──────────────────────────────────────────────
   const stopPlayback = useCallback(() => {
     if (audioRef.current) {
@@ -802,6 +1226,7 @@ const VoiceMaterialLibrary: React.FC = () => {
     }
     setPlayingId(null);
     setCurrentTime(0);
+    setPausedMaterial(null);
   }, []);
 
   const startPlayback = useCallback(
@@ -810,6 +1235,7 @@ const VoiceMaterialLibrary: React.FC = () => {
       stopPlayback();
 
       const audio = new Audio(material.fileUrl);
+      audio.volume = volume;
       audioRef.current = audio;
 
       audio.addEventListener("timeupdate", () => {
@@ -820,6 +1246,7 @@ const VoiceMaterialLibrary: React.FC = () => {
         setPlayingId(null);
         setCurrentTime(0);
         audioRef.current = null;
+        setPausedMaterial(null);
       });
 
       audio.play().catch(() => {
@@ -829,21 +1256,37 @@ const VoiceMaterialLibrary: React.FC = () => {
 
       setPlayingId(material.id);
       setCurrentTime(0);
+      setPausedMaterial(null);
     },
-    [stopPlayback],
+    [stopPlayback, volume],
   );
 
   const handlePlay = useCallback(
     (material: VoiceMaterial) => {
       if (playingId === material.id) return;
+      // 恢复暂停
+      if (
+        pausedMaterial?.id === material.id &&
+        audioRef.current &&
+        audioRef.current.paused
+      ) {
+        audioRef.current.play().catch(() => {});
+        setPlayingId(material.id);
+        setPausedMaterial(null);
+        return;
+      }
       startPlayback(material);
     },
-    [playingId, startPlayback],
+    [playingId, pausedMaterial, startPlayback],
   );
 
-  const handlePause = useCallback(() => {
-    stopPlayback();
-  }, [stopPlayback]);
+  const handlePause = useCallback((material?: VoiceMaterial) => {
+    if (audioRef.current) {
+      audioRef.current.pause();
+    }
+    setPlayingId(null);
+    if (material) setPausedMaterial(material);
+  }, []);
 
   const handleSeek = useCallback(
     (material: VoiceMaterial, time: number) => {
@@ -862,6 +1305,25 @@ const VoiceMaterialLibrary: React.FC = () => {
     [startPlayback],
   );
 
+  const handleVolumeChange = useCallback(
+    (e: React.ChangeEvent<HTMLInputElement>) => {
+      const v = parseFloat(e.target.value);
+      setVolume(v);
+      if (audioRef.current) audioRef.current.volume = v;
+    },
+    [],
+  );
+
+  const toggleMute = useCallback(() => {
+    if (volume > 0) {
+      setVolume(0);
+      if (audioRef.current) audioRef.current.volume = 0;
+    } else {
+      setVolume(0.7);
+      if (audioRef.current) audioRef.current.volume = 0.7;
+    }
+  }, [volume]);
+
   // 组件卸载时清理 audio
   useEffect(() => {
     return () => {
@@ -884,7 +1346,7 @@ const VoiceMaterialLibrary: React.FC = () => {
         name: data.name,
         gender: data.gender,
         description: data.description,
-        tags: data.tags,
+        tagIds: data.tagIds,
       },
       {
         onSuccess: () => {
@@ -903,7 +1365,7 @@ const VoiceMaterialLibrary: React.FC = () => {
       name: data.name,
       gender: data.gender,
       description: data.description,
-      tags: data.tags,
+      tagIds: data.tagIds,
     });
     setEditingMaterial(null);
   };
@@ -931,8 +1393,8 @@ const VoiceMaterialLibrary: React.FC = () => {
     if (filterGender !== "all") {
       list = list.filter((m) => m.gender === filterGender);
     }
-    if (filterTag !== "all") {
-      list = list.filter((m) => m.tags.includes(filterTag));
+    if (filterTagId !== "all") {
+      list = list.filter((m) => m.tagIds.includes(filterTagId));
     }
     if (searchText.trim()) {
       const q = searchText.trim().toLowerCase();
@@ -940,20 +1402,109 @@ const VoiceMaterialLibrary: React.FC = () => {
         (m) =>
           m.name.toLowerCase().includes(q) ||
           m.description.toLowerCase().includes(q) ||
-          m.tags.some((t) => t.toLowerCase().includes(q)),
+          m.tagIds.some((id) =>
+            tagMap.get(id)?.name?.toLowerCase().includes(q),
+          ),
       );
     }
     return list;
-  }, [materials, filterGender, filterTag, searchText]);
+  }, [materials, filterGender, filterTagId, searchText, tagMap]);
 
-  /* ── 所有用到的标签（用于筛选下拉） ────────────────────── */
+  /* ── 标签使用计数（药丸条展示，按 tag ID 统计） ──────────── */
 
-  const allTags = React.useMemo(() => {
-    const set = new Set<string>();
-    materials.forEach((m) => m.tags.forEach((t) => set.add(t)));
-    return Array.from(set).sort();
+  const tagCountMap = React.useMemo(() => {
+    const map: Record<string, number> = {};
+    materials.forEach((m) =>
+      m.tagIds.forEach((id) => {
+        map[id] = (map[id] || 0) + 1;
+      }),
+    );
+    return map;
   }, [materials]);
 
+  /* ── 批量操作 ─────────────────────────────────────────── */
+
+  const batchMode = selectedIds.size > 0;
+  const allSelected =
+    filtered.length > 0 && filtered.every((m) => selectedIds.has(m.id));
+
+  const handleToggleSelect = useCallback((id: string) => {
+    setSelectedIds((prev) => {
+      const next = new Set(prev);
+      if (next.has(id)) next.delete(id);
+      else next.add(id);
+      return next;
+    });
+  }, []);
+
+  const handleSelectAll = useCallback(() => {
+    if (allSelected) setSelectedIds(new Set());
+    else setSelectedIds(new Set(filtered.map((m) => m.id)));
+  }, [allSelected, filtered]);
+
+  const handleBatchDelete = useCallback(async () => {
+    const ids = Array.from(selectedIds);
+    let successCount = 0;
+    for (const id of ids) {
+      try {
+        await deleteAsset(id);
+        successCount++;
+      } catch {
+        /* ignore individual failures */
+      }
+      if (playingId === id) stopPlayback();
+    }
+    queryClient.invalidateQueries({ queryKey: ["assets", "voice"] });
+    setSelectedIds(new Set());
+    message.success(`已批量删除 ${successCount}/${ids.length} 个素材`);
+  }, [selectedIds, playingId, stopPlayback, queryClient]);
+
+  const handleBatchTag = useCallback(
+    async (tagId: string) => {
+      const ids = Array.from(selectedIds);
+      let successCount = 0;
+      for (const id of ids) {
+        try {
+          await tagAsset(id, [tagId]);
+          successCount++;
+        } catch {
+          /* ignore individual failures */
+        }
+      }
+      queryClient.invalidateQueries({ queryKey: ["assets", "voice"] });
+      setSelectedIds(new Set());
+      const tagName = tagMap.get(tagId)?.name ?? tagId;
+      if (successCount === 0) {
+        message.error(`批量打标签失败，请重试`);
+      } else {
+        message.success(
+          `已为 ${successCount}/${ids.length} 个素材添加标签「${tagName}」`,
+        );
+      }
+    },
+    [selectedIds, queryClient, tagMap],
+  );
+
+  /** 批量打标签 — 自定义输入：按名称查找或创建标签，再批量打标 */
+  const handleBatchCustomTag = useCallback(
+    async (name: string) => {
+      // 先查找同名标签（不区分大小写）
+      let existing = tags.find(
+        (t) => t.name.toLowerCase() === name.toLowerCase(),
+      );
+      if (!existing) {
+        try {
+          existing = await createTagMutation.mutateAsync(name);
+        } catch {
+          message.error(`创建标签「${name}」失败`);
+          return;
+        }
+      }
+      await handleBatchTag(existing.id);
+    },
+    [tags, createTagMutation, handleBatchTag],
+  );
+
   /* ── 渲染 ─────────────────────────────────────────────── */
 
   const isUploading = uploadMutation.isPending;
@@ -1003,15 +1554,6 @@ const VoiceMaterialLibrary: React.FC = () => {
               { value: "neutral", label: "中性" },
             ]}
           />
-          <Select
-            value={filterTag}
-            onChange={setFilterTag}
-            style={{ width: 120 }}
-            options={[
-              { value: "all", label: "全部风格" },
-              ...allTags.map((t) => ({ value: t, label: t })),
-            ]}
-          />
         </div>
         <div className="vmat-toolbar-right">
           <span className="vmat-result-count">共 {filtered.length} 个素材</span>
@@ -1036,6 +1578,34 @@ const VoiceMaterialLibrary: React.FC = () => {
         </div>
       </div>
 
+      {/* ── 标签筛选药丸条 ─────────────────────────────────── */}
+      {tags.length > 0 && (
+        <div className="vmat-tag-filter-bar">
+          <button
+            type="button"
+            className={`vmat-filter-pill${filterTagId === "all" ? " active" : ""}`}
+            onClick={() => setFilterTagId("all")}
+            tabIndex={0}
+          >
+            全部
+          </button>
+          {tags.map((tag) => (
+            <button
+              key={tag.id}
+              type="button"
+              className={`vmat-filter-pill${filterTagId === tag.id ? " active" : ""}`}
+              onClick={() => setFilterTagId(tag.id)}
+              tabIndex={0}
+            >
+              {tag.name}
+              <span className="vmat-filter-pill-count">
+                {tagCountMap[tag.id] || 0}
+              </span>
+            </button>
+          ))}
+        </div>
+      )}
+
       {/* 加载中 */}
       {isLoading && (
         <div className="vmat-empty">
@@ -1046,6 +1616,83 @@ const VoiceMaterialLibrary: React.FC = () => {
         </div>
       )}
 
+      {/* 批量操作栏 */}
+      {batchMode && (
+        <div className="vmat-batch-bar">
+          <div className="vmat-batch-bar-left">
+            <div
+              className={`vmat-checkbox${allSelected ? " checked" : ""}`}
+              onClick={handleSelectAll}
+            >
+              {allSelected && <CheckOutlined />}
+            </div>
+            <span className="vmat-select-all" onClick={handleSelectAll}>
+              {allSelected ? "取消全选" : "全选"}
+            </span>
+            <span className="vmat-batch-count">
+              已选择 {selectedIds.size} 项
+            </span>
+          </div>
+          <div className="vmat-batch-bar-right">
+            <Popover
+              content={
+                <div className="vmat-tag-popover">
+                  <div className="vmat-tag-pop-input-row">
+                    <Input
+                      placeholder="输入自定义标签后回车"
+                      value={batchCustomTag}
+                      onChange={(e: React.ChangeEvent<HTMLInputElement>) =>
+                        setBatchCustomTag(e.target.value)
+                      }
+                      onKeyDown={(e: React.KeyboardEvent<HTMLInputElement>) => {
+                        if (e.key === "Enter" && batchCustomTag.trim()) {
+                          handleBatchCustomTag(batchCustomTag.trim());
+                          setBatchCustomTag("");
+                        }
+                      }}
+                    />
+                  </div>
+                  {tags.map((tag) => (
+                    <button
+                      key={tag.id}
+                      type="button"
+                      className="vmat-tag-pop-btn"
+                      onClick={() => handleBatchTag(tag.id)}
+                    >
+                      {tag.name}
+                    </button>
+                  ))}
+                </div>
+              }
+              title="批量打标签"
+              trigger="click"
+            >
+              <Button
+                buttonType="ghost"
+                buttonSize="sm"
+                icon={<TagsOutlined />}
+              >
+                批量打标签
+              </Button>
+            </Popover>
+            <Popconfirm
+              title={`确定删除选中的 ${selectedIds.size} 个素材？`}
+              onConfirm={handleBatchDelete}
+              okText="删除"
+              cancelText="取消"
+            >
+              <Button
+                buttonType="danger"
+                buttonSize="sm"
+                icon={<DeleteOutlined />}
+              >
+                批量删除
+              </Button>
+            </Popconfirm>
+          </div>
+        </div>
+      )}
+
       {/* 内容区 */}
       {!isLoading && filtered.length > 0 && viewMode === "card" && (
         <div className="vmat-grid">
@@ -1053,22 +1700,30 @@ const VoiceMaterialLibrary: React.FC = () => {
             <VoiceMaterialCard
               key={m.id}
               material={m}
+              tagMap={tagMap}
               isPlaying={playingId === m.id}
               currentTime={playingId === m.id ? currentTime : 0}
+              isSelected={selectedIds.has(m.id)}
+              batchMode={batchMode}
+              volume={volume}
               onPlay={() => handlePlay(m)}
-              onPause={handlePause}
+              onPause={() => handlePause(m)}
               onSeek={(t) => handleSeek(m, t)}
               onEdit={() => setEditingMaterial(m)}
               onDelete={() => handleDelete(m.id)}
+              onToggleSelect={handleToggleSelect}
+              onVolumeChange={handleVolumeChange}
+              onToggleMute={toggleMute}
             />
           ))}
         </div>
       )}
 
       {!isLoading && filtered.length > 0 && viewMode === "list" && (
-        <div className="vmat-list">
+        <div className={`vmat-list${batchMode ? " batch-mode" : ""}`}>
           {/* 列表头 */}
           <div className="vmat-list-header">
+            {batchMode && <span className="vmat-lh-checkbox" />}
             <span className="vmat-lh-play" />
             <span className="vmat-lh-info">名称</span>
             <span className="vmat-lh-gender">性别</span>
@@ -1082,13 +1737,17 @@ const VoiceMaterialLibrary: React.FC = () => {
             <VoiceMaterialRow
               key={m.id}
               material={m}
+              tagMap={tagMap}
               isPlaying={playingId === m.id}
               currentTime={playingId === m.id ? currentTime : 0}
+              isSelected={selectedIds.has(m.id)}
+              batchMode={batchMode}
               onPlay={() => handlePlay(m)}
-              onPause={handlePause}
+              onPause={() => handlePause(m)}
               onSeek={(t) => handleSeek(m, t)}
               onEdit={() => setEditingMaterial(m)}
               onDelete={() => handleDelete(m.id)}
+              onToggleSelect={handleToggleSelect}
             />
           ))}
         </div>
@@ -1101,11 +1760,11 @@ const VoiceMaterialLibrary: React.FC = () => {
           </div>
           <h3>暂无配音素材</h3>
           <p>
-            {searchText || filterGender !== "all" || filterTag !== "all"
+            {searchText || filterGender !== "all" || filterTagId !== "all"
               ? "未找到匹配的素材，试试调整筛选条件"
               : "上传音频文件，开始管理配音素材"}
           </p>
-          {!searchText && filterGender === "all" && filterTag === "all" && (
+          {!searchText && filterGender === "all" && filterTagId === "all" && (
             <Button
               buttonType="primary"
               buttonSize="md"
@@ -1131,6 +1790,10 @@ const VoiceMaterialLibrary: React.FC = () => {
           onSubmit={handleUpload}
           onCancel={() => setUploadOpen(false)}
           loading={isUploading}
+          uploadProgress={uploadProgress}
+          tags={tags}
+          tagMap={tagMap}
+          onCreateTag={handleCreateTag}
         />
       </Modal>
 
@@ -1149,6 +1812,9 @@ const VoiceMaterialLibrary: React.FC = () => {
             onSubmit={handleEdit}
             onCancel={() => setEditingMaterial(null)}
             loading={isEditing}
+            tags={tags}
+            tagMap={tagMap}
+            onCreateTag={handleCreateTag}
           />
         )}
       </Modal>
diff --git a/apps/web/src/pages/voice-materials/voice-materials.css b/apps/web/src/pages/voice-materials/voice-materials.css
index bcdb88ba5..fed41ec53 100644
--- a/apps/web/src/pages/voice-materials/voice-materials.css
+++ b/apps/web/src/pages/voice-materials/voice-materials.css
@@ -111,7 +111,9 @@
 
 .vmat-card.playing {
   border-color: var(--primary-500);
-  box-shadow: 0 0 0 1px var(--primary-500), var(--shadow-md);
+  box-shadow:
+    0 0 0 1px var(--primary-500),
+    var(--shadow-md);
 }
 
 /* 性别色带 */
@@ -812,3 +814,449 @@
     justify-content: space-between;
   }
 }
+
+/* ─── 批量操作栏 ─────────────────────────────────────────── */
+
+.vmat-batch-bar {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  padding: 8px 16px;
+  margin-bottom: 16px;
+  background: var(--primary-soft, #eef2ff);
+  border: 1px solid var(--primary-color, #6366f1);
+  border-radius: 8px;
+  animation: vmat-batch-bar-in 0.2s ease;
+}
+
+@keyframes vmat-batch-bar-in {
+  from {
+    opacity: 0;
+    transform: translateY(-8px);
+  }
+  to {
+    opacity: 1;
+    transform: translateY(0);
+  }
+}
+
+.vmat-batch-bar-left {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+}
+
+.vmat-batch-bar-right {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+}
+
+.vmat-select-all {
+  font-size: 13px;
+  color: var(--text-secondary, #64748b);
+  cursor: pointer;
+  user-select: none;
+}
+
+.vmat-select-all:hover {
+  color: var(--primary-color, #6366f1);
+}
+
+.vmat-batch-count {
+  font-size: 13px;
+  font-weight: 500;
+  color: var(--primary-color, #6366f1);
+}
+
+/* ─── 自定义 checkbox ────────────────────────────────────── */
+
+.vmat-checkbox {
+  width: 18px;
+  height: 18px;
+  border: 2px solid var(--neutral-300, #cbd5e1);
+  border-radius: 4px;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  cursor: pointer;
+  transition: all 0.15s ease;
+  background: var(--bg-primary, #fff);
+  flex-shrink: 0;
+}
+
+.vmat-checkbox:hover {
+  border-color: var(--primary-color, #6366f1);
+}
+
+.vmat-checkbox.checked {
+  background: var(--primary-color, #6366f1);
+  border-color: var(--primary-color, #6366f1);
+  color: #fff;
+  font-size: 10px;
+}
+
+/* ─── 卡片 checkbox 覆盖 ─────────────────────────────────── */
+
+.vmat-card-checkbox {
+  position: absolute;
+  top: 8px;
+  left: 8px;
+  z-index: 2;
+}
+
+.vmat-row-checkbox {
+  flex-shrink: 0;
+}
+
+/* ─── 选中态 + 批量模式 ──────────────────────────────────── */
+
+.vmat-card.selected {
+  border-color: var(--primary-color, #6366f1);
+  box-shadow: 0 0 0 1px var(--primary-color, #6366f1);
+}
+
+.vmat-row.selected {
+  background: var(--primary-soft, #eef2ff);
+}
+
+.vmat-card.batch-mode {
+  cursor: pointer;
+}
+
+.vmat-row.batch-mode {
+  cursor: pointer;
+}
+
+/* ─── 进度条拖拽 thumb ───────────────────────────────────── */
+
+.vmat-progress-thumb {
+  position: absolute;
+  top: 50%;
+  transform: translate(-50%, -50%);
+  width: 12px;
+  height: 12px;
+  border-radius: 50%;
+  background: var(--primary-color, #6366f1);
+  border: 2px solid #fff;
+  box-shadow: 0 1px 3px rgba(0, 0, 0, 0.2);
+  pointer-events: none;
+  z-index: 1;
+}
+
+/* ─── 音量控制 ────────────────────────────────────────────── */
+
+.vmat-volume {
+  display: flex;
+  align-items: center;
+  gap: 4px;
+  margin-top: 6px;
+}
+
+.vmat-volume-btn {
+  background: none;
+  border: none;
+  padding: 0;
+  cursor: pointer;
+  color: var(--text-secondary, #64748b);
+  font-size: 14px;
+  display: flex;
+  align-items: center;
+  transition: color 0.15s;
+}
+
+.vmat-volume-btn:hover {
+  color: var(--primary-color, #6366f1);
+}
+
+.vmat-volume-slider {
+  -webkit-appearance: none;
+  appearance: none;
+  width: 48px;
+  height: 4px;
+  border-radius: 2px;
+  background: var(--neutral-200, #e2e8f0);
+  outline: none;
+  cursor: pointer;
+}
+
+.vmat-volume-slider::-webkit-slider-thumb {
+  -webkit-appearance: none;
+  appearance: none;
+  width: 10px;
+  height: 10px;
+  border-radius: 50%;
+  background: var(--primary-color, #6366f1);
+  cursor: pointer;
+}
+
+.vmat-volume-slider::-moz-range-thumb {
+  width: 10px;
+  height: 10px;
+  border-radius: 50%;
+  background: var(--primary-color, #6366f1);
+  cursor: pointer;
+  border: none;
+}
+
+/* ─── 上传进度条 ──────────────────────────────────────────── */
+
+.vmat-upload-progress {
+  position: relative;
+  height: 20px;
+  background: var(--neutral-100, #f1f5f9);
+  border-radius: 10px;
+  overflow: hidden;
+  margin-top: 12px;
+}
+
+.vmat-upload-progress-bar {
+  height: 100%;
+  background: linear-gradient(90deg, var(--primary-color, #6366f1), #818cf8);
+  border-radius: 10px;
+  transition: width 0.3s ease;
+}
+
+.vmat-upload-progress-text {
+  position: absolute;
+  top: 50%;
+  left: 50%;
+  transform: translate(-50%, -50%);
+  font-size: 11px;
+  font-weight: 600;
+  color: var(--text-primary, #1e293b);
+}
+
+/* ─── 批量打标签 Popover ──────────────────────────────────── */
+
+.vmat-tag-popover {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 8px;
+  max-width: 240px;
+}
+
+.vmat-tag-pop-btn {
+  padding: 4px 12px;
+  border: 1px solid var(--neutral-200, #e2e8f0);
+  border-radius: 14px;
+  background: var(--bg-primary, #fff);
+  font-size: 12px;
+  color: var(--text-secondary, #64748b);
+  cursor: pointer;
+  transition: all 0.15s;
+}
+
+.vmat-tag-pop-btn:hover {
+  border-color: var(--primary-color, #6366f1);
+  color: var(--primary-color, #6366f1);
+  background: var(--primary-soft, #eef2ff);
+}
+
+/* ─── 列表 checkbox 列 ────────────────────────────────────── */
+
+.vmat-lh-checkbox {
+  width: 30px;
+}
+
+.vmat-list.batch-mode .vmat-list-header,
+.vmat-list.batch-mode .vmat-row {
+  grid-template-columns: 30px 48px 1fr 80px 160px 120px 60px 70px 80px;
+}
+
+/* ─── 标签筛选药丸条 ─────────────────────────────────────── */
+
+.vmat-tag-filter-bar {
+  display: flex;
+  gap: 8px;
+  overflow-x: auto;
+  padding: 10px 0;
+  margin-bottom: 12px;
+  scrollbar-width: thin;
+}
+
+.vmat-tag-filter-bar::-webkit-scrollbar {
+  height: 4px;
+}
+
+.vmat-tag-filter-bar::-webkit-scrollbar-thumb {
+  background: var(--neutral-300, #cbd5e1);
+  border-radius: 2px;
+}
+
+.vmat-filter-pill {
+  display: inline-flex;
+  align-items: center;
+  gap: 4px;
+  padding: 6px 16px;
+  border: 1px solid var(--border-primary);
+  border-radius: var(--radius-full, 9999px);
+  background: var(--bg-surface);
+  color: var(--text-secondary);
+  font-size: 13px;
+  cursor: pointer;
+  white-space: nowrap;
+  flex-shrink: 0;
+  transition: all 0.2s ease;
+  line-height: 1.4;
+}
+
+.vmat-filter-pill:hover {
+  border-color: var(--primary-300);
+  color: var(--text-primary);
+  background: var(--primary-50);
+}
+
+.vmat-filter-pill:focus-visible {
+  outline: 2px solid var(--primary-500);
+  outline-offset: 2px;
+}
+
+.vmat-filter-pill.active {
+  background: var(--primary-500);
+  color: #fff;
+  border-color: var(--primary-500);
+}
+
+.vmat-filter-pill.active .vmat-filter-pill-count {
+  color: rgba(255, 255, 255, 0.75);
+}
+
+.vmat-filter-pill-count {
+  font-size: 11px;
+  opacity: 0.7;
+  color: var(--text-tertiary);
+}
+
+/* ─── TagSelector ──────────────────────────────────────────── */
+
+.vmat-tag-selector-wrapper {
+  position: relative;
+  display: flex;
+  flex-direction: column;
+  gap: 8px;
+}
+
+.vmat-tag-selector {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 6px;
+  padding: 8px;
+  border: 1px solid var(--border-primary);
+  border-radius: var(--radius-md);
+  background: var(--bg-surface);
+  min-height: 42px;
+  cursor: text;
+  transition:
+    border-color 0.2s,
+    box-shadow 0.2s;
+}
+
+.vmat-tag-selector:focus-within {
+  border-color: var(--primary-500);
+  box-shadow: 0 0 0 2px var(--primary-100);
+}
+
+.vmat-tag-selector-input {
+  border: none;
+  outline: none;
+  flex: 1;
+  min-width: 80px;
+  font-size: 13px;
+  background: transparent;
+  color: var(--text-primary);
+}
+
+.vmat-tag-selector-input::placeholder {
+  color: var(--text-placeholder);
+}
+
+.vmat-tag-suggestions {
+  position: absolute;
+  top: 100%;
+  left: 0;
+  right: 0;
+  z-index: 10;
+  margin-top: 4px;
+  background: var(--bg-surface);
+  border: 1px solid var(--border-primary);
+  border-radius: var(--radius-md);
+  box-shadow: var(--shadow-md, 0 4px 12px rgba(0, 0, 0, 0.1));
+  max-height: 160px;
+  overflow-y: auto;
+}
+
+.vmat-tag-suggestion-item {
+  padding: 8px 12px;
+  font-size: 13px;
+  color: var(--text-primary);
+  cursor: pointer;
+  transition: background 0.15s;
+}
+
+.vmat-tag-suggestion-item:hover {
+  background: var(--primary-50);
+}
+
+.vmat-tag-selector-presets {
+  width: 100%;
+  display: flex;
+  gap: 6px;
+  flex-wrap: wrap;
+  margin-top: 4px;
+  padding-top: 8px;
+  border-top: 1px solid var(--border-light);
+}
+
+.vmat-tag-selector-preset {
+  display: inline-flex;
+  align-items: center;
+  gap: 4px;
+  padding: 2px 10px;
+  font-size: 12px;
+  border: 1px solid var(--border-primary);
+  border-radius: 12px;
+  background: var(--bg-surface);
+  color: var(--text-secondary);
+  cursor: pointer;
+  transition: all 0.15s;
+}
+
+.vmat-tag-selector-preset:hover {
+  border-color: var(--primary-300);
+  color: var(--primary-600);
+  background: var(--primary-50);
+}
+
+.vmat-tag-selector-preset.selected {
+  opacity: 0.4;
+  pointer-events: none;
+}
+
+/* ─── 标签溢出 + 空态 ─────────────────────────────────────── */
+
+.vmat-tag-empty {
+  font-size: 12px;
+  color: var(--primary-500);
+  cursor: pointer;
+  padding: 2px 8px;
+  border: 1px dashed var(--primary-300);
+  border-radius: var(--radius-sm);
+  transition: all 0.15s;
+}
+
+.vmat-tag-empty:hover {
+  background: var(--primary-50);
+  border-color: var(--primary-500);
+}
+
+.vmat-tag-overflow {
+  font-style: italic;
+  opacity: 0.7;
+  cursor: default;
+}
+
+/* ── 批量打标签 Popover 自定义输入 ── */
+.vmat-tag-pop-input-row {
+  margin-bottom: 10px;
+}
diff --git a/apps/worker/worker_app/tasks/__init__.py b/apps/worker/worker_app/tasks/__init__.py
index cf7721f22..7c9c315a4 100755
--- a/apps/worker/worker_app/tasks/__init__.py
+++ b/apps/worker/worker_app/tasks/__init__.py
@@ -41,6 +41,10 @@ def __getattr__(name: str):
         from .tts_synthesis import process_tts_synthesis
 
         return process_tts_synthesis
+    elif name == "process_tts_segment_synthesis":
+        from .tts_synthesis import process_tts_segment_synthesis
+
+        return process_tts_segment_synthesis
     elif name == "run_ai_recommend":
         from .ai_tasks import run_ai_recommend
 
@@ -62,6 +66,7 @@ __all__ = [
     "extract_background_task",
     "process_voice_clone",
     "process_tts_synthesis",
+    "process_tts_segment_synthesis",
     "run_ai_recommend",
     "run_generate_cover",
 ]
diff --git a/apps/worker/worker_app/tasks/tts_synthesis.py b/apps/worker/worker_app/tasks/tts_synthesis.py
index e8f5d5e1f..4fad944a4 100644
--- a/apps/worker/worker_app/tasks/tts_synthesis.py
+++ b/apps/worker/worker_app/tasks/tts_synthesis.py
@@ -104,3 +104,77 @@ def process_tts_synthesis(self: Task, job_id: str) -> dict:
     finally:
         if session is not None:
             session.close()
+
+
+@celery_app.task(bind=True, max_retries=2, name="worker.process_tts_segment_synthesis")
+def process_tts_segment_synthesis(self: Task, job_id: str) -> dict:
+    """分段合成轮询任务 — 轮询多个 CosyVoice 子任务并合并音频。
+
+    与 process_tts_synthesis 类似，但超时更长（300s），
+    因为分段任务需要等待所有子任务完成。
+    """
+    session = None
+    try:
+        session = SessionLocal()
+        repo = SQLAlchemyTTSJobRepository(session)
+        workflow = TTSWorkflowService(
+            repository=repo,
+            cosyvoice_service=CosyVoiceService(),
+        )
+
+        updated_job = workflow.poll_and_process_synthesis(job_id, timeout=300)
+        session.commit()
+
+        logger.info(f"TTS segment synthesis completed: job_id={job_id}, " f"audio_url={updated_job.output_audio_url}")
+        return {
+            "ok": True,
+            "job_id": job_id,
+            "audio_url": updated_job.output_audio_url,
+        }
+
+    except Retry:
+        raise
+
+    except CosyVoiceTimeoutError as e:
+        logger.warning(f"TTS segment synthesis timeout for {job_id}: {e}")
+        if session is not None:
+            session.rollback()
+        raise self.retry(exc=e, countdown=60)
+
+    except CosyVoiceError as e:
+        logger.error(f"TTS segment synthesis failed for {job_id}: {e}")
+        if session is not None:
+            session.rollback()
+        try:
+            if session is not None:
+                job = repo.get(job_id)
+                if job is not None:
+                    job.mark_failed(str(e))
+                    repo.update(job)
+                    session.commit()
+        except Exception as inner_e:
+            logger.error(f"Failed to mark job as failed: {inner_e}")
+            if session is not None:
+                session.rollback()
+        return {"ok": False, "job_id": job_id, "error": str(e)}
+
+    except Exception as e:
+        logger.error(f"TTS segment synthesis unexpected error for {job_id}: {e}")
+        if session is not None:
+            session.rollback()
+        try:
+            if session is not None:
+                job = repo.get(job_id)
+                if job is not None:
+                    job.mark_failed(str(e))
+                    repo.update(job)
+                    session.commit()
+        except Exception as inner_e:
+            logger.error(f"Failed to mark job as failed: {inner_e}")
+            if session is not None:
+                session.rollback()
+        return {"ok": False, "job_id": job_id, "error": str(e)}
+
+    finally:
+        if session is not None:
+            session.close()
diff --git a/docs/schema-metadata-snapshot.json b/docs/schema-metadata-snapshot.json
index e0dacee1f..769beb8ff 100644
--- a/docs/schema-metadata-snapshot.json
+++ b/docs/schema-metadata-snapshot.json
@@ -95,6 +95,39 @@
         "id"
       ]
     },
+    "asset_tags": {
+      "columns": [
+        {
+          "index": false,
+          "name": "asset_id",
+          "nullable": false,
+          "primary_key": true,
+          "type": "VARCHAR(36)",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "tag_id",
+          "nullable": false,
+          "primary_key": true,
+          "type": "VARCHAR(36)",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "created_at",
+          "nullable": false,
+          "primary_key": false,
+          "type": "DATETIME",
+          "unique": false
+        }
+      ],
+      "indexes": [],
+      "primary_key": [
+        "asset_id",
+        "tag_id"
+      ]
+    },
     "assets": {
       "columns": [
         {
@@ -2056,6 +2089,54 @@
         "id"
       ]
     },
+    "tags": {
+      "columns": [
+        {
+          "index": false,
+          "name": "id",
+          "nullable": false,
+          "primary_key": true,
+          "type": "VARCHAR(36)",
+          "unique": false
+        },
+        {
+          "index": true,
+          "name": "user_id",
+          "nullable": false,
+          "primary_key": false,
+          "type": "VARCHAR(36)",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "name",
+          "nullable": false,
+          "primary_key": false,
+          "type": "VARCHAR(100)",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "created_at",
+          "nullable": false,
+          "primary_key": false,
+          "type": "DATETIME",
+          "unique": false
+        }
+      ],
+      "indexes": [
+        {
+          "columns": [
+            "user_id"
+          ],
+          "name": "ix_tags_user_id",
+          "unique": false
+        }
+      ],
+      "primary_key": [
+        "id"
+      ]
+    },
     "template_categories": {
       "columns": [
         {
diff --git a/packages/adapters/in_memory/asset_repository.py b/packages/adapters/in_memory/asset_repository.py
index 7b6874a43..6a3a64043 100755
--- a/packages/adapters/in_memory/asset_repository.py
+++ b/packages/adapters/in_memory/asset_repository.py
@@ -42,3 +42,37 @@ class InMemoryAssetRepository:
             del self._assets[asset_id]
             return True
         return False
+
+    def batch_delete(self, asset_ids: list[str]) -> int:
+        """批量删除素材，返回实际删除数量。"""
+        count = 0
+        for aid in asset_ids:
+            if aid in self._assets:
+                del self._assets[aid]
+                count += 1
+        return count
+
+    def find_by_project(
+        self,
+        project_id: str,
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Asset]:
+        items = [a for a in self._assets.values() if a.project_id == project_id]
+        return items[skip : skip + limit]
+
+    def find_by_id(self, asset_id: str) -> Asset | None:
+        return self._assets.get(asset_id)
+
+    def find_by_tag_ids(
+        self,
+        tag_ids: list[str],
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Asset]:
+        """查找包含所有指定标签的素材。"""
+        if not tag_ids:
+            return []
+        tag_set = set(tag_ids)
+        items = [a for a in self._assets.values() if tag_set.issubset(set(a.tag_ids))]
+        return items[skip : skip + limit]
diff --git a/packages/adapters/in_memory/tag_repository.py b/packages/adapters/in_memory/tag_repository.py
new file mode 100644
index 000000000..451b8848f
--- /dev/null
+++ b/packages/adapters/in_memory/tag_repository.py
@@ -0,0 +1,40 @@
+"""标签 InMemory 仓储实现。"""
+
+from packages.domain import Tag
+
+
+class InMemoryTagRepository:
+    def __init__(self):
+        self._tags: dict[str, Tag] = {}
+
+    def create(self, tag: Tag) -> Tag:
+        self._tags[tag.id] = tag
+        return tag
+
+    def get(self, tag_id: str) -> Tag | None:
+        return self._tags.get(tag_id)
+
+    def find_by_name(self, user_id: str, name: str) -> Tag | None:
+        for tag in self._tags.values():
+            if tag.user_id == user_id and tag.name == name:
+                return tag
+        return None
+
+    def list_by_user(
+        self,
+        user_id: str,
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Tag]:
+        tags = [tag for tag in self._tags.values() if tag.user_id == user_id]
+        tags.sort(key=lambda t: t.created_at, reverse=True)
+        return tags[skip : skip + limit]
+
+    def count_by_user(self, user_id: str) -> int:
+        return sum(1 for tag in self._tags.values() if tag.user_id == user_id)
+
+    def delete(self, tag_id: str) -> bool:
+        if tag_id in self._tags:
+            del self._tags[tag_id]
+            return True
+        return False
diff --git a/packages/adapters/sqlalchemy_impl/asset_repository.py b/packages/adapters/sqlalchemy_impl/asset_repository.py
index d14fa3e9f..ab6083043 100644
--- a/packages/adapters/sqlalchemy_impl/asset_repository.py
+++ b/packages/adapters/sqlalchemy_impl/asset_repository.py
@@ -3,7 +3,7 @@ from datetime import datetime, timezone
 
 from sqlalchemy.orm import Session
 
-from packages.adapters.sqlalchemy_impl.models import AssetModel
+from packages.adapters.sqlalchemy_impl.models import AssetModel, AssetTagModel
 from packages.domain import Asset, AssetStatus, ClassificationStatus
 
 
@@ -87,6 +87,8 @@ class SQLAlchemyAssetRepository:
             updated_at=now,
         )
         self.session.add(model)
+        self.session.flush()
+        self._sync_asset_tags(asset.id, asset.tag_ids)
         self.session.commit()
         return asset
 
@@ -109,6 +111,8 @@ class SQLAlchemyAssetRepository:
         model.quality_score = asset.quality_score
         model.uploaded_by_user_id = asset.uploaded_by_user_id or model.uploaded_by_user_id
         model.updated_at = datetime.now(timezone.utc)
+        self.session.flush()
+        self._sync_asset_tags(asset.id, asset.tag_ids)
         self.session.commit()
         return asset
 
@@ -120,6 +124,14 @@ class SQLAlchemyAssetRepository:
             return True
         return False
 
+    def batch_delete(self, asset_ids: list[str]) -> int:
+        """批量删除素材，返回实际删除数量。"""
+        if not asset_ids:
+            return 0
+        count = self.session.query(AssetModel).filter(AssetModel.id.in_(asset_ids)).delete(synchronize_session=False)
+        self.session.commit()
+        return count
+
     def count_by_project(self, project_id: str) -> int:
         return self.session.query(AssetModel).filter(AssetModel.project_id == project_id).count()
 
@@ -195,6 +207,11 @@ class SQLAlchemyAssetRepository:
                 "audio": "audio/mpeg",
                 "image": "image/jpeg",
             }.get(mime_type, mime_type)
+        # 查询关联的 tag_ids
+        tag_ids = [
+            row.tag_id
+            for row in self.session.query(AssetTagModel.tag_id).filter(AssetTagModel.asset_id == model.id).all()
+        ]
         return Asset(
             id=model.id,
             project_id=model.project_id,
@@ -214,6 +231,39 @@ class SQLAlchemyAssetRepository:
             quality_score=model.quality_score,
             uploaded_by_user_id=model.uploaded_by_user_id,
             metadata=metadata,
+            tag_ids=tag_ids,
             created_at=model.created_at,
             updated_at=model.updated_at,
         )
+
+    def _sync_asset_tags(self, asset_id: str, tag_ids: list[str]) -> None:
+        """同步素材-标签关联表（全量替换）。"""
+        self.session.query(AssetTagModel).filter(AssetTagModel.asset_id == asset_id).delete(synchronize_session=False)
+        for tag_id in tag_ids:
+            self.session.add(AssetTagModel(asset_id=asset_id, tag_id=tag_id))
+
+    def find_by_tag_ids(
+        self,
+        tag_ids: list[str],
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Asset]:
+        """查找包含所有指定标签的素材。"""
+        if not tag_ids:
+            return []
+        from sqlalchemy import func
+
+        # 找出同时拥有所有指定 tag_id 的 asset_id
+        tag_set = set(tag_ids)
+        asset_ids = (
+            self.session.query(AssetTagModel.asset_id)
+            .filter(AssetTagModel.tag_id.in_(tag_set))
+            .group_by(AssetTagModel.asset_id)
+            .having(func.count(AssetTagModel.tag_id) == len(tag_set))
+            .all()
+        )
+        ids = [row[0] for row in asset_ids]
+        if not ids:
+            return []
+        models = self.session.query(AssetModel).filter(AssetModel.id.in_(ids)).offset(skip).limit(limit).all()
+        return [self._to_domain(m) for m in models]
diff --git a/packages/adapters/sqlalchemy_impl/models.py b/packages/adapters/sqlalchemy_impl/models.py
index 982875e00..27a07513e 100755
--- a/packages/adapters/sqlalchemy_impl/models.py
+++ b/packages/adapters/sqlalchemy_impl/models.py
@@ -90,6 +90,29 @@ class AssetModel(Base):
     updated_at = Column(DateTime, nullable=False, default=lambda: datetime.now(timezone.utc))
 
 
+class TagModel(Base):
+    """标签 ORM 模型。"""
+
+    __tablename__ = "tags"
+
+    id = Column(String(36), primary_key=True)
+    user_id = Column(String(36), nullable=False, index=True)
+    name = Column(String(100), nullable=False)
+    created_at = Column(DateTime, nullable=False, default=lambda: datetime.now(timezone.utc))
+
+    __table_args__ = (UniqueConstraint("user_id", "name", name="uq_tags_user_name"),)
+
+
+class AssetTagModel(Base):
+    """素材-标签关联表 ORM 模型。"""
+
+    __tablename__ = "asset_tags"
+
+    asset_id = Column(String(36), primary_key=True)
+    tag_id = Column(String(36), primary_key=True)
+    created_at = Column(DateTime, nullable=False, default=lambda: datetime.now(timezone.utc))
+
+
 class EditTemplateModel(Base):
     """Phase 8 剪辑模板 ORM 模型
 
diff --git a/packages/adapters/sqlalchemy_impl/tag_repository.py b/packages/adapters/sqlalchemy_impl/tag_repository.py
new file mode 100644
index 000000000..4353484ca
--- /dev/null
+++ b/packages/adapters/sqlalchemy_impl/tag_repository.py
@@ -0,0 +1,73 @@
+"""标签 SQLAlchemy 仓储实现。"""
+
+from sqlalchemy.orm import Session
+
+from packages.adapters.sqlalchemy_impl.models import AssetTagModel, TagModel
+from packages.domain import Tag
+
+
+class SQLAlchemyTagRepository:
+    def __init__(self, session: Session):
+        self.session = session
+
+    def create(self, tag: Tag) -> Tag:
+        model = TagModel(
+            id=tag.id,
+            user_id=tag.user_id,
+            name=tag.name,
+            created_at=tag.created_at,
+        )
+        self.session.add(model)
+        self.session.commit()
+        return tag
+
+    def get(self, tag_id: str) -> Tag | None:
+        model = self.session.query(TagModel).filter(TagModel.id == tag_id).first()
+        if model is None:
+            return None
+        return self._to_domain(model)
+
+    def find_by_name(self, user_id: str, name: str) -> Tag | None:
+        model = self.session.query(TagModel).filter(TagModel.user_id == user_id, TagModel.name == name).first()
+        if model is None:
+            return None
+        return self._to_domain(model)
+
+    def list_by_user(
+        self,
+        user_id: str,
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Tag]:
+        models = (
+            self.session.query(TagModel)
+            .filter(TagModel.user_id == user_id)
+            .order_by(TagModel.created_at.desc())
+            .offset(skip)
+            .limit(limit)
+            .all()
+        )
+        return [self._to_domain(m) for m in models]
+
+    def count_by_user(self, user_id: str) -> int:
+        return self.session.query(TagModel).filter(TagModel.user_id == user_id).count()
+
+    def delete(self, tag_id: str) -> bool:
+        # 先清理关联表
+        self.session.query(AssetTagModel).filter(AssetTagModel.tag_id == tag_id).delete(synchronize_session=False)
+        model = self.session.query(TagModel).filter(TagModel.id == tag_id).first()
+        if model is None:
+            self.session.commit()
+            return False
+        self.session.delete(model)
+        self.session.commit()
+        return True
+
+    @staticmethod
+    def _to_domain(model: TagModel) -> Tag:
+        return Tag(
+            id=model.id,
+            user_id=model.user_id,
+            name=model.name,
+            created_at=model.created_at,
+        )
diff --git a/packages/application/tts_job/audio_merger.py b/packages/application/tts_job/audio_merger.py
new file mode 100644
index 000000000..0d583f821
--- /dev/null
+++ b/packages/application/tts_job/audio_merger.py
@@ -0,0 +1,96 @@
+"""FFmpeg 音频合并器 — P1 长文本分段合成。
+
+将多个分段音频文件合并为一个完整音频文件。
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+import shutil
+import subprocess
+import tempfile
+
+logger = logging.getLogger(__name__)
+
+
+class AudioMergeError(Exception):
+    """音频合并异常。"""
+
+    pass
+
+
+class AudioMerger:
+    """使用 FFmpeg 合并多个音频文件。"""
+
+    def merge(self, audio_paths: list[str], output_format: str = "mp3") -> bytes:
+        """合并多个音频文件，返回合并后的音频数据。
+
+        使用 FFmpeg concat demuxer 按顺序拼接音频。
+        所有输入文件必须为相同格式和采样率。
+
+        Args:
+            audio_paths: 音频文件路径列表（按合成顺序）
+            output_format: 输出格式（mp3/wav/pcm）
+
+        Returns:
+            合并后的音频文件字节数据
+
+        Raises:
+            AudioMergeError: 合并失败
+        """
+        if not audio_paths:
+            raise AudioMergeError("没有可合并的音频文件")
+
+        if len(audio_paths) == 1:
+            with open(audio_paths[0], "rb") as f:
+                return f.read()
+
+        temp_dir = tempfile.mkdtemp(prefix="tts_merge_")
+        try:
+            # 生成 concat demuxer 列表文件
+            list_path = os.path.join(temp_dir, "concat_list.txt")
+            with open(list_path, "w") as f:
+                for path in audio_paths:
+                    # FFmpeg concat 文件需要 file: 前缀，路径中的 ' 和 \n 需转义
+                    escaped = path.replace("'", "'\\''").replace("\n", "\\n")
+                    f.write(f"file '{escaped}'\n")
+
+            output_path = os.path.join(temp_dir, f"merged.{output_format}")
+
+            cmd = [
+                "ffmpeg",
+                "-y",
+                "-f",
+                "concat",
+                "-safe",
+                "0",
+                "-i",
+                list_path,
+                "-c",
+                "copy",
+                output_path,
+            ]
+
+            result = subprocess.run(
+                cmd,
+                capture_output=True,
+                text=True,
+                timeout=120,
+            )
+
+            if result.returncode != 0:
+                logger.error(f"FFmpeg 合并失败: stderr={result.stderr}")
+                raise AudioMergeError(f"FFmpeg 合并失败: {result.stderr[:500]}")
+
+            with open(output_path, "rb") as f:
+                return f.read()
+
+        except subprocess.TimeoutExpired:
+            raise AudioMergeError("FFmpeg 合并超时（120 秒）")
+        except AudioMergeError:
+            raise
+        except Exception as e:
+            raise AudioMergeError(f"音频合并失败: {e}")
+        finally:
+            shutil.rmtree(temp_dir, ignore_errors=True)
diff --git a/packages/application/tts_job/streaming_service.py b/packages/application/tts_job/streaming_service.py
new file mode 100644
index 000000000..4c50d4dec
--- /dev/null
+++ b/packages/application/tts_job/streaming_service.py
@@ -0,0 +1,251 @@
+"""P2: TTS 流式合成服务 — WebSocket 实时音频推送。
+
+通过 WebSocket 将合成音频以二进制帧实时推送给客户端。
+- 短文本（≤500 字）：合成完整音频后分块推送
+- 长文本（>500 字）：分段并发合成，逐段推送音频
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+from typing import Any, Optional
+
+import httpx
+
+from packages.application.cosyvoice_service import CosyVoiceError, CosyVoiceService
+from packages.application.tts_job.text_splitter import split_text
+
+logger = logging.getLogger(__name__)
+
+# WebSocket 二进制帧块大小（4KB）
+_AUDIO_CHUNK_SIZE = 4096
+# 分段并发上限
+_MAX_STREAMING_SEGMENT_WORKERS = 5
+# 长文本分段阈值
+_SEGMENT_THRESHOLD = 500
+# WebSocket 最大文本长度
+_MAX_TEXT_LENGTH = 10000
+
+
+class TTSStreamingError(Exception):
+    """TTS 流式合成异常。"""
+
+    pass
+
+
+class TTSStreamingService:
+    """TTS 流式合成服务。
+
+    通过 WebSocket 实时推送合成音频。
+    使用 CosyVoiceService（同步 REST API）合成，
+    通过 asyncio.to_thread 桥接到异步 WebSocket。
+    """
+
+    def __init__(self, cosyvoice_service: CosyVoiceService) -> None:
+        self._cosyvoice = cosyvoice_service
+
+    async def synthesize_and_stream(self, websocket: Any, params: dict) -> None:
+        """根据文本长度选择流式合成策略。
+
+        Args:
+            websocket: FastAPI WebSocket 连接
+            params: 合成参数（text, voice_id, sample_rate, format, speed）
+        """
+        text = params.get("text", "")
+        if not text:
+            await self._send_json(websocket, {"type": "error", "message": "文本不能为空"})
+            return
+
+        if len(text) > _MAX_TEXT_LENGTH:
+            await self._send_json(
+                websocket,
+                {"type": "error", "message": f"文本过长，最大 {_MAX_TEXT_LENGTH} 字"},
+            )
+            return
+
+        if len(text) <= _SEGMENT_THRESHOLD:
+            await self._stream_short_text(websocket, params)
+        else:
+            await self._stream_long_text(websocket, params)
+
+    # ── 短文本流式合成 ────────────────────────────────────────
+
+    async def _stream_short_text(self, websocket: Any, params: dict) -> None:
+        """短文本：合成完整音频后分块推送。"""
+        text = params["text"]
+        voice_id = params.get("voice_id", "")
+        sample_rate = params.get("sample_rate", 0)
+        audio_format = params.get("format", "mp3")
+        speed = params.get("speed", 1.0)
+
+        await self._send_json(
+            websocket,
+            {"type": "started", "segment_count": 1, "total_segments": 1},
+        )
+
+        # 在线程池中执行同步合成
+        try:
+            result = await asyncio.to_thread(
+                self._cosyvoice.submit_synthesize_task,
+                text=text,
+                voice_id=voice_id,
+                sample_rate=sample_rate,
+                format=audio_format,
+                speed=speed,
+            )
+        except CosyVoiceError as e:
+            logger.error(f"流式合成失败: {e}")
+            await self._send_json(websocket, {"type": "error", "message": str(e)})
+            return
+        except Exception as e:
+            logger.error(f"流式合成意外错误: {e}")
+            await self._send_json(websocket, {"type": "error", "message": f"合成失败: {e}"})
+            return
+
+        audio_url = result.get("audio_url", "")
+        if not audio_url:
+            await self._send_json(websocket, {"type": "error", "message": "合成未返回音频 URL"})
+            return
+
+        # 下载并流式推送音频
+        try:
+            audio_data = await asyncio.to_thread(self._download_audio, audio_url)
+            total_bytes = await self._stream_audio_chunks(websocket, audio_data)
+
+            await self._send_json(
+                websocket,
+                {
+                    "type": "done",
+                    "duration": result.get("duration", 0.0),
+                    "file_size": total_bytes,
+                    "format": audio_format,
+                },
+            )
+        except Exception as e:
+            logger.error(f"音频流式推送失败: {e}")
+            await self._send_json(websocket, {"type": "error", "message": f"音频推送失败: {e}"})
+
+    # ── 长文本分段流式合成 ────────────────────────────────────
+
+    async def _stream_long_text(self, websocket: Any, params: dict) -> None:
+        """长文本：分段并发合成，逐段推送音频。"""
+        text = params["text"]
+        voice_id = params.get("voice_id", "")
+        sample_rate = params.get("sample_rate", 0)
+        audio_format = params.get("format", "mp3")
+        speed = params.get("speed", 1.0)
+
+        segments = split_text(text, max_chars=_SEGMENT_THRESHOLD)
+        segment_count = len(segments)
+
+        logger.info(f"流式分段合成: 原文={len(text)}字, 段数={segment_count}")
+
+        await self._send_json(
+            websocket,
+            {"type": "started", "segment_count": segment_count, "total_segments": segment_count},
+        )
+
+        # 并发合成所有分段，按顺序流式推送
+        queue: asyncio.Queue[tuple[int, Optional[bytes], Optional[str]]] = asyncio.Queue()
+        completed_count = 0
+
+        async def _synthesize_one(idx: int, seg_text: str) -> None:
+            """合成单个分段并放入队列。"""
+            try:
+                result = await asyncio.to_thread(
+                    self._cosyvoice.submit_synthesize_task,
+                    text=seg_text,
+                    voice_id=voice_id,
+                    sample_rate=sample_rate,
+                    format=audio_format,
+                    speed=speed,
+                )
+                audio_url = result.get("audio_url", "")
+                if audio_url:
+                    audio_data = await asyncio.to_thread(self._download_audio, audio_url)
+                    await queue.put((idx, audio_data, None))
+                else:
+                    await queue.put((idx, None, "合成未返回音频 URL"))
+            except Exception as e:
+                await queue.put((idx, None, str(e)))
+
+        # 启动并发合成任务
+        workers = [asyncio.create_task(_synthesize_one(idx, seg)) for idx, seg in enumerate(segments)]
+
+        # 按顺序消费队列，流式推送
+        total_bytes = 0
+        total_duration = 0.0
+        consumed = 0
+
+        try:
+            while consumed < segment_count:
+                idx, audio_data, error = await queue.get()
+                consumed += 1
+
+                if error:
+                    logger.error(f"分段 {idx + 1} 合成失败: {error}")
+                    await self._send_json(
+                        websocket,
+                        {"type": "error", "message": f"分段 {idx + 1} 合成失败: {error}"},
+                    )
+                    # 取消剩余 worker
+                    for w in workers:
+                        w.cancel()
+                    return
+
+                if audio_data:
+                    seg_bytes = await self._stream_audio_chunks(websocket, audio_data)
+                    total_bytes += seg_bytes
+
+                await self._send_json(
+                    websocket,
+                    {"type": "segment_done", "segment": idx + 1, "total": segment_count},
+                )
+
+            # 等待所有 worker 完成
+            await asyncio.gather(*workers, return_exceptions=True)
+
+            await self._send_json(
+                websocket,
+                {
+                    "type": "done",
+                    "duration": total_duration,
+                    "file_size": total_bytes,
+                    "format": audio_format,
+                },
+            )
+
+        except Exception as e:
+            logger.error(f"流式分段推送失败: {e}")
+            await self._send_json(websocket, {"type": "error", "message": f"推送失败: {e}"})
+            for w in workers:
+                w.cancel()
+
+    # ── 工具方法 ────────────────────────────────────────────
+
+    def _download_audio(self, url: str) -> bytes:
+        """下载音频数据。"""
+        resp = httpx.get(url, timeout=60.0, follow_redirects=True)
+        resp.raise_for_status()
+        return resp.content
+
+    async def _stream_audio_chunks(self, websocket: Any, audio_data: bytes) -> int:
+        """将音频数据分块通过 WebSocket 推送。
+
+        Returns:
+            推送的总字节数
+        """
+        total = 0
+        for offset in range(0, len(audio_data), _AUDIO_CHUNK_SIZE):
+            chunk = audio_data[offset : offset + _AUDIO_CHUNK_SIZE]
+            await websocket.send_bytes(chunk)
+            total += len(chunk)
+        return total
+
+    async def _send_json(self, websocket: Any, data: dict) -> None:
+        """安全发送 JSON 帧。"""
+        try:
+            await websocket.send_json(data)
+        except Exception:
+            pass
diff --git a/packages/application/tts_job/text_splitter.py b/packages/application/tts_job/text_splitter.py
new file mode 100644
index 000000000..a34135bb1
--- /dev/null
+++ b/packages/application/tts_job/text_splitter.py
@@ -0,0 +1,70 @@
+"""长文本分段工具 — P1 长文本分段合成。
+
+将超过阈值的文本按句子边界分段，供 CosyVoice 并发合成后合并。
+"""
+
+from __future__ import annotations
+
+# 中文句子结束符（含全角/半角）
+_SENTENCE_ENDS = frozenset("。！？；\n.!?;")
+
+
+def split_text(text: str, max_chars: int = 500) -> list[str]:
+    """将文本分段，每段不超过 max_chars 个字符。
+
+    优先在句子边界（句号、问号、感叹号、换行符）处分段。
+    若单个句子超过 max_chars，则在逗号等次级标点处拆分。
+    若仍超长，则硬切。
+
+    Args:
+        text: 待分段文本
+        max_chars: 每段最大字符数
+
+    Returns:
+        分段列表，每段 ≤ max_chars。文本为空时返回空列表。
+    """
+    text = text.strip()
+    if not text:
+        return []
+    if len(text) <= max_chars:
+        return [text]
+
+    segments: list[str] = []
+    current = ""
+
+    for char in text:
+        current += char
+        if char in _SENTENCE_ENDS and len(current) >= 50:
+            # 句子边界且长度合理，切段
+            segments.append(current.strip())
+            current = ""
+        elif len(current) >= max_chars:
+            # 达到上限，强制切段
+            segments.append(current.strip())
+            current = ""
+
+    if current.strip():
+        segments.append(current.strip())
+
+    # 合并过短的段（< 50 字符且不是最后一段），减少 API 调用次数
+    merged: list[str] = []
+    buffer = ""
+    for seg in segments:
+        if buffer:
+            combined = buffer + seg
+            if len(combined) <= max_chars:
+                buffer = combined
+                continue
+            merged.append(buffer)
+            buffer = ""
+        if len(seg) < 50:
+            buffer = seg
+        else:
+            merged.append(seg)
+    if buffer:
+        if merged and len(merged[-1]) + len(buffer) <= max_chars:
+            merged[-1] = merged[-1] + buffer
+        else:
+            merged.append(buffer)
+
+    return [s for s in merged if s]
diff --git a/packages/application/tts_job/workflow.py b/packages/application/tts_job/workflow.py
index b49e60eef..1b6e9c36e 100644
--- a/packages/application/tts_job/workflow.py
+++ b/packages/application/tts_job/workflow.py
@@ -9,19 +9,35 @@
 
 from __future__ import annotations
 
+import io
 import logging
+import os
+import shutil
+import tempfile
+import time
+from concurrent.futures import ThreadPoolExecutor, as_completed
 from typing import Optional
 
+import httpx
+
 from packages.application.cosyvoice_service import (
     CosyVoiceAuthError,
     CosyVoiceError,
     CosyVoiceService,
 )
+from packages.application.tts_job.audio_merger import AudioMergeError, AudioMerger
+from packages.application.tts_job.text_splitter import split_text
 from packages.domain.tts_job import TTSJob, TTSJobStatus
 from packages.ports.tts_job_repository import TTSJobRepository
+from packages.shared.storage import SharedStorageService, get_shared_storage_service
 
 logger = logging.getLogger(__name__)
 
+# 长文本分段阈值：超过此字符数自动分段合成
+_SEGMENT_THRESHOLD = 500
+# 分段并发上限
+_MAX_SEGMENT_WORKERS = 5
+
 
 class TTSWorkflowError(Exception):
     """TTS 合成工作流异常。"""
@@ -46,9 +62,55 @@ class TTSWorkflowService:
         self,
         repository: TTSJobRepository,
         cosyvoice_service: CosyVoiceService,
+        storage_service: Optional[SharedStorageService] = None,
     ) -> None:
         self.repository = repository
         self.cosyvoice_service = cosyvoice_service
+        self._storage_service = storage_service
+
+    @property
+    def _storage(self) -> SharedStorageService:
+        if self._storage_service is None:
+            self._storage_service = get_shared_storage_service()
+        return self._storage_service
+
+    def _transfer_audio_to_oss(
+        self,
+        temp_url: str,
+        user_id: str,
+        job_id: str,
+        audio_format: str = "mp3",
+    ) -> tuple[str, str]:
+        """下载 CosyVoice 临时音频并转存到 OSS。
+
+        Returns:
+            (permanent_url, storage_key) 元组。
+            转存失败时回退到原始临时 URL，storage_key 为空字符串。
+        """
+        storage_key = f"tts-outputs/{user_id}/{job_id}.{audio_format}"
+        content_type_map = {
+            "mp3": "audio/mpeg",
+            "wav": "audio/wav",
+            "pcm": "audio/pcm",
+            "opus": "audio/opus",
+        }
+        content_type = content_type_map.get(audio_format, "application/octet-stream")
+
+        try:
+            # 下载临时音频
+            resp = httpx.get(temp_url, timeout=60.0, follow_redirects=True)
+            resp.raise_for_status()
+            audio_data = resp.content
+
+            # 上传到 OSS
+            file_obj = io.BytesIO(audio_data)
+            permanent_url = self._storage.upload_file(file_obj, storage_key, content_type=content_type)
+            logger.info(f"音频转存 OSS 成功: job_id={job_id}, " f"storage_key={storage_key}, size={len(audio_data)}")
+            return permanent_url, storage_key
+
+        except Exception as e:
+            logger.warning(f"音频转存 OSS 失败，使用临时 URL: " f"job_id={job_id}, error={e}")
+            return temp_url, ""
 
     def start_synthesis(
         self,
@@ -80,6 +142,10 @@ class TTSWorkflowService:
         job.mark_processing()
         job = self.repository.update(job)
 
+        # 长文本自动分段合成
+        if len(job.input_text) > _SEGMENT_THRESHOLD:
+            return self._start_segment_synthesis(job)
+
         try:
             submit_result = self.cosyvoice_service.submit_synthesize_task(
                 text=job.input_text,
@@ -93,17 +159,19 @@ class TTSWorkflowService:
             job_metadata["cosyvoice_task_id"] = submit_result.get("task_id", "")
             job_metadata["cosyvoice_request_id"] = submit_result.get("request_id", "")
 
-            # 如果 CosyVoice 同步返回了 audio_url，直接标记完成
+            # 如果 CosyVoice 同步返回了 audio_url，转存 OSS 后标记完成
             audio_url = submit_result.get("audio_url", "")
             if audio_url:
+                permanent_url, storage_key = self._transfer_audio_to_oss(audio_url, job.user_id, job.id, job.format)
                 job.mark_completed(
-                    output_audio_url=audio_url,
+                    output_audio_url=permanent_url,
+                    output_audio_key=storage_key,
                     duration=submit_result.get("duration", 0.0),
                     file_size=submit_result.get("file_size", 0),
                 )
                 job.metadata = job_metadata
                 job = self.repository.update(job)
-                logger.info(f"TTS 合成同步完成: job_id={job.id}, audio_url={audio_url}")
+                logger.info(f"TTS 合成同步完成: job_id={job.id}, audio_url={permanent_url}")
                 return job
 
             job.metadata = job_metadata
@@ -133,6 +201,11 @@ class TTSWorkflowService:
         if job is None:
             raise TTSJobNotFoundError(f"TTS job {job_id} not found")
 
+        # 检查是否为分段合成任务
+        segment_task_ids = (job.metadata or {}).get("segment_task_ids", [])
+        if segment_task_ids:
+            return self._poll_segment_tasks(job)
+
         task_id = (job.metadata or {}).get("cosyvoice_task_id", "")
         if not task_id:
             raise ValueError(f"TTSJob {job_id} has no cosyvoice_task_id in metadata")
@@ -171,13 +244,17 @@ class TTSWorkflowService:
         if job is None:
             raise TTSJobNotFoundError(f"TTS job {job_id} not found")
 
+        # 转存音频到 OSS，获取永久 URL
+        permanent_url, storage_key = self._transfer_audio_to_oss(audio_url, job.user_id, job.id, job.format)
+
         job.mark_completed(
-            output_audio_url=audio_url,
+            output_audio_url=permanent_url,
+            output_audio_key=storage_key,
             duration=duration,
             file_size=file_size,
         )
         job = self.repository.update(job)
-        logger.info(f"TTS 合成成功: job_id={job_id}, audio_url={audio_url}")
+        logger.info(f"TTS 合成成功: job_id={job_id}, audio_url={permanent_url}")
         return job
 
     def process_synthesis_failure(self, job_id: str, error_message: str) -> TTSJob:
@@ -201,3 +278,224 @@ class TTSWorkflowService:
         job = self.repository.update(job)
         logger.error(f"TTS 合成失败: job_id={job_id}, error={error_message}")
         return job
+
+    # ── P1: 长文本分段合成 ─────────────────────────────────────
+
+    def _upload_merged_to_oss(
+        self, merged_data: bytes, user_id: str, job_id: str, audio_format: str
+    ) -> tuple[str, str]:
+        """上传合并后的音频数据到 OSS。
+
+        Returns:
+            (permanent_url, storage_key) 元组。
+            上传失败时返回 ("", "")。
+        """
+        storage_key = f"tts-outputs/{user_id}/{job_id}.{audio_format}"
+        content_type_map = {
+            "mp3": "audio/mpeg",
+            "wav": "audio/wav",
+            "pcm": "audio/pcm",
+            "opus": "audio/opus",
+        }
+        content_type = content_type_map.get(audio_format, "application/octet-stream")
+        try:
+            file_obj = io.BytesIO(merged_data)
+            permanent_url = self._storage.upload_file(file_obj, storage_key, content_type=content_type)
+            return permanent_url, storage_key
+        except Exception as e:
+            logger.warning(f"分段合并音频转存 OSS 失败: job_id={job_id}, error={e}")
+            return "", ""
+
+    def _start_segment_synthesis(self, job: TTSJob) -> TTSJob:
+        """长文本分段合成入口。
+
+        将文本分段后并发提交到 CosyVoice，根据同步/异步结果走不同路径。
+        """
+        segments = split_text(job.input_text, max_chars=_SEGMENT_THRESHOLD)
+        logger.info(f"长文本分段合成: job_id={job.id}, " f"原文={len(job.input_text)}字, 段数={len(segments)}")
+
+        # 记录分段信息到 metadata
+        job_metadata = dict(job.metadata)
+        job_metadata["segment_count"] = len(segments)
+
+        # 并发提交所有分段
+        results = self._submit_segments_concurrent(segments, job)
+        if results is None:
+            # 提交阶段已失败，_submit_segments_concurrent 内部已标记 failed
+            return self.repository.get(job.id)
+
+        # 判断同步还是异步
+        has_audio_urls = any(r.get("audio_url", "") for r in results)
+        has_task_ids = any(r.get("task_id", "") for r in results)
+
+        if has_audio_urls and not has_task_ids:
+            # 所有分段同步返回音频，直接合并
+            return self._process_segments_sync(job, results)
+
+        # 异步路径：保存各分段的 task_id 供后续轮询
+        segment_task_ids = [r.get("task_id", "") for r in results]
+        segment_audio_urls = [r.get("audio_url", "") for r in results]
+        job_metadata["segment_task_ids"] = segment_task_ids
+        job_metadata["segment_audio_urls"] = segment_audio_urls
+        job_metadata["segment_format"] = job.format
+
+        job.metadata = job_metadata
+        job = self.repository.update(job)
+        logger.info(f"分段合成任务已提交（异步）: job_id={job.id}, " f"段数={len(segments)}")
+        return job
+
+    def _submit_segments_concurrent(self, segments: list[str], job: TTSJob) -> list[dict] | None:
+        """并发提交分段合成任务。
+
+        Returns:
+            各分段的结果列表（保持顺序），提交失败时返回 None。
+        """
+        max_workers = min(len(segments), _MAX_SEGMENT_WORKERS)
+        results: list[dict | None] = [None] * len(segments)
+
+        with ThreadPoolExecutor(max_workers=max_workers) as executor:
+            future_to_idx = {}
+            for idx, segment_text in enumerate(segments):
+                future = executor.submit(
+                    self.cosyvoice_service.submit_synthesize_task,
+                    text=segment_text,
+                    voice_id=job.voice_id,
+                    sample_rate=job.sample_rate,
+                    format=job.format,
+                )
+                future_to_idx[future] = idx
+
+            for future in as_completed(future_to_idx):
+                idx = future_to_idx[future]
+                try:
+                    results[idx] = future.result()
+                except Exception as e:
+                    logger.error(f"分段合成提交失败: job_id={job.id}, " f"segment={idx}, error={e}")
+                    self._handle_segment_failure(job, f"分段 {idx + 1} 合成提交失败: {e}")
+                    return None
+
+        return results  # type: ignore[return-value]
+
+    def _process_segments_sync(self, job: TTSJob, results: list[dict]) -> TTSJob:
+        """同步路径：所有分段已返回 audio_url，下载合并后转存 OSS。"""
+        merged_data, total_duration = self._download_and_merge_segments(results, job)
+
+        # 直接上传合并后的音频 bytes 到 OSS
+        permanent_url, storage_key = self._upload_merged_to_oss(merged_data, job.user_id, job.id, job.format)
+
+        job.mark_completed(
+            output_audio_url=permanent_url,
+            output_audio_key=storage_key,
+            duration=total_duration,
+            file_size=len(merged_data),
+        )
+        job = self.repository.update(job)
+        logger.info(f"分段合成完成: job_id={job.id}, " f"merged_size={len(merged_data)}, duration={total_duration:.1f}")
+        return job
+
+    def _download_and_merge_segments(self, results: list[dict], job: TTSJob) -> tuple[bytes, float]:
+        """下载各分段音频并合并。
+
+        Returns:
+            (merged_audio_bytes, total_duration)
+        """
+        temp_dir = tempfile.mkdtemp(prefix="tts_segments_")
+        try:
+            audio_paths: list[str] = []
+            total_duration = 0.0
+
+            for idx, result in enumerate(results):
+                audio_url = result.get("audio_url", "")
+                if not audio_url:
+                    raise TTSWorkflowError(f"分段 {idx + 1} 没有返回 audio_url")
+
+                total_duration += result.get("duration", 0.0)
+
+                # 下载分段音频到临时文件
+                resp = httpx.get(audio_url, timeout=60.0, follow_redirects=True)
+                resp.raise_for_status()
+
+                seg_path = os.path.join(temp_dir, f"seg_{idx:03d}.{job.format}")
+                with open(seg_path, "wb") as f:
+                    f.write(resp.content)
+                audio_paths.append(seg_path)
+
+            # 合并
+            merger = AudioMerger()
+            merged_data = merger.merge(audio_paths, output_format=job.format)
+            return merged_data, total_duration
+
+        finally:
+            shutil.rmtree(temp_dir, ignore_errors=True)
+
+    def _poll_segment_tasks(self, job: TTSJob) -> TTSJob:
+        """轮询所有分段异步任务，全部完成后合并音频。"""
+        segment_task_ids: list[str] = (job.metadata or {}).get("segment_task_ids", [])
+        segment_audio_urls: list[str] = (job.metadata or {}).get("segment_audio_urls", [])
+        segment_count = len(segment_task_ids)
+
+        poll_start = time.monotonic()
+        poll_timeout = 300.0  # 分段任务超时更长
+        poll_interval = 2.0
+
+        while time.monotonic() - poll_start < poll_timeout:
+            all_done = True
+            results: list[dict | None] = [None] * segment_count
+
+            for idx, task_id in enumerate(segment_task_ids):
+                # 已经有音频的分段跳过轮询
+                if idx < len(segment_audio_urls) and segment_audio_urls[idx]:
+                    results[idx] = {
+                        "audio_url": segment_audio_urls[idx],
+                        "duration": 0.0,
+                        "file_size": 0,
+                    }
+                    continue
+
+                try:
+                    result = self.cosyvoice_service.poll_synthesize_task(task_id, timeout=poll_timeout)
+                    results[idx] = result
+                except Exception as e:
+                    logger.error(f"分段任务轮询失败: job_id={job.id}, " f"segment={idx}, error={e}")
+                    self._handle_segment_failure(job, f"分段 {idx + 1} 轮询失败: {e}")
+                    return self.repository.get(job.id)
+
+                if results[idx] is None:
+                    all_done = False
+
+            if all_done and all(r is not None for r in results):
+                # 所有分段完成，下载合并
+                try:
+                    merged_data, total_duration = self._download_and_merge_segments(results, job)
+
+                    # 转存 OSS
+                    permanent_url, storage_key = self._upload_merged_to_oss(
+                        merged_data, job.user_id, job.id, job.format
+                    )
+
+                    job.mark_completed(
+                        output_audio_url=permanent_url,
+                        output_audio_key=storage_key,
+                        duration=total_duration,
+                        file_size=len(merged_data),
+                    )
+                    job = self.repository.update(job)
+                    logger.info(f"分段合成轮询完成: job_id={job.id}, " f"merged_size={len(merged_data)}")
+                    return job
+
+                except Exception as e:
+                    self._handle_segment_failure(job, f"分段合并失败: {e}")
+                    return self.repository.get(job.id)
+
+            # 等待后重试
+            time.sleep(poll_interval)
+
+        # 超时
+        self._handle_segment_failure(job, "分段合成轮询超时（300 秒）")
+        return self.repository.get(job.id)
+
+    def _handle_segment_failure(self, job: TTSJob, error_message: str) -> None:
+        """分段合成失败处理。"""
+        job.mark_failed(error_message)
+        self.repository.update(job)
+        logger.error(f"分段合成失败: job_id={job.id}, error={error_message}")
diff --git a/packages/domain/__init__.py b/packages/domain/__init__.py
index e0ce5822d..7d2992d69 100755
--- a/packages/domain/__init__.py
+++ b/packages/domain/__init__.py
@@ -24,6 +24,7 @@ from .entities import (
 from .generated_video import GeneratedVideo
 from .generation_task import GenerationTask, GenerationTaskStatus
 from .job import Job, JobStatus, JobType
+from .tag import Tag
 from .template_clip_config import ClipType, TemplateClipConfig, TransitionEffect
 from .title_library import TitleLibraryItem
 from .voice_library import VoiceLibraryItem
@@ -56,6 +57,7 @@ __all__ = [
     "JobStatus",
     "JobType",
     "Project",
+    "Tag",
     "TemplateClipConfig",
     "TransitionEffect",
     "User",
diff --git a/packages/domain/entities.py b/packages/domain/entities.py
index ef568686c..b31897d15 100644
--- a/packages/domain/entities.py
+++ b/packages/domain/entities.py
@@ -162,7 +162,7 @@ class Asset:
     quality_score: float | None = None
     uploaded_by_user_id: str = ""
     metadata: dict[str, Any] = field(default_factory=dict)
-    tags: list[str] = field(default_factory=list)
+    tag_ids: list[str] = field(default_factory=list)
     created_at: datetime = field(default_factory=lambda: datetime.now(timezone.utc))
     updated_at: datetime = field(default_factory=lambda: datetime.now(timezone.utc))
 
@@ -214,23 +214,23 @@ class Asset:
             quality_score=quality_score,
             uploaded_by_user_id=uploaded_by_user_id.strip(),
             metadata=metadata or {},
-            tags=[],
+            tag_ids=[],
         )
 
-    def add_tag(self, tag: str) -> None:
-        """添加标签。空标签会被忽略，自动去重。"""
-        clean_tag = tag.strip()
-        if not clean_tag:
-            raise ValueError("标签不能为空")
-        if clean_tag not in self.tags:
-            self.tags.append(clean_tag)
+    def add_tag(self, tag_id: str) -> None:
+        """添加标签 ID。空 ID 会被忽略，自动去重。"""
+        clean_id = tag_id.strip()
+        if not clean_id:
+            raise ValueError("标签 ID 不能为空")
+        if clean_id not in self.tag_ids:
+            self.tag_ids.append(clean_id)
             self.updated_at = datetime.now(timezone.utc)
 
-    def remove_tag(self, tag: str) -> None:
-        """删除标签。如果标签不存在，不报错（幂等性）。"""
-        clean_tag = tag.strip()
-        if clean_tag in self.tags:
-            self.tags.remove(clean_tag)
+    def remove_tag(self, tag_id: str) -> None:
+        """删除标签 ID。如果标签不存在，不报错（幂等性）。"""
+        clean_id = tag_id.strip()
+        if clean_id in self.tag_ids:
+            self.tag_ids.remove(clean_id)
             self.updated_at = datetime.now(timezone.utc)
 
 
diff --git a/packages/domain/tag.py b/packages/domain/tag.py
new file mode 100644
index 000000000..b83b1e8b6
--- /dev/null
+++ b/packages/domain/tag.py
@@ -0,0 +1,24 @@
+"""标签领域实体。"""
+
+from dataclasses import dataclass, field
+from datetime import datetime, timezone
+from uuid import uuid4
+
+
+@dataclass(slots=True)
+class Tag:
+    id: str
+    user_id: str
+    name: str
+    created_at: datetime = field(default_factory=lambda: datetime.now(timezone.utc))
+
+    @classmethod
+    def create(cls, user_id: str, name: str) -> "Tag":
+        clean_name = name.strip()
+        if not clean_name:
+            raise ValueError("标签名称不能为空")
+        return cls(
+            id=uuid4().hex,
+            user_id=user_id,
+            name=clean_name,
+        )
diff --git a/packages/ports/__init__.py b/packages/ports/__init__.py
index 5d915d17b..a4461e159 100644
--- a/packages/ports/__init__.py
+++ b/packages/ports/__init__.py
@@ -4,6 +4,7 @@ from .asset_library_repository import AssetLibraryRepository
 from .asset_repository import AssetRepository
 from .ingest_job_repository import IngestJobRepository
 from .project_repository import ProjectRepository
+from .tag_repository import TagRepository
 from .title_library_repository import TitleLibraryRepository
 from .voice_library_repository import VoiceLibraryRepository
 
@@ -12,6 +13,7 @@ __all__ = [
     "AssetRepository",
     "IngestJobRepository",
     "ProjectRepository",
+    "TagRepository",
     "TitleLibraryRepository",
     "VoiceLibraryRepository",
 ]
diff --git a/packages/ports/asset_repository.py b/packages/ports/asset_repository.py
index 3bee42356..9f96d1f1c 100644
--- a/packages/ports/asset_repository.py
+++ b/packages/ports/asset_repository.py
@@ -50,6 +50,11 @@ class AssetRepository(ABC):
     def delete(self, asset_id: str) -> bool:
         pass
 
+    @abstractmethod
+    def batch_delete(self, asset_ids: list[str]) -> int:
+        """批量删除素材，返回实际删除数量。"""
+        pass
+
     @abstractmethod
     def count_by_project(self, project_id: str) -> int:
         pass
@@ -78,3 +83,13 @@ class AssetRepository(ABC):
     ) -> list[Asset]:
         """按筛选条件搜索候选素材，按质量分降序排列。"""
         pass
+
+    @abstractmethod
+    def find_by_tag_ids(
+        self,
+        tag_ids: list[str],
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Asset]:
+        """查找包含所有指定标签的素材。"""
+        pass
diff --git a/packages/ports/tag_repository.py b/packages/ports/tag_repository.py
new file mode 100644
index 000000000..901d2b12d
--- /dev/null
+++ b/packages/ports/tag_repository.py
@@ -0,0 +1,36 @@
+"""标签仓储接口定义。"""
+
+from abc import ABC, abstractmethod
+
+from packages.domain import Tag
+
+
+class TagRepository(ABC):
+    @abstractmethod
+    def create(self, tag: Tag) -> Tag:
+        pass
+
+    @abstractmethod
+    def get(self, tag_id: str) -> Tag | None:
+        pass
+
+    @abstractmethod
+    def find_by_name(self, user_id: str, name: str) -> Tag | None:
+        pass
+
+    @abstractmethod
+    def list_by_user(
+        self,
+        user_id: str,
+        skip: int = 0,
+        limit: int = 100,
+    ) -> list[Tag]:
+        pass
+
+    @abstractmethod
+    def count_by_user(self, user_id: str) -> int:
+        pass
+
+    @abstractmethod
+    def delete(self, tag_id: str) -> bool:
+        pass
diff --git a/tests/integration/test_asset_tags.py b/tests/integration/test_asset_tags.py
index 212af3c8a..bc3165203 100644
--- a/tests/integration/test_asset_tags.py
+++ b/tests/integration/test_asset_tags.py
@@ -4,7 +4,7 @@ from packages.domain import Asset
 
 
 def test_add_tag_to_asset():
-    """测试添加标签到 Asset。"""
+    """测试添加标签 ID 到 Asset。"""
     asset = Asset.create(
         project_id="proj-1",
         library_id="lib-1",
@@ -13,16 +13,16 @@ def test_add_tag_to_asset():
         mime_type="video/mp4",
     )
 
-    asset.add_tag("风景")
-    asset.add_tag("自然")
+    asset.add_tag("tag-1")
+    asset.add_tag("tag-2")
 
-    assert len(asset.tags) == 2
-    assert "风景" in asset.tags
-    assert "自然" in asset.tags
+    assert len(asset.tag_ids) == 2
+    assert "tag-1" in asset.tag_ids
+    assert "tag-2" in asset.tag_ids
 
 
 def test_add_duplicate_tag_should_ignore():
-    """测试添加重复标签应自动去重。"""
+    """测试添加重复标签 ID 应自动去重。"""
     asset = Asset.create(
         project_id="proj-1",
         library_id="lib-1",
@@ -31,15 +31,15 @@ def test_add_duplicate_tag_should_ignore():
         mime_type="video/mp4",
     )
 
-    asset.add_tag("风景")
-    asset.add_tag("风景")  # 重复
+    asset.add_tag("tag-1")
+    asset.add_tag("tag-1")  # 重复
 
-    assert len(asset.tags) == 1
-    assert asset.tags.count("风景") == 1
+    assert len(asset.tag_ids) == 1
+    assert asset.tag_ids.count("tag-1") == 1
 
 
 def test_add_empty_tag_should_fail():
-    """测试添加空标签应失败。"""
+    """测试添加空标签 ID 应失败。"""
     asset = Asset.create(
         project_id="proj-1",
         library_id="lib-1",
@@ -48,15 +48,15 @@ def test_add_empty_tag_should_fail():
         mime_type="video/mp4",
     )
 
-    with pytest.raises(ValueError, match="标签不能为空"):
+    with pytest.raises(ValueError, match="标签 ID 不能为空"):
         asset.add_tag("")
 
-    with pytest.raises(ValueError, match="标签不能为空"):
+    with pytest.raises(ValueError, match="标签 ID 不能为空"):
         asset.add_tag("   ")  # 仅空格
 
 
 def test_remove_tag_from_asset():
-    """测试从 Asset 删除标签。"""
+    """测试从 Asset 删除标签 ID。"""
     asset = Asset.create(
         project_id="proj-1",
         library_id="lib-1",
@@ -65,18 +65,18 @@ def test_remove_tag_from_asset():
         mime_type="video/mp4",
     )
 
-    asset.add_tag("风景")
-    asset.add_tag("自然")
+    asset.add_tag("tag-1")
+    asset.add_tag("tag-2")
 
-    asset.remove_tag("风景")
+    asset.remove_tag("tag-1")
 
-    assert len(asset.tags) == 1
-    assert "风景" not in asset.tags
-    assert "自然" in asset.tags
+    assert len(asset.tag_ids) == 1
+    assert "tag-1" not in asset.tag_ids
+    assert "tag-2" in asset.tag_ids
 
 
 def test_remove_nonexistent_tag_should_be_idempotent():
-    """测试删除不存在的标签应幂等（不报错）。"""
+    """测试删除不存在的标签 ID 应幂等（不报错）。"""
     asset = Asset.create(
         project_id="proj-1",
         library_id="lib-1",
@@ -85,10 +85,10 @@ def test_remove_nonexistent_tag_should_be_idempotent():
         mime_type="video/mp4",
     )
 
-    asset.add_tag("风景")
+    asset.add_tag("tag-1")
 
-    # 删除不存在的标签，不应报错
-    asset.remove_tag("不存在的标签")
+    # 删除不存在的标签 ID，不应报错
+    asset.remove_tag("nonexistent-tag")
 
-    assert len(asset.tags) == 1
-    assert "风景" in asset.tags
+    assert len(asset.tag_ids) == 1
+    assert "tag-1" in asset.tag_ids
diff --git a/tests/unit/test_asset_batch_delete.py b/tests/unit/test_asset_batch_delete.py
new file mode 100644
index 000000000..ef9a2f468
--- /dev/null
+++ b/tests/unit/test_asset_batch_delete.py
@@ -0,0 +1,77 @@
+"""批量删除素材 + 分页优化 单元测试。"""
+
+import sys
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).resolve().parents[2] / "apps" / "api"))
+
+from packages.adapters.in_memory.asset_repository import InMemoryAssetRepository
+from packages.domain import Asset, AssetStatus
+
+
+class TestBatchDelete:
+    """batch_delete 仓储方法测试。"""
+
+    def _make_repo_with_assets(self):
+        repo = InMemoryAssetRepository()
+        for i in range(5):
+            asset = Asset.create(
+                project_id="proj-1",
+                library_id="lib-1",
+                name=f"voice_{i}.mp3",
+                storage_key=f"uploads/voice_{i}.mp3",
+                mime_type="audio/mpeg",
+                status=AssetStatus.READY,
+            )
+            repo.create(asset)
+        return repo
+
+    def test_batch_delete_removes_multiple(self):
+        repo = InMemoryAssetRepository()
+        assets = []
+        for i in range(5):
+            asset = Asset.create(
+                project_id="proj-1",
+                library_id="lib-1",
+                name=f"voice_{i}.mp3",
+                storage_key=f"uploads/voice_{i}.mp3",
+                mime_type="audio/mpeg",
+            )
+            repo.create(asset)
+            assets.append(asset)
+
+        ids_to_delete = [assets[0].id, assets[2].id, assets[4].id]
+        deleted_count = repo.batch_delete(ids_to_delete)
+
+        assert deleted_count == 3
+        # 验证确实被删了
+        assert repo.get(assets[0].id) is None
+        assert repo.get(assets[2].id) is None
+        assert repo.get(assets[4].id) is None
+        # 验证其他还在
+        assert repo.get(assets[1].id) is not None
+        assert repo.get(assets[3].id) is not None
+
+    def test_batch_delete_empty_list(self):
+        repo = self._make_repo_with_assets()
+        assert repo.batch_delete([]) == 0
+
+    def test_batch_delete_nonexistent_ids(self):
+        repo = self._make_repo_with_assets()
+        deleted = repo.batch_delete(["nonexistent-1", "nonexistent-2"])
+        assert deleted == 0
+
+    def test_batch_delete_mixed_existing_and_nonexistent(self):
+        repo = InMemoryAssetRepository()
+        asset = Asset.create(
+            project_id="proj-1",
+            library_id="lib-1",
+            name="voice.mp3",
+            storage_key="uploads/voice.mp3",
+            mime_type="audio/mpeg",
+        )
+        repo.create(asset)
+
+        deleted = repo.batch_delete([asset.id, "nonexistent"])
+        assert deleted == 1
+        assert repo.get(asset.id) is None
diff --git a/tests/unit/test_asset_tagging.py b/tests/unit/test_asset_tagging.py
new file mode 100644
index 000000000..2b36b40a4
--- /dev/null
+++ b/tests/unit/test_asset_tagging.py
@@ -0,0 +1,130 @@
+"""素材打标/取消标签单元测试。"""
+
+import pytest
+
+from packages.adapters.in_memory.asset_repository import InMemoryAssetRepository
+from packages.adapters.in_memory.tag_repository import InMemoryTagRepository
+from packages.domain import Asset, Tag
+
+
+@pytest.fixture
+def asset_repo():
+    return InMemoryAssetRepository()
+
+
+@pytest.fixture
+def tag_repo():
+    return InMemoryTagRepository()
+
+
+def _create_asset(asset_repo, **kwargs):
+    defaults = dict(
+        project_id="proj-1",
+        library_id="lib-1",
+        name="video.mp4",
+        storage_key="uploads/abc/video.mp4",
+        mime_type="video/mp4",
+    )
+    defaults.update(kwargs)
+    asset = Asset.create(**defaults)
+    return asset_repo.create(asset)
+
+
+def test_tag_asset(asset_repo, tag_repo):
+    """测试给素材打标签。"""
+    asset = _create_asset(asset_repo)
+    tag = tag_repo.create(Tag.create(user_id="user-1", name="风景"))
+
+    asset.add_tag(tag.id)
+    asset_repo.update(asset)
+
+    loaded = asset_repo.get(asset.id)
+    assert tag.id in loaded.tag_ids
+
+
+def test_untag_asset(asset_repo, tag_repo):
+    """测试取消素材标签。"""
+    asset = _create_asset(asset_repo)
+    tag = tag_repo.create(Tag.create(user_id="user-1", name="风景"))
+
+    asset.add_tag(tag.id)
+    asset_repo.update(asset)
+
+    asset.remove_tag(tag.id)
+    asset_repo.update(asset)
+
+    loaded = asset_repo.get(asset.id)
+    assert tag.id not in loaded.tag_ids
+
+
+def test_tag_multiple_assets(asset_repo, tag_repo):
+    """测试同一标签打给多个素材。"""
+    a1 = _create_asset(asset_repo, name="a.mp4")
+    a2 = _create_asset(asset_repo, name="b.mp4")
+    tag = tag_repo.create(Tag.create(user_id="user-1", name="风景"))
+
+    a1.add_tag(tag.id)
+    a2.add_tag(tag.id)
+    asset_repo.update(a1)
+    asset_repo.update(a2)
+
+    assert tag.id in asset_repo.get(a1.id).tag_ids
+    assert tag.id in asset_repo.get(a2.id).tag_ids
+
+
+def test_find_by_tag_ids(asset_repo, tag_repo):
+    """测试按标签 ID 筛选素材。"""
+    tag1 = tag_repo.create(Tag.create(user_id="user-1", name="风景"))
+    tag2 = tag_repo.create(Tag.create(user_id="user-1", name="自然"))
+
+    a1 = _create_asset(asset_repo, name="a.mp4")
+    a1.add_tag(tag1.id)
+    a1.add_tag(tag2.id)
+    asset_repo.update(a1)
+
+    a2 = _create_asset(asset_repo, name="b.mp4")
+    a2.add_tag(tag1.id)
+    asset_repo.update(a2)
+
+    a3 = _create_asset(asset_repo, name="c.mp4")
+    # 无标签
+
+    # 按 tag1 筛选 → a1, a2
+    result = asset_repo.find_by_tag_ids([tag1.id])
+    ids = {a.id for a in result}
+    assert ids == {a1.id, a2.id}
+
+    # 按 tag1 + tag2 筛选（交集）→ a1
+    result = asset_repo.find_by_tag_ids([tag1.id, tag2.id])
+    ids = {a.id for a in result}
+    assert ids == {a1.id}
+
+    # 空 tag_ids → 空结果
+    assert asset_repo.find_by_tag_ids([]) == []
+
+
+def test_delete_tag_cleans_associations(asset_repo, tag_repo):
+    """测试删除标签后素材的 tag_ids 不受影响（关联表由仓储层清理）。"""
+    asset = _create_asset(asset_repo)
+    tag = tag_repo.create(Tag.create(user_id="user-1", name="风景"))
+
+    asset.add_tag(tag.id)
+    asset_repo.update(asset)
+
+    # 删除标签
+    tag_repo.delete(tag.id)
+    assert tag_repo.get(tag.id) is None
+
+    # 素材的 tag_ids 在内存中仍有，但重新加载后 InMemory 不感知关联表
+    # 实际 SQLAlchemy 实现中 _sync_asset_tags 会在 update 时清理
+
+
+def test_duplicate_tag_id_ignored(asset_repo, tag_repo):
+    """测试重复打同一标签自动去重。"""
+    asset = _create_asset(asset_repo)
+    tag = tag_repo.create(Tag.create(user_id="user-1", name="风景"))
+
+    asset.add_tag(tag.id)
+    asset.add_tag(tag.id)  # 重复
+
+    assert asset.tag_ids.count(tag.id) == 1
diff --git a/tests/unit/test_tag_crud.py b/tests/unit/test_tag_crud.py
new file mode 100644
index 000000000..894bd944a
--- /dev/null
+++ b/tests/unit/test_tag_crud.py
@@ -0,0 +1,102 @@
+"""标签 CRUD 单元测试（使用 InMemoryTagRepository）。"""
+
+import pytest
+
+from packages.adapters.in_memory.tag_repository import InMemoryTagRepository
+from packages.domain import Tag
+
+
+@pytest.fixture
+def tag_repo():
+    return InMemoryTagRepository()
+
+
+def test_create_tag(tag_repo):
+    """测试创建标签。"""
+    tag = Tag.create(user_id="user-1", name="风景")
+    created = tag_repo.create(tag)
+
+    assert created.id == tag.id
+    assert created.user_id == "user-1"
+    assert created.name == "风景"
+
+
+def test_create_tag_strips_whitespace(tag_repo):
+    """测试创建标签时自动去除首尾空格。"""
+    tag = Tag.create(user_id="user-1", name="  风景  ")
+    assert tag.name == "风景"
+
+
+def test_create_tag_empty_name_raises():
+    """测试空名称抛出 ValueError。"""
+    with pytest.raises(ValueError, match="标签名称不能为空"):
+        Tag.create(user_id="user-1", name="")
+
+    with pytest.raises(ValueError, match="标签名称不能为空"):
+        Tag.create(user_id="user-1", name="   ")
+
+
+def test_get_tag(tag_repo):
+    """测试按 ID 获取标签。"""
+    tag = Tag.create(user_id="user-1", name="风景")
+    tag_repo.create(tag)
+
+    found = tag_repo.get(tag.id)
+    assert found is not None
+    assert found.name == "风景"
+
+    assert tag_repo.get("nonexistent") is None
+
+
+def test_find_by_name(tag_repo):
+    """测试按用户 ID + 名称查找标签。"""
+    tag = Tag.create(user_id="user-1", name="风景")
+    tag_repo.create(tag)
+
+    found = tag_repo.find_by_name("user-1", "风景")
+    assert found is not None
+    assert found.id == tag.id
+
+    # 不同用户同名标签不冲突
+    assert tag_repo.find_by_name("user-2", "风景") is None
+
+    # 不存在的名称
+    assert tag_repo.find_by_name("user-1", "不存在") is None
+
+
+def test_list_by_user(tag_repo):
+    """测试按用户列出标签（分页）。"""
+    for i in range(5):
+        tag_repo.create(Tag.create(user_id="user-1", name=f"标签{i}"))
+    # 另一个用户的标签
+    tag_repo.create(Tag.create(user_id="user-2", name="其他用户标签"))
+
+    items = tag_repo.list_by_user("user-1")
+    assert len(items) == 5
+
+    # 分页
+    items_page = tag_repo.list_by_user("user-1", skip=2, limit=2)
+    assert len(items_page) == 2
+
+
+def test_count_by_user(tag_repo):
+    """测试按用户统计标签数量。"""
+    for i in range(3):
+        tag_repo.create(Tag.create(user_id="user-1", name=f"标签{i}"))
+    tag_repo.create(Tag.create(user_id="user-2", name="其他"))
+
+    assert tag_repo.count_by_user("user-1") == 3
+    assert tag_repo.count_by_user("user-2") == 1
+    assert tag_repo.count_by_user("user-3") == 0
+
+
+def test_delete_tag(tag_repo):
+    """测试删除标签。"""
+    tag = Tag.create(user_id="user-1", name="风景")
+    tag_repo.create(tag)
+
+    assert tag_repo.delete(tag.id) is True
+    assert tag_repo.get(tag.id) is None
+
+    # 重复删除返回 False
+    assert tag_repo.delete(tag.id) is False
diff --git a/tests/unit/test_tts_oss_transfer.py b/tests/unit/test_tts_oss_transfer.py
new file mode 100644
index 000000000..0fa0ed7ac
--- /dev/null
+++ b/tests/unit/test_tts_oss_transfer.py
@@ -0,0 +1,299 @@
+"""TTS 音频转存 OSS 单元测试。
+
+验证 TTSWorkflowService 在合成完成后将 CosyVoice 临时音频转存到 OSS，
+存储永久 URL 到 TTSJob.output_audio_url，OSS key 到 output_audio_key。
+"""
+
+from __future__ import annotations
+
+from datetime import datetime, timezone
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from packages.application.cosyvoice_service import CosyVoiceService
+from packages.application.tts_job.workflow import TTSWorkflowService
+from packages.domain.tts_job import TTSJob, TTSJobStatus
+
+
+def _make_job(**kwargs) -> TTSJob:
+    defaults = {
+        "id": "test_job_001",
+        "user_id": "user_001",
+        "input_text": "测试文本",
+        "voice_id": "voice_001",
+        "voice_model": "",
+        "project_id": "",
+        "voice_clone_profile_id": "",
+        "status": TTSJobStatus.PENDING,
+        "output_audio_url": "",
+        "output_audio_key": "",
+        "duration": 0.0,
+        "file_size": 0,
+        "sample_rate": 22050,
+        "format": "mp3",
+        "error_message": "",
+        "retry_count": 0,
+        "max_retries": 3,
+        "metadata": {},
+        "started_at": None,
+        "completed_at": None,
+        "created_at": datetime.now(timezone.utc),
+        "updated_at": datetime.now(timezone.utc),
+    }
+    defaults.update(kwargs)
+    return TTSJob(**defaults)
+
+
+def _make_workflow(
+    cosyvoice_service: MagicMock | None = None,
+    repo: MagicMock | None = None,
+    storage: MagicMock | None = None,
+) -> TTSWorkflowService:
+    if cosyvoice_service is None:
+        cosyvoice_service = MagicMock(spec=CosyVoiceService)
+    if repo is None:
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+    if storage is None:
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/tts-outputs/user_001/test_job_001.mp3"
+    return TTSWorkflowService(
+        repository=repo,
+        cosyvoice_service=cosyvoice_service,
+        storage_service=storage,
+    )
+
+
+class TestTransferAudioToOSS:
+    """测试 _transfer_audio_to_oss 方法。"""
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_success_download_and_upload(self, mock_httpx: MagicMock) -> None:
+        """成功下载音频并上传到 OSS，返回永久 URL 和 storage_key。"""
+        mock_resp = MagicMock()
+        mock_resp.content = b"fake audio data"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/tts-outputs/user_001/job_123.mp3"
+
+        workflow = _make_workflow(storage=storage)
+        url, key = workflow._transfer_audio_to_oss(
+            "https://cosyvoice-temp.com/audio.mp3",
+            "user_001",
+            "job_123",
+            "mp3",
+        )
+
+        assert url == "https://oss.example.com/tts-outputs/user_001/job_123.mp3"
+        assert key == "tts-outputs/user_001/job_123.mp3"
+
+        mock_httpx.get.assert_called_once_with(
+            "https://cosyvoice-temp.com/audio.mp3",
+            timeout=60.0,
+            follow_redirects=True,
+        )
+        storage.upload_file.assert_called_once()
+        call_args = storage.upload_file.call_args
+        assert call_args[0][1] == "tts-outputs/user_001/job_123.mp3"
+        assert call_args[1]["content_type"] == "audio/mpeg"
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_download_failure_fallback(self, mock_httpx: MagicMock) -> None:
+        """下载失败时回退到原始临时 URL，storage_key 为空。"""
+        mock_httpx.get.side_effect = Exception("Network error")
+
+        workflow = _make_workflow()
+        url, key = workflow._transfer_audio_to_oss(
+            "https://cosyvoice-temp.com/audio.mp3",
+            "user_001",
+            "job_123",
+        )
+
+        assert url == "https://cosyvoice-temp.com/audio.mp3"
+        assert key == ""
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_upload_failure_fallback(self, mock_httpx: MagicMock) -> None:
+        """上传 OSS 失败时回退到原始临时 URL。"""
+        mock_resp = MagicMock()
+        mock_resp.content = b"fake audio data"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        storage = MagicMock()
+        storage.upload_file.side_effect = Exception("OSS bucket error")
+
+        workflow = _make_workflow(storage=storage)
+        url, key = workflow._transfer_audio_to_oss(
+            "https://cosyvoice-temp.com/audio.mp3",
+            "user_001",
+            "job_123",
+        )
+
+        assert url == "https://cosyvoice-temp.com/audio.mp3"
+        assert key == ""
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_wav_content_type(self, mock_httpx: MagicMock) -> None:
+        """wav 格式使用正确的 content_type。"""
+        mock_resp = MagicMock()
+        mock_resp.content = b"fake wav data"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/audio.wav"
+
+        workflow = _make_workflow(storage=storage)
+        workflow._transfer_audio_to_oss(
+            "https://cosyvoice-temp.com/audio.wav",
+            "user_001",
+            "job_456",
+            "wav",
+        )
+
+        call_args = storage.upload_file.call_args
+        assert call_args[1]["content_type"] == "audio/wav"
+        assert call_args[0][1] == "tts-outputs/user_001/job_456.wav"
+
+
+class TestProcessSynthesisResultWithOSS:
+    """测试 process_synthesis_result 集成 OSS 转存。"""
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_stores_permanent_url_and_key(self, mock_httpx: MagicMock) -> None:
+        """合成结果存 OSS 永久 URL 和 storage_key。"""
+        mock_resp = MagicMock()
+        mock_resp.content = b"audio bytes"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/tts-outputs/user_001/test_job_001.mp3"
+
+        repo = MagicMock()
+        job = _make_job(status=TTSJobStatus.PROCESSING)
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(repo=repo, storage=storage)
+        result = workflow.process_synthesis_result(
+            "test_job_001",
+            audio_url="https://cosyvoice-temp.com/expiring.mp3",
+            duration=5.0,
+            file_size=50000,
+        )
+
+        assert result.status == TTSJobStatus.COMPLETED
+        assert result.output_audio_url == "https://oss.example.com/tts-outputs/user_001/test_job_001.mp3"
+        assert result.output_audio_key == "tts-outputs/user_001/test_job_001.mp3"
+        assert result.duration == 5.0
+        assert result.file_size == 50000
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_fallback_to_temp_url_on_oss_failure(self, mock_httpx: MagicMock) -> None:
+        """OSS 转存失败时，使用 CosyVoice 临时 URL（不阻塞合成流程）。"""
+        mock_httpx.get.side_effect = Exception("Download failed")
+
+        repo = MagicMock()
+        job = _make_job(status=TTSJobStatus.PROCESSING)
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(repo=repo)
+        result = workflow.process_synthesis_result(
+            "test_job_001",
+            audio_url="https://cosyvoice-temp.com/expiring.mp3",
+        )
+
+        assert result.status == TTSJobStatus.COMPLETED
+        assert result.output_audio_url == "https://cosyvoice-temp.com/expiring.mp3"
+        assert result.output_audio_key == ""
+
+
+class TestStartSynthesisSyncWithOSS:
+    """测试 start_synthesis 同步路径的 OSS 转存。"""
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_sync_path_transfers_to_oss(self, mock_httpx: MagicMock) -> None:
+        """CosyVoice 同步返回 audio_url 时，也走 OSS 转存。"""
+        mock_resp = MagicMock()
+        mock_resp.content = b"sync audio bytes"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/tts-outputs/user_001/test_job_001.mp3"
+
+        service = MagicMock(spec=CosyVoiceService)
+        service.submit_synthesize_task.return_value = {
+            "task_id": "",
+            "audio_url": "https://cosyvoice-temp.com/sync.mp3",
+            "duration": 2.0,
+            "file_size": 20000,
+            "request_id": "req_sync",
+        }
+
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo, storage=storage)
+        job = workflow.start_synthesis("test_job_001")
+
+        assert job.status == TTSJobStatus.COMPLETED
+        assert job.output_audio_url == "https://oss.example.com/tts-outputs/user_001/test_job_001.mp3"
+        assert job.output_audio_key == "tts-outputs/user_001/test_job_001.mp3"
+        assert job.duration == 2.0
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_sync_path_oss_failure_stores_temp_url(self, mock_httpx: MagicMock) -> None:
+        """同步路径 OSS 失败时，降级存储临时 URL。"""
+        mock_httpx.get.side_effect = Exception("Network error")
+
+        service = MagicMock(spec=CosyVoiceService)
+        service.submit_synthesize_task.return_value = {
+            "task_id": "",
+            "audio_url": "https://cosyvoice-temp.com/sync.mp3",
+            "duration": 2.0,
+            "file_size": 20000,
+            "request_id": "req_sync",
+        }
+
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo)
+        job = workflow.start_synthesis("test_job_001")
+
+        assert job.status == TTSJobStatus.COMPLETED
+        assert job.output_audio_url == "https://cosyvoice-temp.com/sync.mp3"
+        assert job.output_audio_key == ""
+
+    def test_async_path_no_oss_transfer(self) -> None:
+        """异步路径（返回 task_id，无 audio_url）不触发 OSS 转存。"""
+        service = MagicMock(spec=CosyVoiceService)
+        service.submit_synthesize_task.return_value = {
+            "task_id": "cosy_task_async",
+            "audio_url": "",
+            "duration": 0.0,
+            "file_size": 0,
+            "request_id": "req_async",
+        }
+
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+
+        storage = MagicMock()
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo, storage=storage)
+        job = workflow.start_synthesis("test_job_001")
+
+        assert job.status == TTSJobStatus.PROCESSING
+        # 异步路径不应调用 OSS 上传
+        storage.upload_file.assert_not_called()
diff --git a/tests/unit/test_tts_save_to_library.py b/tests/unit/test_tts_save_to_library.py
new file mode 100644
index 000000000..fc62b0e5c
--- /dev/null
+++ b/tests/unit/test_tts_save_to_library.py
@@ -0,0 +1,344 @@
+"""最后一公里：TTS 合成结果保存到配音库 单元测试。
+
+覆盖：
+- 正常保存已完成 TTS job 到配音库
+- 自动携带元信息（音色名、时长、语速等）
+- 自定义名称
+- TTS job 不存在 → 404
+- TTS job 未完成 → 400
+- 配音库配额已满 → 429
+"""
+
+from __future__ import annotations
+
+from datetime import datetime, timezone
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from packages.domain.tts_job import TTSJob, TTSJobStatus
+from packages.domain.voice_library import VoiceLibraryItem
+
+
+def _make_completed_job(**kwargs) -> TTSJob:
+    """构造一个已完成的 TTSJob。"""
+    defaults = {
+        "id": "tts_job_001",
+        "user_id": "user_001",
+        "input_text": "你好世界",
+        "voice_id": "voice_001",
+        "voice_model": "CosyVoice-v1",
+        "project_id": "proj_001",
+        "voice_clone_profile_id": "",
+        "status": TTSJobStatus.COMPLETED,
+        "output_audio_url": "https://oss.example.com/audio.mp3",
+        "output_audio_key": "tts-outputs/user_001/tts_job_001.mp3",
+        "duration": 5.5,
+        "file_size": 88000,
+        "sample_rate": 22050,
+        "format": "mp3",
+        "error_message": "",
+        "retry_count": 0,
+        "max_retries": 3,
+        "metadata": {"speed": 1.0, "language": "zh-CN"},
+        "started_at": datetime(2026, 7, 7, 10, 0, 0, tzinfo=timezone.utc),
+        "completed_at": datetime(2026, 7, 7, 10, 0, 5, tzinfo=timezone.utc),
+        "created_at": datetime(2026, 7, 7, 10, 0, 0, tzinfo=timezone.utc),
+        "updated_at": datetime(2026, 7, 7, 10, 0, 5, tzinfo=timezone.utc),
+    }
+    defaults.update(kwargs)
+    return TTSJob(**defaults)
+
+
+def _make_voice_library_item(**kwargs) -> VoiceLibraryItem:
+    """构造一个配音库条目。"""
+    defaults = {
+        "id": "voice_lib_001",
+        "user_id": "user_001",
+        "name": "TTS-tts_job_",
+        "text": "你好世界",
+        "voice_provider": "cosyvoice",
+        "voice_id": "voice_001",
+        "voice_name": "CosyVoice-v1",
+        "audio_url": "https://oss.example.com/audio.mp3",
+        "duration": 5.5,
+        "file_size": 88000,
+        "status": "completed",
+        "project_id": "proj_001",
+        "tags": [],
+        "metadata_": {
+            "source": "tts_job",
+            "tts_job_id": "tts_job_001",
+            "format": "mp3",
+            "sample_rate": 22050,
+            "speed": 1.0,
+            "language": "zh-CN",
+        },
+        "created_at": datetime(2026, 7, 7, 10, 1, 0, tzinfo=timezone.utc),
+        "updated_at": datetime(2026, 7, 7, 10, 1, 0, tzinfo=timezone.utc),
+    }
+    defaults.update(kwargs)
+    return VoiceLibraryItem(**defaults)
+
+
+class TestSaveToLibraryMapping:
+    """测试 TTSJob → VoiceLibraryItem 字段映射。"""
+
+    def test_completed_job_maps_correctly(self) -> None:
+        """已完成的 TTS job 字段正确映射到配音库条目。"""
+        job = _make_completed_job()
+
+        # 验证 is_completed 属性
+        assert job.is_completed is True
+
+        # 验证关键字段映射
+        assert job.output_audio_url == "https://oss.example.com/audio.mp3"
+        assert job.duration == 5.5
+        assert job.file_size == 88000
+        assert job.voice_id == "voice_001"
+        assert job.voice_model == "CosyVoice-v1"
+        assert job.input_text == "你好世界"
+        assert job.format == "mp3"
+        assert job.sample_rate == 22050
+
+    def test_metadata_carries_speed_and_format(self) -> None:
+        """元信息携带语速、格式等。"""
+        job = _make_completed_job()
+
+        metadata = {
+            "source": "tts_job",
+            "tts_job_id": job.id,
+            "format": job.format,
+            "sample_rate": job.sample_rate,
+        }
+        if job.metadata:
+            for key in ("speed", "language"):
+                if key in job.metadata:
+                    metadata[key] = job.metadata[key]
+
+        assert metadata["format"] == "mp3"
+        assert metadata["sample_rate"] == 22050
+        assert metadata["speed"] == 1.0
+        assert metadata["language"] == "zh-CN"
+        assert metadata["source"] == "tts_job"
+
+    def test_name_auto_generated_when_empty(self) -> None:
+        """未提供名称时自动生成。"""
+        job = _make_completed_job()
+        name = None  # 模拟未提供名称
+        generated_name = name or f"TTS-{job.id[:8]}"
+        assert generated_name == "TTS-tts_job_"
+
+    def test_name_uses_custom_when_provided(self) -> None:
+        """提供自定义名称时使用自定义名称。"""
+        custom_name = "我的配音"
+        generated_name = custom_name or "TTS-fallback"
+        assert generated_name == "我的配音"
+
+
+class TestSaveToLibraryNotCompleted:
+    """测试未完成 job 不能保存。"""
+
+    def test_pending_job_not_completed(self) -> None:
+        """pending 状态的 job 不能保存。"""
+        job = _make_completed_job(status=TTSJobStatus.PENDING)
+        assert job.is_completed is False
+
+    def test_processing_job_not_completed(self) -> None:
+        """processing 状态的 job 不能保存。"""
+        job = _make_completed_job(status=TTSJobStatus.PROCESSING)
+        assert job.is_completed is False
+
+    def test_failed_job_not_completed(self) -> None:
+        """failed 状态的 job 不能保存。"""
+        job = _make_completed_job(status=TTSJobStatus.FAILED)
+        assert job.is_completed is False
+
+    def test_completed_without_url_not_completed(self) -> None:
+        """status=completed 但没有 audio_url 的 job 不算完成。"""
+        job = _make_completed_job(
+            status=TTSJobStatus.COMPLETED,
+            output_audio_url="",
+        )
+        assert job.is_completed is False
+
+
+class TestSaveToLibraryQuota:
+    """测试配额检查。"""
+
+    def test_quota_exceeded_raises(self) -> None:
+        """配音库配额已满时抛出 QuotaExceededError。"""
+        from packages.application.voice_library.use_cases import QuotaExceededError
+
+        error = QuotaExceededError(dimension="max_voiceovers", limit=10, used=10)
+        assert "10/10" in str(error)
+
+    def test_quota_under_limit_passes(self) -> None:
+        """配额未满时不报错。"""
+        from packages.domain.quota import QuotaDimension, quota_checker
+
+        result = quota_checker.check("free", QuotaDimension.MAX_VOICEOVERS.value, 5)
+        assert result.allowed is True
+
+
+class TestSaveToLibraryCreateCommand:
+    """测试 CreateVoiceLibraryCommand 构建。"""
+
+    def test_command_fields_from_tts_job(self) -> None:
+        """从 TTSJob 构建的 Command 字段正确。"""
+        from packages.application.voice_library.commands import CreateVoiceLibraryCommand
+
+        job = _make_completed_job()
+        metadata_ = {
+            "source": "tts_job",
+            "tts_job_id": job.id,
+            "format": job.format,
+            "sample_rate": job.sample_rate,
+            "speed": 1.0,
+            "language": "zh-CN",
+        }
+
+        command = CreateVoiceLibraryCommand(
+            user_id=job.user_id,
+            name=f"TTS-{job.id[:8]}",
+            text=job.input_text,
+            voice_provider="cosyvoice",
+            voice_id=job.voice_id,
+            voice_name=job.voice_model,
+            audio_url=job.output_audio_url,
+            duration=job.duration,
+            file_size=job.file_size,
+            status="completed",
+            project_id=job.project_id,
+            tags=[],
+            metadata_=metadata_,
+        )
+
+        assert command.user_id == "user_001"
+        assert command.name == "TTS-tts_job_"
+        assert command.text == "你好世界"
+        assert command.voice_provider == "cosyvoice"
+        assert command.voice_id == "voice_001"
+        assert command.voice_name == "CosyVoice-v1"
+        assert command.audio_url == "https://oss.example.com/audio.mp3"
+        assert command.duration == 5.5
+        assert command.file_size == 88000
+        assert command.status == "completed"
+        assert command.project_id == "proj_001"
+        assert command.metadata_["source"] == "tts_job"
+
+    def test_command_with_empty_project_id(self) -> None:
+        """project_id 为空时传空字符串。"""
+        from packages.application.voice_library.commands import CreateVoiceLibraryCommand
+
+        job = _make_completed_job(project_id="")
+
+        command = CreateVoiceLibraryCommand(
+            user_id=job.user_id,
+            name="test",
+            text=job.input_text,
+            voice_provider="cosyvoice",
+            voice_id=job.voice_id,
+            voice_name="",
+            audio_url=job.output_audio_url,
+            duration=job.duration,
+            file_size=job.file_size,
+            status="completed",
+            project_id=job.project_id or "",
+            tags=[],
+            metadata_={},
+        )
+
+        assert command.project_id == ""
+
+    def test_command_voice_name_fallback(self) -> None:
+        """voice_model 为空时 voice_name 回退为空字符串。"""
+        from packages.application.voice_library.commands import CreateVoiceLibraryCommand
+
+        job = _make_completed_job(voice_model="")
+
+        command = CreateVoiceLibraryCommand(
+            user_id=job.user_id,
+            name="test",
+            text=job.input_text,
+            voice_provider="cosyvoice",
+            voice_id=job.voice_id,
+            voice_name=job.voice_model or "",
+            audio_url=job.output_audio_url,
+            duration=job.duration,
+            file_size=job.file_size,
+            status="completed",
+            project_id="",
+            tags=[],
+            metadata_={},
+        )
+
+        assert command.voice_name == ""
+
+
+class TestSaveToLibraryUseCase:
+    """测试 CreateVoiceLibraryUseCase 调用。"""
+
+    def test_use_case_creates_item(self) -> None:
+        """UseCase 正确创建配音库条目。"""
+        from packages.application.voice_library.commands import CreateVoiceLibraryCommand
+        from packages.application.voice_library.use_cases import CreateVoiceLibraryUseCase
+
+        repo = MagicMock()
+        repo.count_by_user.return_value = 0  # 配额未满
+
+        expected_item = _make_voice_library_item()
+        repo.create.side_effect = lambda item: item
+
+        use_case = CreateVoiceLibraryUseCase(repo)
+        command = CreateVoiceLibraryCommand(
+            user_id="user_001",
+            name="test",
+            text="你好",
+            voice_provider="cosyvoice",
+            voice_id="v1",
+            voice_name="Voice1",
+            audio_url="https://example.com/audio.mp3",
+            duration=3.0,
+            file_size=5000,
+            status="completed",
+            project_id="",
+            tags=[],
+            metadata_={},
+        )
+
+        item = use_case.execute(command, plan_name="free")
+
+        repo.create.assert_called_once()
+        assert item is not None
+
+    def test_use_case_quota_exceeded(self) -> None:
+        """UseCase 配额已满时抛出 QuotaExceededError。"""
+        from packages.application.voice_library.commands import CreateVoiceLibraryCommand
+        from packages.application.voice_library.use_cases import (
+            CreateVoiceLibraryUseCase,
+            QuotaExceededError,
+        )
+
+        repo = MagicMock()
+        repo.count_by_user.return_value = 100  # 超过 premium 配额
+
+        use_case = CreateVoiceLibraryUseCase(repo)
+        command = CreateVoiceLibraryCommand(
+            user_id="user_001",
+            name="test",
+            text="你好",
+            voice_provider="cosyvoice",
+            voice_id="v1",
+            voice_name="Voice1",
+            audio_url="https://example.com/audio.mp3",
+            duration=3.0,
+            file_size=5000,
+            status="completed",
+            project_id="",
+            tags=[],
+            metadata_={},
+        )
+
+        with pytest.raises(QuotaExceededError):
+            use_case.execute(command, plan_name="premium")
diff --git a/tests/unit/test_tts_segment_synthesis.py b/tests/unit/test_tts_segment_synthesis.py
new file mode 100644
index 000000000..f903b5d51
--- /dev/null
+++ b/tests/unit/test_tts_segment_synthesis.py
@@ -0,0 +1,485 @@
+"""P1 长文本分段合成单元测试。
+
+覆盖：
+- text_splitter.split_text 分段逻辑
+- audio_merger.AudioMerger 合并逻辑
+- workflow 分段合成路径（同步 / 异步 / 失败）
+"""
+
+from __future__ import annotations
+
+import os
+import tempfile
+from datetime import datetime, timezone
+from unittest.mock import MagicMock, call, patch
+
+import pytest
+
+from packages.application.cosyvoice_service import CosyVoiceError, CosyVoiceService
+from packages.application.tts_job.audio_merger import AudioMergeError, AudioMerger
+from packages.application.tts_job.text_splitter import split_text
+from packages.application.tts_job.workflow import TTSWorkflowService
+from packages.domain.tts_job import TTSJob, TTSJobStatus
+
+# ── text_splitter ─────────────────────────────────────────────
+
+
+class TestSplitText:
+    """测试文本分段工具。"""
+
+    def test_short_text_no_split(self) -> None:
+        """短文本不拆分。"""
+        assert split_text("你好世界", max_chars=500) == ["你好世界"]
+
+    def test_empty_text(self) -> None:
+        """空文本返回空列表。"""
+        assert split_text("") == []
+        assert split_text("   ") == []
+
+    def test_exact_threshold(self) -> None:
+        """恰好等于阈值不拆分。"""
+        text = "a" * 500
+        assert split_text(text, max_chars=500) == [text]
+
+    def test_split_at_sentence_boundary(self) -> None:
+        """在句子边界处分段。"""
+        text = "第一句话。" * 60  # 300 chars
+        text += "第二句话。" * 60  # 300 chars → total 600
+        segments = split_text(text, max_chars=500)
+        assert len(segments) >= 2
+        for seg in segments:
+            assert len(seg) <= 500
+
+    def test_split_at_newline(self) -> None:
+        """在换行符处分段。"""
+        text = "段落一\n" * 100  # 300 chars
+        text += "段落二\n" * 100  # 300 chars
+        segments = split_text(text, max_chars=500)
+        assert len(segments) >= 2
+
+    def test_long_sentence_hard_split(self) -> None:
+        """超长句子硬切。"""
+        text = "a" * 1200
+        segments = split_text(text, max_chars=500)
+        assert len(segments) >= 3
+        for seg in segments:
+            assert len(seg) <= 500
+
+    def test_merge_short_segments(self) -> None:
+        """短段合并减少 API 调用。"""
+        # 多个短句子应该被合并
+        text = "你好。" * 120  # 360 chars, each sentence 3 chars
+        segments = split_text(text, max_chars=500)
+        # 短段应该被合并，段数应该比较少
+        assert len(segments) < 120
+
+    def test_preserves_order(self) -> None:
+        """分段保持原始顺序。"""
+        text = "第一段。第二段。第三段。" + "x" * 490
+        segments = split_text(text, max_chars=500)
+        # 第一个段应该以 "第一段" 开头
+        assert segments[0].startswith("第一段")
+
+
+# ── audio_merger ─────────────────────────────────────────────
+
+
+class TestAudioMerger:
+    """测试 FFmpeg 音频合并器。"""
+
+    def test_empty_list_raises(self) -> None:
+        """空列表抛出 AudioMergeError。"""
+        merger = AudioMerger()
+        with pytest.raises(AudioMergeError, match="没有可合并"):
+            merger.merge([])
+
+    def test_single_file_returns_bytes(self) -> None:
+        """单文件直接返回内容。"""
+        with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as f:
+            f.write(b"fake audio content")
+            f.flush()
+            path = f.name
+
+        try:
+            merger = AudioMerger()
+            data = merger.merge([path])
+            assert data == b"fake audio content"
+        finally:
+            os.unlink(path)
+
+    @patch("packages.application.tts_job.audio_merger.subprocess.run")
+    def test_ffmpeg_called_correctly(self, mock_run: MagicMock) -> None:
+        """多文件调用 FFmpeg concat。"""
+        mock_run.return_value = MagicMock(returncode=0)
+
+        # 创建临时文件
+        paths = []
+        for i in range(3):
+            with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as f:
+                f.write(b"audio")
+                paths.append(f.name)
+
+        try:
+            merger = AudioMerger()
+            # Mock open for reading the merged output
+            with patch("builtins.open", create=True) as mock_open:
+                mock_open.return_value.__enter__ = lambda s: s
+                mock_open.return_value.read = lambda: b"merged audio"
+                try:
+                    merger.merge(paths, output_format="mp3")
+                except (FileNotFoundError, OSError):
+                    pass  # Expected since we're mocking
+
+            # 验证 FFmpeg 被调用
+            mock_run.assert_called_once()
+            cmd = mock_run.call_args[0][0]
+            assert cmd[0] == "ffmpeg"
+            assert "-f" in cmd
+            assert "concat" in cmd
+        finally:
+            for p in paths:
+                os.unlink(p)
+
+    @patch("packages.application.tts_job.audio_merger.subprocess.run")
+    def test_ffmpeg_failure_raises(self, mock_run: MagicMock) -> None:
+        """FFmpeg 失败抛出 AudioMergeError。"""
+        mock_run.return_value = MagicMock(returncode=1, stderr="error details")
+
+        paths = []
+        for i in range(2):
+            with tempfile.NamedTemporaryFile(suffix=".mp3", delete=False) as f:
+                f.write(b"audio")
+                paths.append(f.name)
+
+        try:
+            merger = AudioMerger()
+            with pytest.raises(AudioMergeError, match="FFmpeg 合并失败"):
+                merger.merge(paths)
+        finally:
+            for p in paths:
+                os.unlink(p)
+
+
+# ── workflow segment methods ─────────────────────────────────
+
+
+def _make_job(**kwargs) -> TTSJob:
+    defaults = {
+        "id": "test_job_seg",
+        "user_id": "user_001",
+        "input_text": "x" * 600,  # > 500 threshold
+        "voice_id": "voice_001",
+        "voice_model": "",
+        "project_id": "",
+        "voice_clone_profile_id": "",
+        "status": TTSJobStatus.PENDING,
+        "output_audio_url": "",
+        "output_audio_key": "",
+        "duration": 0.0,
+        "file_size": 0,
+        "sample_rate": 22050,
+        "format": "mp3",
+        "error_message": "",
+        "retry_count": 0,
+        "max_retries": 3,
+        "metadata": {},
+        "started_at": None,
+        "completed_at": None,
+        "created_at": datetime.now(timezone.utc),
+        "updated_at": datetime.now(timezone.utc),
+    }
+    defaults.update(kwargs)
+    return TTSJob(**defaults)
+
+
+def _make_workflow(
+    cosyvoice_service: MagicMock | None = None,
+    repo: MagicMock | None = None,
+    storage: MagicMock | None = None,
+) -> TTSWorkflowService:
+    if cosyvoice_service is None:
+        cosyvoice_service = MagicMock(spec=CosyVoiceService)
+    if repo is None:
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+    if storage is None:
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/merged.mp3"
+    return TTSWorkflowService(
+        repository=repo,
+        cosyvoice_service=cosyvoice_service,
+        storage_service=storage,
+    )
+
+
+class TestStartSegmentSynthesis:
+    """测试 _start_segment_synthesis 分段合成入口。"""
+
+    def test_short_text_no_segment(self) -> None:
+        """短文本不触发分段。"""
+        service = MagicMock(spec=CosyVoiceService)
+        service.submit_synthesize_task.return_value = {
+            "task_id": "task_1",
+            "audio_url": "",
+        }
+        repo = MagicMock()
+        repo.get.return_value = _make_job(input_text="短文本")
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo)
+        job = workflow.start_synthesis("test_job_seg")
+
+        # 短文本走普通路径，不调用分段
+        assert job.status == TTSJobStatus.PROCESSING
+
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_long_text_sync_segments(self, mock_httpx: MagicMock) -> None:
+        """长文本同步分段：所有段立即返回 audio_url，直接合并。"""
+        # Mock 分段音频下载
+        mock_resp = MagicMock()
+        mock_resp.content = b"segment audio"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        service = MagicMock(spec=CosyVoiceService)
+        # 每个分段都同步返回 audio_url
+        service.submit_synthesize_task.side_effect = [
+            {"task_id": "", "audio_url": "https://temp.com/seg1.mp3", "duration": 2.0, "file_size": 1000},
+            {"task_id": "", "audio_url": "https://temp.com/seg2.mp3", "duration": 3.0, "file_size": 1500},
+        ]
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/merged.mp3"
+
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo, storage=storage)
+
+        # Mock AudioMerger 避免真实 FFmpeg 调用
+        with patch("packages.application.tts_job.workflow.AudioMerger") as MockMerger:
+            mock_merger = MagicMock()
+            mock_merger.merge.return_value = b"merged audio data"
+            MockMerger.return_value = mock_merger
+
+            job = workflow.start_synthesis("test_job_seg")
+
+        assert job.status == TTSJobStatus.COMPLETED
+        assert job.output_audio_url == "https://oss.example.com/merged.mp3"
+        assert job.duration == 5.0  # 2.0 + 3.0
+
+    def test_long_text_async_segments(self) -> None:
+        """长文本异步分段：返回 task_id，存入 metadata。"""
+        service = MagicMock(spec=CosyVoiceService)
+        # 每个分段返回 task_id（异步）
+        service.submit_synthesize_task.side_effect = [
+            {"task_id": "seg_task_1", "audio_url": "", "duration": 0.0, "file_size": 0},
+            {"task_id": "seg_task_2", "audio_url": "", "duration": 0.0, "file_size": 0},
+        ]
+
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo)
+        job = workflow.start_synthesis("test_job_seg")
+
+        assert job.status == TTSJobStatus.PROCESSING
+        assert "segment_task_ids" in job.metadata
+        assert job.metadata["segment_task_ids"] == ["seg_task_1", "seg_task_2"]
+
+    def test_segment_submit_failure_marks_failed(self) -> None:
+        """分段提交失败时标记 job 为 failed。"""
+        service = MagicMock(spec=CosyVoiceService)
+        service.submit_synthesize_task.side_effect = CosyVoiceError("API error")
+
+        repo = MagicMock()
+        repo.get.return_value = _make_job()
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo)
+        job = workflow.start_synthesis("test_job_seg")
+
+        assert job.status == TTSJobStatus.FAILED
+
+
+class TestUploadMergedToOSS:
+    """测试 _upload_merged_to_oss 辅助方法。"""
+
+    def test_success(self) -> None:
+        """成功上传返回 URL 和 key。"""
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/merged.mp3"
+
+        workflow = _make_workflow(storage=storage)
+        url, key = workflow._upload_merged_to_oss(b"audio data", "user_001", "job_001", "mp3")
+
+        assert url == "https://oss.example.com/merged.mp3"
+        assert key == "tts-outputs/user_001/job_001.mp3"
+        storage.upload_file.assert_called_once()
+        call_args = storage.upload_file.call_args
+        assert call_args[1]["content_type"] == "audio/mpeg"
+
+    def test_failure_returns_empty(self) -> None:
+        """上传失败返回空字符串。"""
+        storage = MagicMock()
+        storage.upload_file.side_effect = Exception("OSS error")
+
+        workflow = _make_workflow(storage=storage)
+        url, key = workflow._upload_merged_to_oss(b"audio data", "user_001", "job_001", "mp3")
+
+        assert url == ""
+        assert key == ""
+
+
+class TestHandleSegmentFailure:
+    """测试 _handle_segment_failure。"""
+
+    def test_marks_job_failed(self) -> None:
+        """标记 job 为 failed 并更新。"""
+        repo = MagicMock()
+        job = _make_job(status=TTSJobStatus.PROCESSING)
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(repo=repo)
+        workflow._handle_segment_failure(job, "分段 1 合成失败")
+
+        assert job.status == TTSJobStatus.FAILED
+        assert "分段 1 合成失败" in job.error_message
+        repo.update.assert_called_once()
+
+
+class TestPollSegmentTasks:
+    """测试 _poll_segment_tasks 异步轮询。"""
+
+    @patch("packages.application.tts_job.workflow.time")
+    @patch("packages.application.tts_job.workflow.httpx")
+    def test_all_segments_done(self, mock_httpx: MagicMock, mock_time: MagicMock) -> None:
+        """所有分段完成后合并并标记完成。"""
+        # Mock time.monotonic 让循环只执行一次
+        mock_time.monotonic.side_effect = [0.0, 1.0, 2.0]
+        mock_time.sleep = MagicMock()
+
+        # Mock 下载分段音频
+        mock_resp = MagicMock()
+        mock_resp.content = b"seg audio"
+        mock_resp.raise_for_status.return_value = None
+        mock_httpx.get.return_value = mock_resp
+
+        service = MagicMock(spec=CosyVoiceService)
+        service.poll_synthesize_task.side_effect = [
+            {"audio_url": "https://temp.com/seg1.mp3", "duration": 2.0, "file_size": 100},
+            {"audio_url": "https://temp.com/seg2.mp3", "duration": 3.0, "file_size": 200},
+        ]
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/merged.mp3"
+
+        repo = MagicMock()
+        job = _make_job(
+            status=TTSJobStatus.PROCESSING,
+            metadata={
+                "segment_task_ids": ["task_1", "task_2"],
+                "segment_audio_urls": ["", ""],
+                "segment_format": "mp3",
+            },
+        )
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo, storage=storage)
+
+        with patch("packages.application.tts_job.workflow.AudioMerger") as MockMerger:
+            mock_merger = MagicMock()
+            mock_merger.merge.return_value = b"merged data"
+            MockMerger.return_value = mock_merger
+
+            result = workflow._poll_segment_tasks(job)
+
+        assert result.status == TTSJobStatus.COMPLETED
+
+    @patch("packages.application.tts_job.workflow.time")
+    def test_segment_poll_failure(self, mock_time: MagicMock) -> None:
+        """分段轮询失败时标记 job failed。"""
+        mock_time.monotonic.side_effect = [0.0, 1.0]
+        mock_time.sleep = MagicMock()
+
+        service = MagicMock(spec=CosyVoiceService)
+        service.poll_synthesize_task.side_effect = CosyVoiceError("Poll failed")
+
+        repo = MagicMock()
+        job = _make_job(
+            status=TTSJobStatus.PROCESSING,
+            metadata={
+                "segment_task_ids": ["task_1"],
+                "segment_audio_urls": [""],
+            },
+        )
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo)
+        result = workflow._poll_segment_tasks(job)
+
+        assert result.status == TTSJobStatus.FAILED
+
+
+class TestPollAndProcessSynthesisSegmentDetection:
+    """测试 poll_and_process_synthesis 正确识别分段任务。"""
+
+    def test_detects_segment_task(self) -> None:
+        """metadata 中有 segment_task_ids 时走分段轮询路径。"""
+        service = MagicMock(spec=CosyVoiceService)
+
+        repo = MagicMock()
+        job = _make_job(
+            status=TTSJobStatus.PROCESSING,
+            metadata={
+                "segment_task_ids": ["task_1", "task_2"],
+                "segment_audio_urls": ["", ""],
+            },
+        )
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo)
+
+        with patch.object(workflow, "_poll_segment_tasks") as mock_poll:
+            mock_poll.return_value = job
+            workflow.poll_and_process_synthesis("test_job_seg")
+            mock_poll.assert_called_once()
+
+    def test_normal_task_no_segment(self) -> None:
+        """普通任务不走分段路径。"""
+        service = MagicMock(spec=CosyVoiceService)
+        service.poll_synthesize_task.return_value = {
+            "audio_url": "https://temp.com/audio.mp3",
+            "duration": 5.0,
+            "file_size": 5000,
+        }
+
+        repo = MagicMock()
+        job = _make_job(
+            status=TTSJobStatus.PROCESSING,
+            metadata={"cosyvoice_task_id": "task_normal"},
+        )
+        repo.get.return_value = job
+        repo.update.side_effect = lambda j: j
+
+        storage = MagicMock()
+        storage.upload_file.return_value = "https://oss.example.com/audio.mp3"
+
+        workflow = _make_workflow(cosyvoice_service=service, repo=repo, storage=storage)
+
+        with patch("packages.application.tts_job.workflow.httpx") as mock_httpx:
+            mock_resp = MagicMock()
+            mock_resp.content = b"audio"
+            mock_resp.raise_for_status.return_value = None
+            mock_httpx.get.return_value = mock_resp
+
+            result = workflow.poll_and_process_synthesis("test_job_seg")
+
+        assert result.status == TTSJobStatus.COMPLETED
diff --git a/tests/unit/test_tts_streaming.py b/tests/unit/test_tts_streaming.py
new file mode 100644
index 000000000..39a20ff36
--- /dev/null
+++ b/tests/unit/test_tts_streaming.py
@@ -0,0 +1,239 @@
+"""P2 WebSocket 流式合成单元测试。
+
+覆盖：
+- TTSStreamingService 流式合成逻辑
+- 短文本流式合成
+- 长文本分段流式合成
+- 错误处理
+"""
+
+from __future__ import annotations
+
+import asyncio
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+
+from packages.application.cosyvoice_service import CosyVoiceError, CosyVoiceService
+from packages.application.tts_job.streaming_service import (
+    TTSStreamingError,
+    TTSStreamingService,
+)
+
+
+class MockWebSocket:
+    """Mock WebSocket for testing."""
+
+    def __init__(self):
+        self.sent_json = []
+        self.sent_bytes = []
+        self.accepted = False
+
+    async def accept(self):
+        self.accepted = True
+
+    async def send_json(self, data):
+        self.sent_json.append(data)
+
+    async def send_bytes(self, data):
+        self.sent_bytes.append(data)
+
+    async def receive_json(self):
+        return {}
+
+
+class TestTTSStreamingService:
+    """测试 TTS 流式合成服务。"""
+
+    @pytest.mark.asyncio
+    async def test_empty_text_returns_error(self):
+        """空文本返回错误。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        params = {"text": "", "voice_id": "test_voice"}
+
+        await service.synthesize_and_stream(ws, params)
+
+        assert len(ws.sent_json) == 1
+        assert ws.sent_json[0]["type"] == "error"
+        assert "文本不能为空" in ws.sent_json[0]["message"]
+
+    @pytest.mark.asyncio
+    async def test_text_too_long_returns_error(self):
+        """超长文本返回错误。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        params = {"text": "x" * 10001, "voice_id": "test_voice"}
+
+        await service.synthesize_and_stream(ws, params)
+
+        assert len(ws.sent_json) == 1
+        assert ws.sent_json[0]["type"] == "error"
+        assert "文本过长" in ws.sent_json[0]["message"]
+
+    @pytest.mark.asyncio
+    async def test_short_text_stream_success(self):
+        """短文本流式合成成功。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        cosyvoice.submit_synthesize_task.return_value = {
+            "task_id": "task_1",
+            "audio_url": "https://temp.com/audio.mp3",
+            "duration": 5.0,
+            "file_size": 10000,
+        }
+
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        params = {"text": "测试文本", "voice_id": "test_voice", "format": "mp3"}
+
+        with patch.object(service, "_download_audio", return_value=b"fake audio data"):
+            await service.synthesize_and_stream(ws, params)
+
+        # 验证发送了 started 帧
+        assert ws.sent_json[0]["type"] == "started"
+        assert ws.sent_json[0]["segment_count"] == 1
+
+        # 验证发送了二进制音频数据
+        assert len(ws.sent_bytes) > 0
+
+        # 验证发送了 done 帧
+        assert ws.sent_json[-1]["type"] == "done"
+        assert ws.sent_json[-1]["format"] == "mp3"
+
+    @pytest.mark.asyncio
+    async def test_short_text_cosyvoice_error(self):
+        """短文本合成时 CosyVoice 报错。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        cosyvoice.submit_synthesize_task.side_effect = CosyVoiceError("API error")
+
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        params = {"text": "测试文本", "voice_id": "test_voice"}
+
+        await service.synthesize_and_stream(ws, params)
+
+        # 验证发送了 started 帧和 error 帧
+        assert len(ws.sent_json) == 2
+        assert ws.sent_json[0]["type"] == "started"
+        assert ws.sent_json[1]["type"] == "error"
+        assert "API error" in ws.sent_json[1]["message"]
+
+    @pytest.mark.asyncio
+    async def test_short_text_no_audio_url(self):
+        """短文本合成未返回 audio_url。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        cosyvoice.submit_synthesize_task.return_value = {
+            "task_id": "task_1",
+            "audio_url": "",
+        }
+
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        params = {"text": "测试文本", "voice_id": "test_voice"}
+
+        await service.synthesize_and_stream(ws, params)
+
+        # 验证发送了 started 帧和 error 帧
+        assert len(ws.sent_json) == 2
+        assert ws.sent_json[0]["type"] == "started"
+        assert ws.sent_json[1]["type"] == "error"
+        assert "未返回音频 URL" in ws.sent_json[1]["message"]
+
+    @pytest.mark.asyncio
+    async def test_long_text_stream_success(self):
+        """长文本分段流式合成成功。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        # 每个分段都返回 audio_url
+        cosyvoice.submit_synthesize_task.side_effect = [
+            {"task_id": "", "audio_url": "https://temp.com/seg1.mp3", "duration": 2.0},
+            {"task_id": "", "audio_url": "https://temp.com/seg2.mp3", "duration": 3.0},
+        ]
+
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        # 超过 500 字的文本
+        params = {"text": "x" * 600, "voice_id": "test_voice", "format": "mp3"}
+
+        with patch.object(service, "_download_audio", return_value=b"segment audio"):
+            await service.synthesize_and_stream(ws, params)
+
+        # 验证发送了 started 帧（多段）
+        assert ws.sent_json[0]["type"] == "started"
+        assert ws.sent_json[0]["segment_count"] >= 2
+
+        # 验证发送了 segment_done 帧
+        segment_done_count = sum(1 for msg in ws.sent_json if msg["type"] == "segment_done")
+        assert segment_done_count >= 2
+
+        # 验证发送了 done 帧
+        assert ws.sent_json[-1]["type"] == "done"
+
+    @pytest.mark.asyncio
+    async def test_long_text_segment_failure(self):
+        """长文本分段合成失败。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        cosyvoice.submit_synthesize_task.side_effect = CosyVoiceError("Segment error")
+
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        params = {"text": "x" * 600, "voice_id": "test_voice"}
+
+        await service.synthesize_and_stream(ws, params)
+
+        # 验证发送了错误帧
+        error_msgs = [msg for msg in ws.sent_json if msg["type"] == "error"]
+        assert len(error_msgs) > 0
+        assert "合成失败" in error_msgs[0]["message"]
+
+    @pytest.mark.asyncio
+    async def test_stream_audio_chunks(self):
+        """音频分块推送。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        audio_data = b"x" * 10000  # 10KB
+
+        total = await service._stream_audio_chunks(ws, audio_data)
+
+        assert total == 10000
+        # 验证分块发送（4KB per chunk）
+        assert len(ws.sent_bytes) == 3  # 4096 + 4096 + 1808
+
+    @pytest.mark.asyncio
+    async def test_send_json_suppresses_exceptions(self):
+        """_send_json 抑制异常。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        service = TTSStreamingService(cosyvoice)
+
+        ws = MockWebSocket()
+        ws.send_json = AsyncMock(side_effect=Exception("Send failed"))
+
+        # 不应该抛出异常
+        await service._send_json(ws, {"type": "test"})
+
+    @pytest.mark.asyncio
+    async def test_download_audio(self):
+        """下载音频数据。"""
+        cosyvoice = MagicMock(spec=CosyVoiceService)
+        service = TTSStreamingService(cosyvoice)
+
+        with patch("packages.application.tts_job.streaming_service.httpx") as mock_httpx:
+            mock_resp = MagicMock()
+            mock_resp.content = b"audio data"
+            mock_resp.raise_for_status.return_value = None
+            mock_httpx.get.return_value = mock_resp
+
+            result = service._download_audio("https://example.com/audio.mp3")
+
+            assert result == b"audio data"
+            mock_httpx.get.assert_called_once()