feat: 封面模板对接后端 CRUD API，移除 mock 数据

- 新增 api/cover-templates.ts：封装 GET/POST/PUT/DELETE /cover-templates
- useStep6Cover.ts：弹窗打开时 fetch 模板列表，CRUD 操作走 API
- CoverSettingsModal.tsx：增加 loading/error 状态展示 + 删除按钮（非系统模板）
- Step6CoverSettings.tsx：透传 loading/error/delete 属性
- constants.ts：删除 MOCK_COVER_TEMPLATES 硬编码数据

apps/api/app/api/routes/generation_preview.py

@@ -312,7 +312,7 @@ def create_preview_generation_task(
                 project_id="",
                 asset_library_id="",
                 strategy_id=strategy_id,
-                voice_library_id="",
+                voice_library_id=request.voice_library_id,
                 template_id=request.template_id,
                 asset_ids=list(request.asset_ids),
                 title_ids=list(request.title_ids),

apps/api/app/schemas/generation_task.py

@@ -161,6 +161,9 @@ class CreatePreviewGenerationTaskRequest(BaseModel):
     asset_ids: list[str] = Field(default_factory=list)
     title_ids: list[str] = Field(default_factory=list)
     voice_ids: list[str] = Field(default_factory=list)
+    voice_library_id: str = Field(
+        default="", description="配音素材库ID（用户上传的音频或AI配音），对应配音选择页面选择的配音素材"
+    )
     video_title: str = Field(default="", description="生成视频的标题/名称，为空则使用默认命名")
     duration: float = Field(default=0.0, ge=0, description="期望视频时长（秒），0 表示由模板决定")
     video_ratio: str = Field(default="", description="视频比例，如 16:9 / 9:16，为空使用模板默认")

apps/web/src/api/cover-templates.ts

@@ -0,0 +1,53 @@
+/**
+ * 封面模板 CRUD API
+ * 后端路由: /api/v1/cover-templates
+ */
+import apiClient from "./client"
+import type { CoverTemplate } from "@/pages/editing-planner/types"
+
+export interface CoverTemplateListResponse {
+  items: CoverTemplate[]
+  total: number
+}
+
+export interface CoverTemplateCreateRequest {
+  name: string
+  config?: {
+    background_enabled?: boolean
+    background_color?: string
+    portrait_enabled?: boolean
+    title_text?: string
+    subtitle_text?: string
+    mask_enabled?: boolean
+  }
+}
+
+export type CoverTemplateUpdateRequest = Partial<CoverTemplateCreateRequest>
+
+/** 获取封面模板列表 */
+export async function fetchCoverTemplates(): Promise<CoverTemplateListResponse> {
+  const response = await apiClient.get<CoverTemplateListResponse>("/cover-templates")
+  return response.data
+}
+
+/** 创建封面模板 */
+export async function createCoverTemplate(
+  data: CoverTemplateCreateRequest,
+): Promise<CoverTemplate> {
+  const response = await apiClient.post<CoverTemplate>("/cover-templates", data)
+  return response.data
+}
+
+/** 更新封面模板 */
+export async function updateCoverTemplate(
+  id: string,
+  data: CoverTemplateUpdateRequest,
+): Promise<CoverTemplate> {
+  const response = await apiClient.put<CoverTemplate>(`/cover-templates/${id}`, data)
+  return response.data
+}
+
+/** 删除封面模板（系统模板不可删） */
+export async function deleteCoverTemplate(id: string): Promise<void> {
+  await apiClient.delete(`/cover-templates/${id}`)
+}

apps/web/src/api/generation/confirm.ts

@@ -0,0 +1,14 @@
+import apiClient from "../client"
+import type { ConfirmGenerationRequest, ConfirmGenerationResponse } from "./types"
+
+/** 确认生成 — 基于预览任务创建正式生成任务 */
+export const confirmGeneration = async (
+  taskId: string,
+  params: ConfirmGenerationRequest,
+): Promise<ConfirmGenerationResponse> => {
+  const response = await apiClient.post<ConfirmGenerationResponse>(
+    `/tasks/${taskId}/confirm`,
+    params,
+  )
+  return response.data
+}

apps/web/src/api/generation/index.ts

@@ -3,6 +3,10 @@ export type {
   CreatePreviewRequest,
   CreatePreviewResponse,
   PreviewTaskResponse,
+  ConfirmGenerationRequest,
+  ConfirmGenerationResponse,
+  ConfirmGenerationTaskItem,
 } from "./types"
 
 export { createPreview, getPreviewStatus } from "./preview"
+export { confirmGeneration } from "./confirm"

apps/web/src/api/generation/types.ts

@@ -7,6 +7,8 @@ export interface CreatePreviewRequest {
   asset_ids: string[]
   title_ids?: string[]
   voice_ids?: string[]
+  /** 配音素材库ID（用户上传的音频或AI配音），对应配音选择页面选择的配音素材 */
+  voice_library_id?: string
   video_title?: string
   duration?: number
   video_ratio?: string
@@ -45,3 +47,47 @@ export interface PreviewTaskResponse {
   finished_at?: string
   generate_duration?: number
 }
+
+/** 确认生成请求体 — 基于预览任务创建正式生成任务 */
+export interface ConfirmGenerationRequest {
+  /** 输出视频宽度，默认 1080 */
+  output_width?: number
+  /** 输出视频高度，默认 1920 */
+  output_height?: number
+  /** 自定义封面图片 URL */
+  cover_url?: string
+  /** 自定义视频标题 */
+  custom_title?: string
+}
+
+/** 确认生成响应 */
+export interface ConfirmGenerationResponse {
+  items: ConfirmGenerationTaskItem[]
+  total: number
+}
+
+/** 确认生成返回的任务项 */
+export interface ConfirmGenerationTaskItem {
+  id: string
+  project_id: string
+  asset_library_id: string
+  strategy_id: string
+  voice_library_id: string
+  template_id: string
+  asset_ids: string[]
+  title_ids: string[]
+  voice_ids: string[]
+  source_edit_plan_id: string
+  asset_select_mode: string
+  batch_id: string
+  is_preview: boolean
+  source_task_id: string
+  output_width: number
+  output_height: number
+  cover_url: string
+  custom_title: string
+  status: string
+  progress: number
+  result_count: number
+  error_message: string
+}

apps/web/src/pages/editing-planner/types/cover.ts

@@ -30,3 +30,20 @@ export const DEFAULT_COVER_CONFIG: CoverConfig = {
   ai_suggested_time: null,
   thumbnail_url: "",
 }
+
+/** 封面模板 */
+export interface CoverTemplate {
+  id: string
+  name: string
+  thumbnail_url: string
+  is_system: boolean
+  created_at: string
+  config?: {
+    background_enabled?: boolean
+    background_color?: string
+    portrait_enabled?: boolean
+    title_text?: string
+    subtitle_text?: string
+    mask_enabled?: boolean
+  }
+}

apps/web/src/pages/editing-planner/types/index.ts

@@ -72,7 +72,7 @@ export {
 } from "./sticker"
 
 /* 封面 */
-export { type CoverMode, type CoverConfig, DEFAULT_COVER_CONFIG } from "./cover"
+export { type CoverMode, type CoverConfig, DEFAULT_COVER_CONFIG, type CoverTemplate } from "./cover"
 
 /* 片段数据 */
 export { type ClipType, type ClipData } from "./clip"

apps/web/src/pages/generate/GeneratePage.tsx

@@ -106,6 +106,7 @@ const GeneratePage: React.FC = () => {
     duration,
     videoRatio,
     voiceIds: previewVoiceIds,
+    voiceLibraryId: selectedVoice || undefined,
     previewCount,
   })
 

apps/web/src/pages/generate/components/Step6CoverSettings.tsx

@@ -1,9 +1,9 @@
-import React, { useEffect, useRef } from "react"
+import React from "react"
 import type { CoverConfig } from "../../editing-planner/types"
 import { useStep6Cover } from "../hooks/useStep6Cover"
-import { CoverModeSelector } from "./cover-settings/CoverModeSelector"
-import { FrameCoverPicker } from "./cover-settings/FrameCoverPicker"
-import { UploadCoverPicker } from "./cover-settings/UploadCoverPicker"
+import Button from "@/components/ui/Button"
+import CoverSettingsModal from "./cover-settings/CoverSettingsModal"
+import CoverEditorModal from "./cover-settings/CoverEditorModal"
 
 interface Step6CoverSettingsProps {
   coverSettings: CoverConfig
@@ -18,15 +18,21 @@ interface Step6CoverSettingsProps {
 const Step6CoverSettings: React.FC<Step6CoverSettingsProps> = (props) => {
   const {
     coverSettings,
-    formatTime,
-    toggleEnabled,
-    setMode,
-    setFrameTime,
-    handleUpload,
     generateAutoCover,
-    totalDuration,
-    COVER_MODE_LABELS,
-    COVER_MODE_ICONS,
+    showCoverSettings,
+    setShowCoverSettings,
+    showCoverEditor,
+    setShowCoverEditor,
+    selectedTemplateId,
+    selectedTemplateName,
+    editingTemplate,
+    coverTemplates,
+    templatesLoading,
+    templatesError,
+    handleSelectTemplate,
+    handleEditTemplate,
+    handleSaveTemplate,
+    handleDeleteTemplate,
   } = useStep6Cover({
     coverSettings: props.coverSettings,
     onCoverSettingsChange: props.onCoverSettingsChange,
@@ -35,32 +41,9 @@ const Step6CoverSettings: React.FC<Step6CoverSettingsProps> = (props) => {
     selectedTemplate: props.selectedTemplate,
   })
 
-  // 进入 auto 模式时自动触发智能封面生成
-  const autoTriggeredRef = useRef(false)
-  useEffect(() => {
-    // 切换模式、禁用封面或素材变更时重置触发标记
-    if (coverSettings.mode !== "auto" || !coverSettings.enabled) {
-      autoTriggeredRef.current = false
-      return
-    }
-    // 有素材且未生成过封面时自动触发
-    if (
-      coverSettings.mode === "auto" &&
-      !coverSettings.thumbnail_url &&
-      !autoTriggeredRef.current &&
-      props.assetIds &&
-      props.assetIds.length > 0
-    ) {
-      autoTriggeredRef.current = true
-      generateAutoCover()
-    }
-  }, [
-    coverSettings.enabled,
-    coverSettings.mode,
-    coverSettings.thumbnail_url,
-    generateAutoCover,
-    props.assetIds,
-  ])
+  const handleAutoGenerate = () => {
+    generateAutoCover()
+  }
 
   // 预览图：优先 thumbnail_url，其次 upload_url
   const previewUrl = coverSettings.thumbnail_url || coverSettings.upload_url
@@ -69,70 +52,52 @@ const Step6CoverSettings: React.FC<Step6CoverSettingsProps> = (props) => {
     <div className="xx-form-section">
       <h3>🖼️ 选择封面</h3>
 
-      <div className="xx-cover-header">
-        <span className="xx-cover-header-label">启用自定义封面</span>
-        <label className="xx-switch">
-          <input
-            type="checkbox"
-            checked={coverSettings.enabled}
-            onChange={(e) => toggleEnabled(e.target.checked)}
-          />
-          <span className="xx-switch-slider" />
-        </label>
+      <div className="xx-cover-actions">
+        <Button buttonType="primary" onClick={handleAutoGenerate}>
+          ✨ 自动生成封面
+        </Button>
+        <Button buttonType="ghost" onClick={() => setShowCoverSettings(true)}>
+          ⚙️ 封面设置
+        </Button>
       </div>
 
-      {coverSettings.enabled && (
-        <>
-          <div className="xx-section-title">封面来源</div>
-          <CoverModeSelector
-            mode={coverSettings.mode}
-            onModeChange={setMode}
-            modeLabels={COVER_MODE_LABELS}
-            modeIcons={COVER_MODE_ICONS}
-          />
+      <div className="xx-cover-selected-template">已选模板: {selectedTemplateName}</div>
 
-          {coverSettings.mode === "auto" && (
-            <div className="xx-cover-auto">
-              <div className="xx-cover-auto-badge">
-                <span style={{ fontSize: 24 }}>🤖</span>
-                <span>AI 智能选帧</span>
-              </div>
-            </div>
-          )}
-
-          {coverSettings.mode === "frame" && (
-            <FrameCoverPicker
-              frameTime={coverSettings.frame_time}
-              totalDuration={totalDuration}
-              formatTime={formatTime}
-              onFrameTimeChange={setFrameTime}
-            />
-          )}
-
-          {coverSettings.mode === "upload" && (
-            <UploadCoverPicker uploadUrl={coverSettings.upload_url} onUpload={handleUpload} />
-          )}
-
-          <div className="xx-section-title">封面预览</div>
-          <div className="xx-cover-preview-box">
-            {previewUrl ? (
-              <img src={previewUrl} alt="封面预览" className="xx-cover-preview-img" />
-            ) : (
-              <div className="xx-cover-preview-placeholder">
-                <span style={{ fontSize: 28 }}>🖼️</span>
-                <span>
-                  {coverSettings.mode === "auto"
-                    ? "AI 正在选择..."
-                    : coverSettings.mode === "frame"
-                      ? `帧 ${formatTime(coverSettings.frame_time)}`
-                      : "未上传封面"}
-                </span>
-              </div>
-            )}
-            <div className="xx-cover-preview-ratio">9:16</div>
+      <div className="xx-section-title">封面预览</div>
+      <div className="xx-cover-preview-box">
+        {previewUrl ? (
+          <img src={previewUrl} alt="封面预览" className="xx-cover-preview-img" />
+        ) : (
+          <div className="xx-cover-preview-placeholder">
+            <span style={{ fontSize: 28 }}>🖼️</span>
+            <span>点击"自动生成封面"或选择模板</span>
           </div>
-        </>
-      )}
+        )}
+        <div className="xx-cover-preview-ratio">9:16</div>
+      </div>
+
+      <CoverSettingsModal
+        open={showCoverSettings}
+        onClose={() => setShowCoverSettings(false)}
+        templates={coverTemplates}
+        loading={templatesLoading}
+        error={templatesError}
+        selectedTemplateId={selectedTemplateId}
+        onSelectTemplate={handleSelectTemplate}
+        onEditTemplate={handleEditTemplate}
+        onDeleteTemplate={handleDeleteTemplate}
+        onCreateNew={() => {
+          setShowCoverSettings(false)
+          setShowCoverEditor(true)
+        }}
+      />
+
+      <CoverEditorModal
+        open={showCoverEditor}
+        onClose={() => setShowCoverEditor(false)}
+        template={editingTemplate}
+        onSave={handleSaveTemplate}
+      />
     </div>
   )
 }

apps/web/src/pages/generate/components/cover-settings/CoverEditorModal.tsx

@@ -0,0 +1,125 @@
+import React, { useState } from "react"
+import type { CoverTemplate } from "../../../editing-planner/types"
+import Modal from "@/components/ui/Modal"
+import Button from "@/components/ui/Button"
+
+interface CoverEditorModalProps {
+  open: boolean
+  onClose: () => void
+  template: CoverTemplate | null
+  onSave: (template: CoverTemplate) => void
+}
+
+interface SectionState {
+  basic: boolean
+  portrait: boolean
+  background: boolean
+  title: boolean
+  subtitle: boolean
+  mask: boolean
+}
+
+const CoverEditorModal: React.FC<CoverEditorModalProps> = ({ open, onClose, template, onSave }) => {
+  const [name, setName] = useState(template?.name || "")
+  const [sections, setSections] = useState<SectionState>({
+    basic: true,
+    portrait: false,
+    background: false,
+    title: true,
+    subtitle: true,
+    mask: false,
+  })
+
+  const toggleSection = (key: keyof SectionState) => {
+    setSections((prev) => ({ ...prev, [key]: !prev[key] }))
+  }
+
+  const handleSave = () => {
+    if (!template) return
+    onSave({ ...template, name })
+    onClose()
+  }
+
+  return (
+    <Modal
+      open={open}
+      onCancel={onClose}
+      width={1000}
+      title="自定义封面编辑器"
+      centered
+      footer={null}
+    >
+      <div className="xx-cover-editor-header">
+        <input
+          className="xx-cover-editor-name-input"
+          value={name}
+          onChange={(e) => setName(e.target.value)}
+          placeholder="输入模板名称"
+        />
+        <div className="xx-cover-editor-header-actions">
+          <Button buttonType="ghost" onClick={onClose}>
+            取消
+          </Button>
+          <Button buttonType="primary" onClick={handleSave}>
+            保存模板
+          </Button>
+        </div>
+      </div>
+
+      <div className="xx-cover-editor-layout">
+        {/* 左侧折叠面板 */}
+        <div className="xx-cover-editor-left">
+          {[
+            { key: "basic" as const, label: "基础设置" },
+            { key: "portrait" as const, label: "人像设置" },
+            { key: "background" as const, label: "背景设置", toggle: true },
+            { key: "title" as const, label: "主标题" },
+            { key: "subtitle" as const, label: "副标题" },
+            { key: "mask" as const, label: "蒙版", toggle: true },
+          ].map((item) => (
+            <div key={item.key} className="xx-cover-editor-section">
+              <div
+                className="xx-cover-editor-section-header"
+                onClick={() => toggleSection(item.key)}
+              >
+                <span>{item.label}</span>
+                <span>{sections[item.key] ? "▾" : "▸"}</span>
+              </div>
+              {sections[item.key] && (
+                <div className="xx-cover-editor-section-body">
+                  {item.toggle ? (
+                    <label style={{ display: "flex", alignItems: "center", gap: 6 }}>
+                      <input type="checkbox" defaultChecked={false} />
+                      已开启
+                    </label>
+                  ) : (
+                    <span style={{ color: "var(--text-tertiary)" }}>暂无配置项</span>
+                  )}
+                </div>
+              )}
+            </div>
+          ))}
+        </div>
+
+        {/* 右侧画布预览 */}
+        <div className="xx-cover-editor-right">
+          <div className="xx-cover-editor-canvas">
+            {/* 人像占位 */}
+            <div className="xx-cover-editor-portrait">
+              {/* 四角拖拽手柄 */}
+              <span className="xx-cover-editor-handle" style={{ top: -4, left: -4 }} />
+              <span className="xx-cover-editor-handle" style={{ top: -4, right: -4 }} />
+              <span className="xx-cover-editor-handle" style={{ bottom: -4, left: -4 }} />
+              <span className="xx-cover-editor-handle" style={{ bottom: -4, right: -4 }} />
+            </div>
+            {/* 文字占位 */}
+            <div className="xx-cover-editor-title-placeholder">主标题文字</div>
+            <div className="xx-cover-editor-subtitle-placeholder">副标题文字</div>
+          </div>
+        </div>
+      </div>
+    </Modal>
+  )
+}
+
+export default CoverEditorModal

apps/web/src/pages/generate/components/cover-settings/CoverSettingsModal.tsx

@@ -0,0 +1,112 @@
+import React from "react"
+import type { CoverTemplate } from "../../../editing-planner/types"
+import Modal from "@/components/ui/Modal"
+import Button from "@/components/ui/Button"
+
+interface CoverSettingsModalProps {
+  open: boolean
+  onClose: () => void
+  templates: CoverTemplate[]
+  loading?: boolean
+  error?: string | null
+  selectedTemplateId: string
+  onSelectTemplate: (id: string) => void
+  onEditTemplate: (template: CoverTemplate) => void
+  onDeleteTemplate: (id: string) => void
+  onCreateNew: () => void
+}
+
+const GRADIENT_MAP: Record<string, string> = {
+  default: "linear-gradient(135deg, #e0e0e0, #c0c0c0)",
+  "bold-red": "linear-gradient(135deg, #ef4444, #b91c1c)",
+  "elegant-black": "linear-gradient(135deg, #374151, #111827)",
+  "gradient-blue": "linear-gradient(135deg, #3b82f6, #1d4ed8)",
+  "gradient-purple": "linear-gradient(135deg, #8b5cf6, #6d28d9)",
+  "warm-orange": "linear-gradient(135deg, #f97316, #ea580c)",
+  "fresh-green": "linear-gradient(135deg, #22c55e, #15803d)",
+  "tech-blue": "linear-gradient(135deg, #06b6d4, #0e7490)",
+}
+
+const CoverSettingsModal: React.FC<CoverSettingsModalProps> = ({
+  open,
+  onClose,
+  templates,
+  loading = false,
+  error = null,
+  selectedTemplateId,
+  onSelectTemplate,
+  onEditTemplate,
+  onDeleteTemplate,
+  onCreateNew,
+}) => {
+  return (
+    <Modal open={open} onCancel={onClose} width={800} title="封面设置" centered footer={null}>
+      <div className="xx-cover-modal-toolbar">
+        <Button buttonType="primary">选择素材文件</Button>
+        <Button buttonType="ghost">导出全部</Button>
+        <Button buttonType="primary" onClick={onCreateNew}>
+          创建新模板
+        </Button>
+      </div>
+
+      {loading && (
+        <div style={{ textAlign: "center", padding: "40px 0", color: "var(--text-secondary)" }}>
+          加载中...
+        </div>
+      )}
+
+      {error && !loading && (
+        <div style={{ textAlign: "center", padding: "40px 0", color: "#ef4444" }}>{error}</div>
+      )}
+
+      {!loading && !error && (
+        <div className="xx-cover-template-grid">
+          {templates.map((tpl) => (
+            <div
+              key={tpl.id}
+              className={`xx-cover-template-card${selectedTemplateId === tpl.id ? " selected" : ""}`}
+              onClick={() => onSelectTemplate(tpl.id)}
+            >
+              <div
+                className="xx-cover-template-thumb"
+                style={{ background: GRADIENT_MAP[tpl.id] || GRADIENT_MAP.default }}
+              >
+                🖼️
+              </div>
+              <div className="xx-cover-template-info">
+                <div className="xx-cover-template-name">
+                  {tpl.name}
+                  {tpl.is_system && <span className="xx-cover-template-badge">✨ 系统模板</span>}
+                </div>
+                <div className="xx-cover-template-date">{tpl.created_at}</div>
+                <div className="xx-cover-template-actions" onClick={(e) => e.stopPropagation()}>
+                  <Button buttonType="ghost" buttonSize="sm" onClick={() => onEditTemplate(tpl)}>
+                    编辑
+                  </Button>
+                  {!tpl.is_system && (
+                    <Button
+                      buttonType="ghost"
+                      buttonSize="sm"
+                      onClick={() => {
+                        if (confirm("确定删除此模板？")) {
+                          onDeleteTemplate(tpl.id)
+                        }
+                      }}
+                    >
+                      删除
+                    </Button>
+                  )}
+                  <Button buttonType="ghost" buttonSize="sm">
+                    导出
+                  </Button>
+                </div>
+              </div>
+            </div>
+          ))}
+        </div>
+      )}
+    </Modal>
+  )
+}
+
+export default CoverSettingsModal

apps/web/src/pages/generate/generate.css

@@ -2383,7 +2383,6 @@
   font-size: 14px;
 }
 
-
 /* ── 加载中状态 ── */
 .xx-preview-loading {
   text-align: center;
@@ -2889,3 +2888,196 @@
 .ant-modal-close {
   color: #fff !important;
 }
+
+/* ── 封面设置区域改造样式 ── */
+
+/* 封面操作按钮区 */
+.xx-cover-actions {
+  display: flex;
+  gap: 12px;
+  margin-bottom: 12px;
+}
+
+/* 已选模板文字 */
+.xx-cover-selected-template {
+  font-size: 13px;
+  color: var(--text-secondary);
+  margin-bottom: 16px;
+}
+
+/* 封面设置弹窗 - 工具栏 */
+.xx-cover-modal-toolbar {
+  display: flex;
+  gap: 8px;
+  margin-bottom: 20px;
+  flex-wrap: wrap;
+}
+
+/* 封面模板网格 */
+.xx-cover-template-grid {
+  display: grid;
+  grid-template-columns: repeat(4, 1fr);
+  gap: 16px;
+}
+
+/* 封面模板卡片 */
+.xx-cover-template-card {
+  border: 2px solid var(--border-color);
+  border-radius: var(--radius-md);
+  overflow: hidden;
+  cursor: pointer;
+  transition: border-color 0.2s;
+}
+.xx-cover-template-card:hover {
+  border-color: var(--primary-color);
+}
+.xx-cover-template-card.selected {
+  border-color: var(--primary-color);
+  box-shadow: 0 0 0 2px rgba(102, 126, 234, 0.2);
+}
+
+/* 卡片缩略图 */
+.xx-cover-template-thumb {
+  aspect-ratio: 9/16;
+  background: linear-gradient(135deg, #f0f0f0, #e0e0e0);
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  font-size: 32px;
+  color: #ccc;
+}
+
+/* 卡片信息区 */
+.xx-cover-template-info {
+  padding: 8px;
+}
+.xx-cover-template-name {
+  display: flex;
+  align-items: center;
+  gap: 4px;
+  font-size: 13px;
+  font-weight: 500;
+  margin-bottom: 2px;
+}
+.xx-cover-template-badge {
+  font-size: 11px;
+  color: #7c3aed;
+  background: rgba(124, 58, 237, 0.1);
+  padding: 1px 6px;
+  border-radius: 4px;
+  white-space: nowrap;
+}
+.xx-cover-template-date {
+  font-size: 11px;
+  color: var(--text-tertiary);
+  margin-bottom: 6px;
+}
+.xx-cover-template-actions {
+  display: flex;
+  gap: 4px;
+}
+
+/* 封面编辑器 */
+.xx-cover-editor-layout {
+  display: flex;
+  gap: 20px;
+  min-height: 500px;
+}
+.xx-cover-editor-left {
+  width: 280px;
+  flex-shrink: 0;
+}
+.xx-cover-editor-right {
+  flex: 1;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  background: #f5f5f5;
+  border-radius: var(--radius-md);
+}
+.xx-cover-editor-canvas {
+  width: 225px;
+  height: 400px;
+  background: #ddd;
+  position: relative;
+  border-radius: 8px;
+  overflow: hidden;
+}
+.xx-cover-editor-portrait {
+  position: absolute;
+  top: 20%;
+  left: 15%;
+  width: 70%;
+  height: 45%;
+  background: #a8d4f0;
+  border: 2px solid #333;
+}
+.xx-cover-editor-handle {
+  position: absolute;
+  width: 8px;
+  height: 8px;
+  background: #333;
+  border: 1px solid white;
+}
+
+/* 编辑器折叠面板 */
+.xx-cover-editor-section {
+  border: 1px solid var(--border-color);
+  border-radius: var(--radius-sm);
+  margin-bottom: 8px;
+}
+.xx-cover-editor-section-header {
+  padding: 10px 12px;
+  font-size: 13px;
+  font-weight: 500;
+  cursor: pointer;
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  background: var(--bg-tertiary);
+}
+.xx-cover-editor-section-body {
+  padding: 12px;
+  font-size: 12px;
+  color: var(--text-secondary);
+}
+
+/* 编辑器顶部 */
+.xx-cover-editor-header {
+  margin-bottom: 16px;
+}
+.xx-cover-editor-name-input {
+  width: 100%;
+  padding: 8px 12px;
+  border: 1px solid var(--border-color);
+  border-radius: var(--radius-sm);
+  font-size: 14px;
+  margin-bottom: 12px;
+}
+.xx-cover-editor-header-actions {
+  display: flex;
+  justify-content: flex-end;
+  gap: 8px;
+}
+
+/* 编辑器画布内文字占位 */
+.xx-cover-editor-title-placeholder {
+  position: absolute;
+  top: 72%;
+  left: 50%;
+  transform: translateX(-50%);
+  font-size: 16px;
+  font-weight: 700;
+  color: #fff;
+  text-shadow: 1px 1px 3px rgba(0, 0, 0, 0.6);
+  white-space: nowrap;
+}
+.xx-cover-editor-subtitle-placeholder {
+  position: absolute;
+  top: 82%;
+  left: 50%;
+  transform: translateX(-50%);
+  font-size: 11px;
+  color: rgba(255, 255, 255, 0.85);
+  white-space: nowrap;
+}

apps/web/src/pages/generate/hooks/generate-video/types.ts

@@ -19,6 +19,8 @@ export interface UseGenerateVideoProps {
   autoSubtitles: boolean
   bgm: boolean
   generateCount: number
+  /** 预览任务的 task_id（用于新确认生成 API） */
+  previewTaskId?: string
 }
 
 /** 生成阶段 */

apps/web/src/pages/generate/hooks/useGenerateVideo.ts

@@ -6,6 +6,7 @@ import { useState, useCallback } from "react"
 import { message } from "antd"
 import type { GeneratedVideo } from "@/api/template-editor"
 import { generateEditPlan, updateEditPlan, getEditPlan } from "@/api/template-editor"
+import { confirmGeneration } from "@/api/generation"
 import type { UseGenerateVideoProps } from "./generate-video/types"
 import { getGenerationPhase } from "./generate-video/phase"
 import { useGenerationPolling } from "./generate-video/useGenerationPolling"
@@ -55,6 +56,25 @@ export function useGenerateVideo(props: UseGenerateVideoProps) {
     clearTimer()
 
     try {
+      // 新流程：使用确认生成 API（基于预览任务）
+      if (props.previewTaskId) {
+        // 解析分辨率
+        const [widthStr, heightStr] = (props.videoRatio || "1080x1920").split("x")
+        const outputWidth = parseInt(widthStr, 10) || 1080
+        const outputHeight = parseInt(heightStr, 10) || 1920
+
+        await confirmGeneration(props.previewTaskId, {
+          output_width: outputWidth,
+          output_height: outputHeight,
+          cover_url: props.coverSettings.upload_url || "",
+          custom_title: props.titleSettings.title || "",
+        })
+
+        startPolling()
+        return
+      }
+
+      // 旧流程：使用 EditPlan API（向后兼容）
       const payload = buildEditPlanPayload(props)
 
       // 获取或创建草稿

apps/web/src/pages/generate/hooks/useStep4Preview.ts

@@ -35,6 +35,8 @@ interface UseStep4PreviewProps {
   videoRatio: string
   /** 配音 voice_ids（传给后端，让预览包含配音音频） */
   voiceIds?: string[]
+  /** 配音素材库ID（用户选择的上传音频或AI配音素材） */
+  voiceLibraryId?: string
   /** 要生成的预览数量 */
   previewCount?: number
 }
@@ -84,6 +86,7 @@ export function useStep4Preview({
   duration,
   videoRatio,
   voiceIds,
+  voiceLibraryId,
   previewCount = 1,
 }: UseStep4PreviewProps) {
   const templateName = useMemo(
@@ -332,6 +335,7 @@ export function useStep4Preview({
           duration: duration || undefined,
           video_ratio: videoRatio,
           voice_ids: voiceIds && voiceIds.length > 0 ? voiceIds : undefined,
+          voice_library_id: voiceLibraryId || undefined,
         })
 
         if (startTimeRef.current === 0) return
@@ -355,6 +359,7 @@ export function useStep4Preview({
     duration,
     videoRatio,
     voiceIds,
+    voiceLibraryId,
     previewCount,
     clearPollTimer,
     pollPreviewStatus,

apps/web/src/pages/generate/hooks/useStep6Cover.ts

@@ -1,11 +1,16 @@
 /**
  * Step 6 封面设置 Hook
- * 封装封面设置的交互逻辑
+ * 封装封面设置的交互逻辑，对接后端封面模板 CRUD API
  */
-import { useCallback, useRef } from "react"
-import type { CoverConfig } from "../../editing-planner/types"
-import { COVER_MODE_LABELS, COVER_MODE_ICONS, DEFAULT_COVER_SETTINGS } from "../constants"
+import { useCallback, useEffect, useRef, useState } from "react"
+import type { CoverConfig, CoverTemplate } from "../../editing-planner/types"
 import { generateCover } from "@/api/template-editor"
+import {
+  fetchCoverTemplates,
+  createCoverTemplate,
+  updateCoverTemplate,
+  deleteCoverTemplate,
+} from "@/api/cover-templates"
 
 interface UseStep6CoverProps {
   coverSettings: CoverConfig
@@ -26,50 +31,38 @@ export function useStep6Cover({
 }: UseStep6CoverProps) {
   const generatingRef = useRef(false)
 
-  const formatTime = useCallback((seconds: number) => {
-    const m = Math.floor(seconds / 60)
-    const s = Math.floor(seconds % 60)
-    const ms = Math.floor((seconds % 1) * 10)
-    return `${m.toString().padStart(2, "0")}:${s.toString().padStart(2, "0")}.${ms}`
+  // ── 封面设置弹窗状态 ──
+  const [showCoverSettings, setShowCoverSettings] = useState(false)
+  const [showCoverEditor, setShowCoverEditor] = useState(false)
+  const [selectedTemplateId, setSelectedTemplateId] = useState("default")
+  const [editingTemplate, setEditingTemplate] = useState<CoverTemplate | null>(null)
+  const [coverTemplates, setCoverTemplates] = useState<CoverTemplate[]>([])
+
+  // ── API 加载状态 ──
+  const [templatesLoading, setTemplatesLoading] = useState(false)
+  const [templatesError, setTemplatesError] = useState<string | null>(null)
+
+  /** 从后端加载封面模板列表 */
+  const loadTemplates = useCallback(async () => {
+    setTemplatesLoading(true)
+    setTemplatesError(null)
+    try {
+      const res = await fetchCoverTemplates()
+      setCoverTemplates(res.items || [])
+    } catch (err) {
+      console.error("[Step6] 加载封面模板失败:", err)
+      setTemplatesError("加载模板失败，请稍后重试")
+    } finally {
+      setTemplatesLoading(false)
+    }
   }, [])
 
-  const toggleEnabled = useCallback(
-    (enabled: boolean) => {
-      onCoverSettingsChange({ ...coverSettings, enabled })
-    },
-    [coverSettings, onCoverSettingsChange],
-  )
-
-  const setMode = useCallback(
-    (mode: CoverConfig["mode"]) => {
-      onCoverSettingsChange({ ...coverSettings, mode })
-    },
-    [coverSettings, onCoverSettingsChange],
-  )
-
-  const setFrameTime = useCallback(
-    (frameTime: number) => {
-      onCoverSettingsChange({ ...coverSettings, frame_time: frameTime })
-    },
-    [coverSettings, onCoverSettingsChange],
-  )
-
-  const handleUpload = useCallback(
-    (file: File) => {
-      const reader = new FileReader()
-      reader.onload = (ev) => {
-        const url = ev.target?.result as string
-        onCoverSettingsChange({
-          ...coverSettings,
-          upload_url: url,
-          thumbnail_url: url,
-          mode: "upload",
-        })
-      }
-      reader.readAsDataURL(file)
-    },
-    [coverSettings, onCoverSettingsChange],
-  )
+  /** 弹窗打开时加载模板列表 */
+  useEffect(() => {
+    if (showCoverSettings) {
+      loadTemplates()
+    }
+  }, [showCoverSettings, loadTemplates])
 
   /** 调用后端智能封面 API，生成封面并更新预览 */
   const generateAutoCover = useCallback(async () => {
@@ -95,20 +88,82 @@ export function useStep6Cover({
     }
   }, [selectedTemplate, assetIds, coverSettings, onCoverSettingsChange])
 
+  // ── 模板操作方法 ──
+  const handleSelectTemplate = useCallback((id: string) => {
+    setSelectedTemplateId(id)
+  }, [])
+
+  const handleEditTemplate = useCallback((tpl: CoverTemplate) => {
+    setEditingTemplate(tpl)
+    setShowCoverEditor(true)
+  }, [])
+
+  /** 保存模板（创建或更新） */
+  const handleSaveTemplate = useCallback(
+    async (tpl: CoverTemplate) => {
+      try {
+        if (tpl.id && coverTemplates.some((t) => t.id === tpl.id)) {
+          const updated = await updateCoverTemplate(tpl.id, {
+            name: tpl.name,
+            config: tpl.config,
+          })
+          setCoverTemplates((prev) => prev.map((t) => (t.id === tpl.id ? { ...t, ...updated } : t)))
+        } else {
+          const created = await createCoverTemplate({
+            name: tpl.name,
+            config: tpl.config,
+          })
+          setCoverTemplates((prev) => [...prev, created])
+        }
+        setShowCoverEditor(false)
+      } catch (err) {
+        console.error("[Step6] 保存模板失败:", err)
+      }
+    },
+    [coverTemplates],
+  )
+
+  /** 删除模板 */
+  const handleDeleteTemplate = useCallback(
+    async (id: string) => {
+      try {
+        await deleteCoverTemplate(id)
+        setCoverTemplates((prev) => prev.filter((t) => t.id !== id))
+        if (selectedTemplateId === id) {
+          setSelectedTemplateId("default")
+        }
+      } catch (err) {
+        console.error("[Step6] 删除模板失败:", err)
+      }
+    },
+    [selectedTemplateId],
+  )
+
+  const selectedTemplateName =
+    coverTemplates.find((t) => t.id === selectedTemplateId)?.name || "默认"
+
   const totalDuration = duration || 30
 
   return {
     coverSettings,
-    formatTime,
-    toggleEnabled,
-    setMode,
-    setFrameTime,
-    handleUpload,
     generateAutoCover,
     totalDuration,
-    COVER_MODE_LABELS,
-    COVER_MODE_ICONS,
-    DEFAULT_COVER_SETTINGS,
+    showCoverSettings,
+    setShowCoverSettings,
+    showCoverEditor,
+    setShowCoverEditor,
+    selectedTemplateId,
+    setSelectedTemplateId,
+    editingTemplate,
+    coverTemplates,
+    templatesLoading,
+    templatesError,
+    selectedTemplateName,
+    handleSelectTemplate,
+    handleEditTemplate,
+    handleSaveTemplate,
+    handleDeleteTemplate,
+    loadTemplates,
   }
 }
 

apps/worker/video_processing/unified_render_service.py

@@ -1494,9 +1494,17 @@ class UnifiedRenderService:
             raise ValueError("没有可渲染的图层")
 
         # 收集所有 clips（按图层顺序，同层按 order）
+        # 排除纯音频 clips — 它们由 mix_audio() 独立处理，不应出现在视频 filter_complex 中
+        # 例如：voice.mp3 没有视频流，如果加入 all_clips 会生成 [N:v] 引用导致 FFmpeg 报错
         all_clips: list[ResolvedClip] = []
         for layer in layers:
-            all_clips.extend(layer.clips)
+            for clip in layer.clips:
+                if clip.clip_type == "audio":
+                    continue
+                all_clips.append(clip)
+
+        if not all_clips:
+            raise ValueError("没有可渲染的视频片段（所有片段均为纯音频）")
 
         # 构建输入参数
         input_args: list[str] = []
@@ -1580,10 +1588,14 @@ class UnifiedRenderService:
             filter_parts.append(filter_str)
             preprocessed_labels.append(label)
 
-        # Step 2: 同层 clips 用 xfade 串联
+        # Step 2: 同层 clips 用 xfade 串联（跳过纯音频层，由 mix_audio() 独立处理）
         layer_output_labels: dict[str, str] = {}
         for layer in layers:
-            layer_clip_indices = [all_clips.index(c) for c in layer.clips]
+            # 音频层不参与视频 filter_complex，跳过
+            video_clips_in_layer = [c for c in layer.clips if c.clip_type != "audio"]
+            if not video_clips_in_layer:
+                continue
+            layer_clip_indices = [all_clips.index(c) for c in video_clips_in_layer]
             layer_labels = [preprocessed_labels[i] for i in layer_clip_indices]
             # 使用调速后的实际时长，与 Step 1 的调速处理保持一致
             layer_durations = [UnifiedRenderService._clip_adjusted_duration(all_clips[i]) for i in layer_clip_indices]

apps/worker/worker_app/tasks/generation.py

@@ -1058,6 +1058,7 @@ def _download_all_assets(
     task_asset_ids: list[str],
     voice_library_id: str,
     task_id: str,
+    voice_ids: list[str] | None = None,
 ) -> tuple[list[Path], str | None]:
     """下载视频素材和配音素材。
 
@@ -1066,6 +1067,9 @@ def _download_all_assets(
 
     Note: gen_task 不传入下载函数（session 已关闭），
     主函数在下载前后已有汇总日志。
+
+    配音下载逻辑：优先使用 voice_library_id（配音素材库资产）；
+    若为空则 fallback 到 voice_ids[0]（前端选择的音频 asset_id）。
     """
     logger.info("[task_id=%s] [下载素材] 开始下载视频素材", task_id)
     download_start = time.monotonic()
@@ -1085,11 +1089,24 @@ def _download_all_assets(
     )
 
     audio_path: str | None = None
-    if voice_library_id:
+    # 配音下载：优先 voice_library_id，fallback 到 voice_ids[0]
+    effective_voice_id = voice_library_id
+    if not effective_voice_id and voice_ids:
+        effective_voice_id = voice_ids[0]
+        logger.info(
+            "[task_id=%s] [下载配音] voice_library_id 为空，fallback 到 voice_ids[0]=%s",
+            task_id,
+            effective_voice_id,
+        )
+    if effective_voice_id:
         local_audio = temp_path / "voice.mp3"
-        if _download_voice_asset(voice_library_id, local_audio):
+        if _download_voice_asset(effective_voice_id, local_audio):
             audio_path = str(local_audio)
-            logger.info("[task_id=%s] [下载配音] 配音下载成功", task_id)
+            logger.info(
+                "[task_id=%s] [下载配音] 配音下载成功 (source=%s)",
+                task_id,
+                "voice_library_id" if voice_library_id else "voice_ids",
+            )
 
     return downloaded_videos, audio_path
 
@@ -1422,6 +1439,7 @@ def generate_video(self, task_id: str) -> dict:
                 task_asset_ids=task_asset_ids,
                 voice_library_id=voice_library_id,
                 task_id=task_id,
+                voice_ids=task_info.get("voice_ids", []),
             )
 
             if gen_task:

tests/unit/test_audio_only_clips_in_filter_complex.py

@@ -0,0 +1,262 @@
+"""测试 _build_filter_complex 正确排除纯音频 clips.
+
+Bug: voice.mp3（纯音频文件）被错误地加入视频 filter_complex，
+导致 FFmpeg 尝试访问 [N:v] 视频流时报错 "Stream specifier ':v' matches no streams".
+
+修复：_build_filter_complex 在收集 clips 时跳过 clip_type="audio" 的 clips，
+因为音频 clips 由 mix_audio() 独立处理。
+"""
+
+from pathlib import Path
+from unittest.mock import MagicMock
+
+import pytest
+
+
+def _make_render_service(tmp_path):
+    """创建一个最小化的 UnifiedRenderService 实例."""
+    from video_processing.unified_render_service import UnifiedRenderService
+
+    plan = MagicMock()
+    plan.id = "test_plan"
+    plan.config = {}
+    plan.strategy_id = "test_strategy"
+
+    clips = []
+    asset_path_map = {}
+
+    service = UnifiedRenderService(
+        plan=plan,
+        clips=clips,
+        asset_path_map=asset_path_map,
+        work_dir=tmp_path,
+        output_width=480,
+        output_height=854,
+        output_fps=30,
+    )
+    return service
+
+
+def _make_video_clip(clip_id: str, local_path: Path, duration: float = 5.0):
+    """创建一个视频 clip."""
+    from video_processing.unified_render_service import ResolvedClip
+
+    return ResolvedClip(
+        clip_id=clip_id,
+        asset_id=f"asset_{clip_id}",
+        local_path=local_path,
+        clip_type="video",
+        order=0,
+        start_time=0.0,
+        duration=duration,
+        config={},
+        actual_duration=duration,
+    )
+
+
+def _make_audio_clip(clip_id: str, local_path: Path, duration: float = 5.0):
+    """创建一个纯音频 clip."""
+    from video_processing.unified_render_service import ResolvedClip
+
+    return ResolvedClip(
+        clip_id=clip_id,
+        asset_id=f"asset_{clip_id}",
+        local_path=local_path,
+        clip_type="audio",
+        order=0,
+        start_time=0.0,
+        duration=duration,
+        config={"volume": 1.0},
+        actual_duration=duration,
+    )
+
+
+class TestBuildFilterComplexExcludesAudioClips:
+    """_build_filter_complex 应该排除 clip_type='audio' 的 clips."""
+
+    def test_audio_clip_not_in_filter_complex(self, tmp_path):
+        """纯音频 clip 不应出现在 filter_complex 中."""
+        render_service = _make_render_service(tmp_path)
+        from video_processing.unified_render_service import RenderLayer
+
+        # 准备视频和音频文件
+        video_path = tmp_path / "video.mp4"
+        video_path.write_bytes(b"\x00")
+        audio_path = tmp_path / "voice.mp3"
+        audio_path.write_bytes(b"\x00")
+
+        video_clip = _make_video_clip("clip_0", video_path, duration=5.0)
+        audio_clip = _make_audio_clip("voice_library_main", audio_path, duration=5.0)
+
+        video_layer = RenderLayer(role="main", z_index=1)
+        video_layer.clips.append(video_clip)
+
+        audio_layer = RenderLayer(role="audio", z_index=2)
+        audio_layer.clips.append(audio_clip)
+
+        layers = [video_layer, audio_layer]
+
+        # 执行
+        filter_complex, input_args = render_service._build_filter_complex(layers)
+
+        # 验证：filter_complex 只包含视频 clip 的处理（[0:v]），不包含音频 clip（[1:v]）
+        assert "[0:v]" in filter_complex
+        assert "[1:v]" not in filter_complex  # voice.mp3 不应该有视频滤镜
+
+        # 验证：input_args 只包含视频文件，不包含音频文件
+        assert str(video_path) in " ".join(input_args)
+        assert str(audio_path) not in " ".join(input_args)
+
+    def test_multiple_video_clips_with_audio(self, tmp_path):
+        """多个视频 clips + 音频 clip 时，filter_complex 只处理视频."""
+        render_service = _make_render_service(tmp_path)
+        from video_processing.unified_render_service import RenderLayer
+
+        # 准备文件
+        video_paths = [tmp_path / f"video_{i}.mp4" for i in range(3)]
+        for p in video_paths:
+            p.write_bytes(b"\x00")
+        audio_path = tmp_path / "voice.mp3"
+        audio_path.write_bytes(b"\x00")
+
+        video_layer = RenderLayer(role="main", z_index=1)
+        for i, vp in enumerate(video_paths):
+            clip = _make_video_clip(f"clip_{i}", vp, duration=3.0)
+            clip.order = i
+            video_layer.clips.append(clip)
+
+        audio_layer = RenderLayer(role="audio", z_index=2)
+        audio_clip = _make_audio_clip("voice_main", audio_path, duration=9.0)
+        audio_layer.clips.append(audio_clip)
+
+        layers = [video_layer, audio_layer]
+
+        # 执行
+        filter_complex, input_args = render_service._build_filter_complex(layers)
+
+        # 验证：只有 3 个视频输入
+        assert "[0:v]" in filter_complex
+        assert "[1:v]" in filter_complex
+        assert "[2:v]" in filter_complex
+        assert "[3:v]" not in filter_complex  # 音频不应该出现
+
+        # 验证：input_args 只有 3 个 -i
+        input_files = [arg for arg in input_args if not arg.startswith("-")]
+        assert len(input_files) == 3
+        assert str(audio_path) not in input_files
+
+    def test_only_audio_clips_raises_error(self, tmp_path):
+        """只有音频 clips 时应该抛出 ValueError."""
+        render_service = _make_render_service(tmp_path)
+        from video_processing.unified_render_service import RenderLayer
+
+        audio_path = tmp_path / "voice.mp3"
+        audio_path.write_bytes(b"\x00")
+
+        audio_layer = RenderLayer(role="audio", z_index=2)
+        audio_clip = _make_audio_clip("voice_main", audio_path, duration=5.0)
+        audio_layer.clips.append(audio_clip)
+
+        layers = [audio_layer]
+
+        with pytest.raises(ValueError, match="没有可渲染的视频片段"):
+            render_service._build_filter_complex(layers)
+
+    def test_tts_audio_clip_excluded(self, tmp_path):
+        """TTS 配音 clip（clip_type='audio', config.tts=True）也不应出现在 filter_complex."""
+        render_service = _make_render_service(tmp_path)
+        from video_processing.unified_render_service import RenderLayer, ResolvedClip
+
+        video_path = tmp_path / "video.mp4"
+        video_path.write_bytes(b"\x00")
+        tts_audio_path = tmp_path / "tts_segment.wav"
+        tts_audio_path.write_bytes(b"\x00")
+
+        video_clip = _make_video_clip("clip_0", video_path, duration=10.0)
+        tts_clip = ResolvedClip(
+            clip_id="tts_0.000",
+            asset_id="tts_voiceover",
+            local_path=tts_audio_path,
+            clip_type="audio",
+            order=0,
+            start_time=0.0,
+            duration=3.0,
+            config={"volume": 1.0, "tts": True},
+            actual_duration=3.0,
+        )
+
+        video_layer = RenderLayer(role="main", z_index=1)
+        video_layer.clips.append(video_clip)
+
+        audio_layer = RenderLayer(role="audio", z_index=2)
+        audio_layer.clips.append(tts_clip)
+
+        layers = [video_layer, audio_layer]
+
+        filter_complex, input_args = render_service._build_filter_complex(layers)
+
+        # TTS 音频不应出现在 filter_complex
+        assert "[1:v]" not in filter_complex
+        input_files = [arg for arg in input_args if not arg.startswith("-")]
+        assert str(tts_audio_path) not in input_files
+
+    def test_video_clip_with_audio_config_not_excluded(self, tmp_path):
+        """clip_type='video' 的 clip 不应被排除（即使它有音频流）."""
+        render_service = _make_render_service(tmp_path)
+        from video_processing.unified_render_service import RenderLayer
+
+        video_path = tmp_path / "video.mp4"
+        video_path.write_bytes(b"\x00")
+
+        video_clip = _make_video_clip("clip_0", video_path, duration=5.0)
+        video_layer = RenderLayer(role="main", z_index=1)
+        video_layer.clips.append(video_clip)
+
+        layers = [video_layer]
+
+        filter_complex, input_args = render_service._build_filter_complex(layers)
+
+        # 视频 clip 应该被处理
+        assert "[0:v]" in filter_complex
+        input_files = [arg for arg in input_args if not arg.startswith("-")]
+        assert str(video_path) in input_files
+
+    def test_voice_library_clip_with_voice_library_flag(self, tmp_path):
+        """voice_library=True 的 clip（来自 _maybe_add_voice_library_layer）应被排除."""
+        render_service = _make_render_service(tmp_path)
+        from video_processing.unified_render_service import RenderLayer, ResolvedClip
+
+        video_path = tmp_path / "video.mp4"
+        video_path.write_bytes(b"\x00")
+        voice_path = tmp_path / "voice.mp3"
+        voice_path.write_bytes(b"\x00")
+
+        video_clip = _make_video_clip("clip_0", video_path, duration=14.0)
+        # 模拟 _maybe_add_voice_library_layer 创建的 clip
+        voice_clip = ResolvedClip(
+            clip_id="voice_library_main",
+            asset_id="voice_library",
+            local_path=voice_path,
+            clip_type="audio",
+            order=0,
+            start_time=0.0,
+            duration=14.0,
+            config={"volume": 1.0, "voice_library": True},
+            actual_duration=14.0,
+        )
+
+        video_layer = RenderLayer(role="main", z_index=1)
+        video_layer.clips.append(video_clip)
+
+        audio_layer = RenderLayer(role="audio", z_index=2)
+        audio_layer.clips.append(voice_clip)
+
+        layers = [video_layer, audio_layer]
+
+        filter_complex, input_args = render_service._build_filter_complex(layers)
+
+        # 只有视频 clip 被处理
+        assert "[0:v]" in filter_complex
+        assert "[1:v]" not in filter_complex
+        # voice.mp3 不在输入中
+        assert str(voice_path) not in " ".join(input_args)

tests/unit/test_preview_voice_ids_fallback.py

@@ -0,0 +1,151 @@
+"""Tests for voice_ids fallback in _download_all_assets.
+
+When voice_library_id is empty but voice_ids is non-empty, the Worker
+should fallback to voice_ids[0] as the audio asset_id.
+"""
+
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+
+class TestDownloadAllAssetsVoiceIdsFallback:
+    """_download_all_assets 配音下载 fallback 逻辑测试。"""
+
+    @patch("worker_app.tasks.generation._download_voice_asset")
+    @patch("worker_app.tasks.generation._download_library_assets")
+    def test_voice_library_id_takes_priority(self, mock_download_videos, mock_download_voice, tmp_path):
+        """voice_library_id 存在时优先使用，不 fallback 到 voice_ids。"""
+        from worker_app.tasks.generation import _download_all_assets
+
+        mock_download_videos.return_value = [tmp_path / "v1.mp4"]
+        mock_download_voice.return_value = True
+
+        videos, audio = _download_all_assets(
+            temp_path=tmp_path,
+            asset_library_id="lib-1",
+            project_id="proj-1",
+            task_asset_ids=["a1"],
+            voice_library_id="voice-lib-123",
+            task_id="task-1",
+            voice_ids=["voice-ids-456"],
+        )
+
+        assert audio is not None
+        mock_download_voice.assert_called_once()
+        call_args = mock_download_voice.call_args
+        assert call_args[0][0] == "voice-lib-123"  # first positional arg
+
+    @patch("worker_app.tasks.generation._download_voice_asset")
+    @patch("worker_app.tasks.generation._download_library_assets")
+    def test_fallback_to_voice_ids_when_voice_library_id_empty(
+        self, mock_download_videos, mock_download_voice, tmp_path
+    ):
+        """voice_library_id 为空时 fallback 到 voice_ids[0]。"""
+        from worker_app.tasks.generation import _download_all_assets
+
+        mock_download_videos.return_value = [tmp_path / "v1.mp4"]
+        mock_download_voice.return_value = True
+
+        videos, audio = _download_all_assets(
+            temp_path=tmp_path,
+            asset_library_id="lib-1",
+            project_id="proj-1",
+            task_asset_ids=["a1"],
+            voice_library_id="",  # 空字符串
+            task_id="task-2",
+            voice_ids=["voice-asset-789"],
+        )
+
+        assert audio is not None
+        mock_download_voice.assert_called_once()
+        call_args = mock_download_voice.call_args
+        assert call_args[0][0] == "voice-asset-789"
+
+    @patch("worker_app.tasks.generation._download_voice_asset")
+    @patch("worker_app.tasks.generation._download_library_assets")
+    def test_no_audio_when_both_empty(self, mock_download_videos, mock_download_voice, tmp_path):
+        """voice_library_id 和 voice_ids 都为空时，不下载音频。"""
+        from worker_app.tasks.generation import _download_all_assets
+
+        mock_download_videos.return_value = [tmp_path / "v1.mp4"]
+
+        videos, audio = _download_all_assets(
+            temp_path=tmp_path,
+            asset_library_id="lib-1",
+            project_id="proj-1",
+            task_asset_ids=["a1"],
+            voice_library_id="",
+            task_id="task-3",
+            voice_ids=[],
+        )
+
+        assert audio is None
+        mock_download_voice.assert_not_called()
+
+    @patch("worker_app.tasks.generation._download_voice_asset")
+    @patch("worker_app.tasks.generation._download_library_assets")
+    def test_no_audio_when_voice_ids_none(self, mock_download_videos, mock_download_voice, tmp_path):
+        """voice_ids 为 None 时，不触发 fallback。"""
+        from worker_app.tasks.generation import _download_all_assets
+
+        mock_download_videos.return_value = [tmp_path / "v1.mp4"]
+
+        videos, audio = _download_all_assets(
+            temp_path=tmp_path,
+            asset_library_id="lib-1",
+            project_id="proj-1",
+            task_asset_ids=["a1"],
+            voice_library_id="",
+            task_id="task-4",
+            voice_ids=None,
+        )
+
+        assert audio is None
+        mock_download_voice.assert_not_called()
+
+    @patch("worker_app.tasks.generation._download_voice_asset")
+    @patch("worker_app.tasks.generation._download_library_assets")
+    def test_voice_library_id_empty_string_fallback(self, mock_download_videos, mock_download_voice, tmp_path):
+        """voice_library_id 为空字符串且 voice_ids 有多个元素时，取第一个。"""
+        from worker_app.tasks.generation import _download_all_assets
+
+        mock_download_videos.return_value = [tmp_path / "v1.mp4"]
+        mock_download_voice.return_value = True
+
+        videos, audio = _download_all_assets(
+            temp_path=tmp_path,
+            asset_library_id="lib-1",
+            project_id="proj-1",
+            task_asset_ids=["a1"],
+            voice_library_id="",
+            task_id="task-5",
+            voice_ids=["first-id", "second-id", "third-id"],
+        )
+
+        assert audio is not None
+        call_args = mock_download_voice.call_args
+        assert call_args[0][0] == "first-id"
+
+    @patch("worker_app.tasks.generation._download_voice_asset")
+    @patch("worker_app.tasks.generation._download_library_assets")
+    def test_backward_compat_no_voice_ids_param(self, mock_download_videos, mock_download_voice, tmp_path):
+        """不传 voice_ids 参数时，行为与之前一致（向后兼容）。"""
+        from worker_app.tasks.generation import _download_all_assets
+
+        mock_download_videos.return_value = [tmp_path / "v1.mp4"]
+        mock_download_voice.return_value = True
+
+        # 不传 voice_ids
+        videos, audio = _download_all_assets(
+            temp_path=tmp_path,
+            asset_library_id="lib-1",
+            project_id="proj-1",
+            task_asset_ids=["a1"],
+            voice_library_id="voice-lib-999",
+            task_id="task-6",
+        )
+
+        assert audio is not None
+        mock_download_voice.assert_called_once_with("voice-lib-999", tmp_path / "voice.mp3")

tests/unit/test_preview_voice_library_id.py

@@ -0,0 +1,74 @@
+"""预览生成 voice_library_id 透传修复测试.
+
+Bug: 预览生成接口硬编码 voice_library_id=""，导致用户选择的上传音频
+在预览渲染时从未下载和混入，预览视频无声。
+
+Fix: CreatePreviewGenerationTaskRequest 增加 voice_library_id 字段，
+预览端点透传 request.voice_library_id。
+"""
+
+from __future__ import annotations
+
+import os
+import sys
+from pathlib import Path
+
+os.environ.setdefault("JWT_SECRET_KEY", "unit-test-secret-key-for-testing")
+os.environ.setdefault("DATABASE_URL", "sqlite:///test.db")
+
+sys.path.insert(0, str(Path(__file__).resolve().parents[2] / "apps" / "api"))
+
+
+class TestPreviewVoiceLibraryIdSchema:
+    """CreatePreviewGenerationTaskRequest voice_library_id 字段测试."""
+
+    def test_default_empty_string(self):
+        """不传 voice_library_id 时默认为空字符串."""
+        from app.schemas.generation_task import CreatePreviewGenerationTaskRequest
+
+        req = CreatePreviewGenerationTaskRequest(
+            template_id="tmpl_1",
+            asset_ids=["a1"],
+        )
+        assert req.voice_library_id == ""
+
+    def test_accepts_voice_library_id(self):
+        """传入 voice_library_id 正常接收."""
+        from app.schemas.generation_task import CreatePreviewGenerationTaskRequest
+
+        req = CreatePreviewGenerationTaskRequest(
+            template_id="tmpl_1",
+            asset_ids=["a1"],
+            voice_library_id="asset_abc123",
+        )
+        assert req.voice_library_id == "asset_abc123"
+
+    def test_accepts_empty_voice_library_id(self):
+        """显式传空字符串也正常."""
+        from app.schemas.generation_task import CreatePreviewGenerationTaskRequest
+
+        req = CreatePreviewGenerationTaskRequest(
+            template_id="tmpl_1",
+            asset_ids=["a1"],
+            voice_library_id="",
+        )
+        assert req.voice_library_id == ""
+
+    def test_all_fields_including_voice_library_id(self):
+        """包含 voice_library_id 的完整请求."""
+        from app.schemas.generation_task import CreatePreviewGenerationTaskRequest
+
+        req = CreatePreviewGenerationTaskRequest(
+            template_id="tmpl_123",
+            asset_ids=["a1", "a2"],
+            title_ids=["t1"],
+            voice_ids=["v1"],
+            voice_library_id="voice_asset_456",
+            video_title="测试预览",
+            duration=30.0,
+            video_ratio="9:16",
+            bgm_config={"enabled": True, "volume": 0.5},
+        )
+        assert req.voice_library_id == "voice_asset_456"
+        assert req.asset_ids == ["a1", "a2"]
+        assert req.bgm_config["enabled"] is True