fix: 移除配音选择时长过滤逻辑，所有配音素材完整显示 (#1697)

Co-authored-by: xiaoxia <dev@xiaoxiajianji.com>
Co-committed-by: xiaoxia <dev@xiaoxiajianji.com>

apps/web/src/pages/editing-planner/components/EditorClipList.tsx

@@ -40,13 +40,6 @@ const clipTypeLabel: Record<ClipType | string, string> = {
   pip: "混剪",
 }
 
-const formatDuration = (sec: number) => {
-  if (sec < 60) return `${sec.toFixed(1)}s`
-  const m = Math.floor(sec / 60)
-  const s = (sec % 60).toFixed(0)
-  return `${m}m${s.padStart(2, "0")}s`
-}
-
 const EditorClipList: React.FC<EditorClipListProps> = ({
   clips,
   selectedClipId,
@@ -102,7 +95,6 @@ const EditorClipList: React.FC<EditorClipListProps> = ({
                   {clipTypeLabel[clip.type] || "片段"}
                 </span>
               </span>
-              <span className="ep-clip-item-duration">{formatDuration(clip.duration)}</span>
             </div>
 
             {/* 文案预览 */}

apps/web/src/pages/editing-planner/components/PreviewPlayer.tsx

@@ -87,7 +87,7 @@ const PreviewPlayer: React.FC<PreviewPlayerProps> = ({
                 WebkitTextStroke: "1px rgba(0,0,0,0.6)",
                 top:
                   titleConfig.position === "top"
-                    ? "8px"
+                    ? "6.25%"
                     : titleConfig.position === "center"
                       ? "50%"
                       : "auto",

apps/web/src/pages/editing-planner/components/TimelinePanel.tsx

@@ -7,7 +7,6 @@
  * - ClipCard - 片段卡片
  * - ClipTrack - 片段轨道（播放头+片段列表+添加卡片）
  * - TimelineHeader - 时间线头部（标题+缩放+操作按钮）
- * - AddClipPicker - 添加片段选择器
  * - TrimPreview - 裁剪预览 tooltip
  * - ContextMenu - 右键菜单
  *
@@ -27,7 +26,6 @@ import { usePlayheadDrag } from "./timeline/hooks/usePlayheadDrag"
 import { TimeRuler } from "./timeline/TimeRuler"
 import { ClipTrack } from "./timeline/ClipTrack"
 import { TimelineHeader } from "./timeline/TimelineHeader"
-import { AddClipPicker } from "./timeline/AddClipPicker"
 import { TrimPreview } from "./timeline/TrimPreview"
 import { ContextMenu } from "./timeline/ContextMenu"
 
@@ -100,16 +98,7 @@ const TimelinePanel: React.FC<TimelinePanelProps> = ({
     handleContextSplit,
     handleContextResetTrim,
     handleContextDelete,
-    showAddPicker,
-    pickerRef,
     addCardRef,
-    pickerPos,
-    availableTypes,
-    addType,
-    addDuration,
-    setAddType,
-    setAddDuration,
-    handleTogglePicker,
     handleConfirmAdd,
     hoveredClipId,
     setHoveredClipId,
@@ -177,7 +166,7 @@ const TimelinePanel: React.FC<TimelinePanelProps> = ({
           onClipMouseLeave={() => setHoveredClipId(null)}
           onTrimHandleMouseDown={handleTrimHandleMouseDown}
           onClipRemove={onClipRemove}
-          onTogglePicker={handleTogglePicker}
+          onTogglePicker={handleConfirmAdd}
         />
       )}
 
@@ -204,20 +193,6 @@ const TimelinePanel: React.FC<TimelinePanelProps> = ({
           onDelete={handleContextDelete}
         />
       )}
-
-      {/* 类型+时长选择面板 */}
-      {showAddPicker && (
-        <AddClipPicker
-          pickerRef={pickerRef}
-          position={pickerPos}
-          availableTypes={availableTypes}
-          addType={addType}
-          addDuration={addDuration}
-          onTypeChange={setAddType}
-          onDurationChange={setAddDuration}
-          onConfirm={handleConfirmAdd}
-        />
-      )}
     </div>
   )
 }

apps/web/src/pages/editing-planner/components/timeline/AddClipPicker.tsx

@@ -7,12 +7,8 @@ interface AddClipPickerProps {
   position: { top: number; right: number }
   availableTypes: ClipType[]
   addType: ClipType
-  addDuration: number
   onTypeChange: (type: ClipType) => void
-  onDurationChange: (duration: number) => void
   onConfirm: () => void
-  minDuration?: number
-  maxDuration?: number
 }
 
 export const AddClipPicker: React.FC<AddClipPickerProps> = ({
@@ -20,12 +16,8 @@ export const AddClipPicker: React.FC<AddClipPickerProps> = ({
   position,
   availableTypes,
   addType,
-  addDuration,
   onTypeChange,
-  onDurationChange,
   onConfirm,
-  minDuration = 1,
-  maxDuration = 120,
 }) => {
   return (
     <div
@@ -53,24 +45,6 @@ export const AddClipPicker: React.FC<AddClipPickerProps> = ({
         ))}
       </div>
 
-      {/* 时长输入 */}
-      <div className="ep-add-clip-duration-row">
-        <span className="ep-add-clip-type-label">时长:</span>
-        <input
-          type="number"
-          className="ep-duration-input"
-          min={minDuration}
-          max={maxDuration}
-          value={addDuration}
-          onChange={(e) =>
-            onDurationChange(
-              Math.max(minDuration, Math.min(maxDuration, Number(e.target.value) || minDuration)),
-            )
-          }
-        />
-        <span className="ep-add-clip-duration-unit">秒</span>
-      </div>
-
       {/* 确认按钮 */}
       <button className="ep-add-clip-confirm-btn" onClick={onConfirm}>
         添加

apps/web/src/pages/editing-planner/components/timeline/ClipCard.tsx

@@ -105,14 +105,11 @@ export const ClipCard: React.FC<ClipCardProps> = ({
           <span className="ep-clip-name">
             {CLIP_TYPE_LABELS[clip.type] || "片段"} {idx + 1}
           </span>
-          <span className="ep-clip-duration">
-            {clip.duration}s
-            {hasTrim && (
-              <span className="ep-trim-indicator" title="已裁剪">
-                ✂
-              </span>
-            )}
-          </span>
+          {hasTrim && (
+            <span className="ep-trim-indicator" title="已裁剪">
+              ✂
+            </span>
+          )}
         </div>
 
         {/* 速度徽章 */}

apps/web/src/pages/editing-planner/hooks/timeline-menus/useAddPicker.ts

@@ -26,8 +26,6 @@ export function useAddPicker({ currentMode, onAddClip }: UseAddPickerOptions) {
   }, [currentMode])
 
   const [addType, setAddType] = useState<ClipType>(defaultAddType)
-  const [addDuration, setAddDuration] = useState<number>(DEFAULT_ADD_DURATION)
-
   useEffect(() => {
     if (!availableTypes.includes(addType)) {
       setAddType(defaultAddType)
@@ -95,9 +93,9 @@ export function useAddPicker({ currentMode, onAddClip }: UseAddPickerOptions) {
   }, [showAddPicker])
 
   const handleConfirmAdd = useCallback(() => {
-    onAddClip(addType, addDuration)
+    onAddClip(addType, DEFAULT_ADD_DURATION)
     setShowAddPicker(false)
-  }, [onAddClip, addType, addDuration])
+  }, [onAddClip, addType])
 
   return {
     showAddPicker,
@@ -107,9 +105,8 @@ export function useAddPicker({ currentMode, onAddClip }: UseAddPickerOptions) {
     pickerPos,
     availableTypes,
     addType,
-    addDuration,
+    addDuration: DEFAULT_ADD_DURATION,
     setAddType,
-    setAddDuration,
     handleTogglePicker,
     handleConfirmAdd,
   }

apps/web/src/pages/editing-planner/hooks/useTimelineMenus.ts

@@ -35,7 +35,6 @@ export const useTimelineMenus = (
     addType,
     addDuration,
     setAddType,
-    setAddDuration,
     handleTogglePicker,
     handleConfirmAdd,
   } = useAddPicker({ currentMode, onAddClip })
@@ -57,7 +56,6 @@ export const useTimelineMenus = (
     addType,
     addDuration,
     setAddType,
-    setAddDuration,
     handleTogglePicker,
     handleConfirmAdd,
     // 悬停状态

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

@@ -16,10 +16,6 @@ import { useQuery } from "@tanstack/react-query"
 import { useCloneProgress } from "@/hooks/useCloneProgress"
 import CloneModal from "@/components/voice/CloneModal"
 import GenerateHeader from "./components/GenerateHeader"
-import {
-  calculateTotalVideoDuration,
-  estimateTotalVideoDuration,
-} from "./utils/calculateTotalVideoDuration"
 import FrontendPreviewPlayer from "./components/FrontendPreviewPlayer"
 import GenerateStepsBar from "./components/GenerateStepsBar"
 import GenerateStepContent from "./components/GenerateStepContent"
@@ -128,7 +124,6 @@ const GeneratePage: React.FC = () => {
       cancelled = true
       controller.abort()
     }
-    // eslint-disable-next-line react-hooks/exhaustive-deps
   }, [selectedVoice, selectedClonedVoice, titleSettings.title, voiceMaterials])
 
   /* ── 克隆声音 ── */
@@ -174,13 +169,6 @@ const GeneratePage: React.FC = () => {
     [previewAssetsReady, currentTemplate],
   )
 
-  /* ── 视频总时长计算 ── */
-  const totalVideoDuration = useMemo(() => {
-    const exact = calculateTotalVideoDuration(previewAssets, currentTemplate ?? undefined)
-    if (exact > 0) return exact
-    return estimateTotalVideoDuration(currentTemplate ?? undefined)
-  }, [previewAssets, currentTemplate])
-
   /* ── 视频生成核心逻辑 ── */
   const {
     generating,
@@ -291,7 +279,6 @@ const GeneratePage: React.FC = () => {
             onCoverSettingsChange={setCoverSettings}
             selectedVoice={selectedVoice}
             onSelectedVoiceChange={setSelectedVoice}
-            totalVideoDuration={totalVideoDuration}
             onServerClipsChange={setServerClips}
             voiceMode={voiceMode}
             onVoiceModeChange={setVoiceMode}

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

@@ -591,6 +591,8 @@ const FrontendPreviewPlayer: React.FC<FrontendPreviewPlayerProps> = ({
           <div
             style={{
               position: "absolute",
+              width: `${100 - 2 * titleSidePct}%`,
+              maxWidth: `${100 - 2 * titleSidePct}%`,
               ...(customTitleXPct != null && customTitleYPct != null
                 ? {
                     left: `${customTitleXPct}%`,
@@ -599,13 +601,13 @@ const FrontendPreviewPlayer: React.FC<FrontendPreviewPlayerProps> = ({
                     textAlign: "center" as const,
                   }
                 : {
-                    left: `${titleSidePct}%`,
-                    right: `${titleSidePct}%`,
+                    left: "50%",
+                    transform: "translateX(-50%)",
                     textAlign: "center" as const,
                     ...(titleSettings.position === "top"
                       ? { top: `${titleTopPct}%` }
                       : titleSettings.position === "center"
-                        ? { top: "50%", transform: "translateY(-50%)" }
+                        ? { top: "50%", transform: "translate(-50%, -50%)" }
                         : { bottom: `${titleBottomPct}%` }),
                   }),
               pointerEvents: "auto",

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

@@ -49,7 +49,6 @@ export interface GenerateStepContentProps {
   /* 配音 */
   selectedVoice: string
   onSelectedVoiceChange: (id: string) => void
-  totalVideoDuration?: number
   onServerClipsChange: (clips: EditPlanClip[]) => void
   voiceMode: "preset" | "custom" | "clone"
   onVoiceModeChange: (mode: "preset" | "custom" | "clone") => void
@@ -104,7 +103,6 @@ export const GenerateStepContent: React.FC<GenerateStepContentProps> = (props) =
     onCoverSettingsChange,
     selectedVoice,
     onSelectedVoiceChange,
-    totalVideoDuration,
     onServerClipsChange,
     voiceMode,
     selectedClonedVoice,
@@ -151,7 +149,6 @@ export const GenerateStepContent: React.FC<GenerateStepContentProps> = (props) =
         <Step3VoiceSelect
           selectedVoice={selectedVoice}
           onSelectedVoiceChange={onSelectedVoiceChange}
-          totalVideoDuration={totalVideoDuration}
         />
       )
     case 4:

apps/web/src/pages/generate/components/Step1TemplateSelect.tsx

@@ -47,9 +47,7 @@ const Step1TemplateSelect: React.FC<Step1TemplateSelectProps> = (props) => {
                 🎬
               </div>
               <h4>{tpl.name}</h4>
-              <p>
-                {tpl.estimated_duration}s · {tpl.segments.length}片段
-              </p>
+              <p>{tpl.segments.length}片段</p>
               {tpl.tags.length > 0 && (
                 <div
                   style={{

apps/web/src/pages/generate/components/Step5VoiceSelect.tsx

@@ -5,18 +5,15 @@
 import React, { useState, useRef, useCallback } from "react"
 import { useNavigate } from "react-router-dom"
 import { useQuery } from "@tanstack/react-query"
-import { AudioOutlined, SoundOutlined, WarningOutlined } from "@ant-design/icons"
-import { Modal } from "antd"
+import { AudioOutlined, SoundOutlined } from "@ant-design/icons"
 import { getAssetsByKind } from "@/api/assets"
 import type { AssetItem } from "@/api/assets"
 
 interface Step5VoiceSelectProps {
   selectedVoice: string
   onSelectedVoiceChange: (id: string) => void
-  totalVideoDuration?: number
 }
 
-/** 格式化时长 mm:ss */
 /** 获取素材实际时长（优先顶层 duration，fallback 到 metadata.duration） */
 const getDuration = (item: AssetItem): number => {
   return item.duration ?? (item.metadata?.duration as number) ?? 0
@@ -34,13 +31,6 @@ const isAiVoice = (item: AssetItem): boolean => {
   return (!duration || duration <= 0) && (!size || size <= 0)
 }
 
-const formatDuration = (seconds?: number): string => {
-  if (!seconds || seconds <= 0) return "00:00"
-  const m = Math.floor(seconds / 60)
-  const s = Math.floor(seconds % 60)
-  return `${String(m).padStart(2, "0")}:${String(s).padStart(2, "0")}`
-}
-
 /** 格式化文件大小 */
 const formatFileSize = (bytes?: number): string => {
   if (!bytes || bytes <= 0) return "未知"
@@ -53,13 +43,10 @@ const formatFileSize = (bytes?: number): string => {
 const Step5VoiceSelect: React.FC<Step5VoiceSelectProps> = ({
   selectedVoice,
   onSelectedVoiceChange,
-  totalVideoDuration = 0,
 }) => {
   const navigate = useNavigate()
   const [playingId, setPlayingId] = useState<string | null>(null)
   const audioRef = useRef<HTMLAudioElement | null>(null)
-  const [durationWarningOpen, setDurationWarningOpen] = useState(false)
-  const [pendingVoiceId, setPendingVoiceId] = useState<string | null>(null)
 
   // 获取用户上传的配音素材
   const { data: materials = [], isLoading } = useQuery({
@@ -100,38 +87,14 @@ const Step5VoiceSelect: React.FC<Step5VoiceSelectProps> = ({
     [playingId],
   )
 
-  /** 选中素材（含时长校验） */
+  /** 选中素材（直接选中，不再做时长校验弹窗） */
   const handleSelect = useCallback(
     (id: string) => {
-      // 如果启用了时长校验，且配音时长不足（AI 音色按脚本实时合成，不参与时长校验）
-      if (totalVideoDuration > 0) {
-        const material = materials.find((m) => m.id === id)
-        if (material && !isAiVoice(material) && getDuration(material) < totalVideoDuration) {
-          setPendingVoiceId(id)
-          setDurationWarningOpen(true)
-          return
-        }
-      }
       onSelectedVoiceChange(id)
     },
-    [onSelectedVoiceChange, totalVideoDuration, materials],
+    [onSelectedVoiceChange],
   )
 
-  /** 确认使用时长不足的配音 */
-  const handleConfirmUseAnyway = useCallback(() => {
-    if (pendingVoiceId) {
-      onSelectedVoiceChange(pendingVoiceId)
-    }
-    setDurationWarningOpen(false)
-    setPendingVoiceId(null)
-  }, [pendingVoiceId, onSelectedVoiceChange])
-
-  /** 取消选择 */
-  const handleCancelSelection = useCallback(() => {
-    setDurationWarningOpen(false)
-    setPendingVoiceId(null)
-  }, [])
-
   /** 跳转到配音库上传 */
   const handleGoToUpload = useCallback(() => {
     navigate("/app/voices?tab=material&upload=1")
@@ -277,7 +240,7 @@ const Step5VoiceSelect: React.FC<Step5VoiceSelectProps> = ({
                 {item.name}
               </div>
 
-              {/* 时长 + 大小 */}
+              {/* 文件大小 */}
               <div
                 style={{
                   display: "flex",
@@ -289,65 +252,13 @@ const Step5VoiceSelect: React.FC<Step5VoiceSelectProps> = ({
               >
                 {isAiVoice(item) ? (
                   <span style={{ color: "#1677ff", fontWeight: 500 }}>AI 音色</span>
-                ) : (
-                  <span style={{ display: "flex", alignItems: "center", gap: 4 }}>
-                    {formatDuration(getDuration(item))}
-                    {totalVideoDuration > 0 && getDuration(item) < Number(totalVideoDuration) && (
-                      <span
-                        style={{
-                          color: "#ff4d4f",
-                          fontSize: 11,
-                          fontWeight: 500,
-                          display: "inline-flex",
-                          alignItems: "center",
-                          gap: 2,
-                        }}
-                      >
-                        <WarningOutlined />
-                        时长不足
-                      </span>
-                    )}
-                  </span>
-                )}
+                ) : null}
                 <span>{isAiVoice(item) ? "按文本合成" : formatFileSize(getFileSize(item))}</span>
               </div>
             </div>
           )
         })}
       </div>
-
-      {/* 时长不足警告弹窗 */}
-      <Modal
-        title={
-          <span style={{ display: "flex", alignItems: "center", gap: 8 }}>
-            <WarningOutlined style={{ color: "#faad14" }} />
-            配音时长不足
-          </span>
-        }
-        open={durationWarningOpen}
-        onOk={handleConfirmUseAnyway}
-        onCancel={handleCancelSelection}
-        okText="仍要使用"
-        cancelText="重新选择"
-        okButtonProps={{ danger: true }}
-      >
-        {(() => {
-          const pendingMaterial = pendingVoiceId
-            ? materials.find((m) => m.id === pendingVoiceId)
-            : null
-          return (
-            <p>
-              该配音时长（
-              <strong>
-                {pendingMaterial ? formatDuration(getDuration(pendingMaterial)) : "--"}
-              </strong>
-              ）短于视频总时长（
-              <strong>{formatDuration(totalVideoDuration)}</strong>
-              ），播放时配音可能提前结束，建议选择更长的配音素材。
-            </p>
-          )
-        })()}
-      </Modal>
     </div>
   )
 }

apps/worker/video_processing/subtitle_generator.py

@@ -213,7 +213,7 @@ def generate_ass_from_timeline(
             t_shadow.get("offset_x", 2) if t_shadow.get("enabled", False) else 0,
             t_shadow.get("offset_y", 2) if t_shadow.get("enabled", False) else 0,
         )
-        t_alignment = position_to_ass_alignment(title_cfg.get("position", "top"))
+        t_alignment = position_to_ass_alignment(title_cfg.get("position", "bottom"))
 
         title_style_line = build_ass_style(
             "TitleStyle",

apps/worker/video_processing/unified_render_service.py

@@ -198,8 +198,13 @@ class UnifiedRenderService:
         # 2. 分组为 RenderLayers
         layers = self._group_clips_into_layers(resolved)
 
+        # 2.5 配音时长对齐：如果有配音素材，调整片段时长以匹配配音时长
+        voice_duration = self._get_voice_audio_duration()
+        if voice_duration > 0:
+            self._align_clips_to_voice_duration(layers, voice_duration)
+
         # 3. 计算视频总时长（用于字幕显示时长）
-        video_duration = self._estimate_total_duration(layers)
+        video_duration_final = self._estimate_total_duration(layers)
         # Debug: 输出各图层时长明细
         for layer in layers:
             layer_total = sum(UnifiedRenderService._clip_adjusted_duration(c) for c in layer.clips)
@@ -215,16 +220,16 @@ class UnifiedRenderService:
                 self.transition_duration,
                 ", ".join(clip_details),
             )
-        logger.info("[debug] estimated video_duration=%.3f", video_duration)
+        logger.info("[debug] estimated video_duration=%.3f", video_duration_final)
 
         # 3.5 TTS 配音生成（如果配置了）
-        self._maybe_add_voiceover_layer(layers, video_duration=video_duration)
+        self._maybe_add_voiceover_layer(layers, video_duration=video_duration_final)
 
         # 3.6 配音素材库音频（如果传入了本地路径）
-        self._maybe_add_voice_library_layer(layers, video_duration=video_duration)
+        self._maybe_add_voice_library_layer(layers, video_duration=video_duration_final)
 
         # 4. 生成 ASS 字幕文件（如果有 title/subtitle 配置）
-        ass_path = self._maybe_generate_ass(video_duration)
+        ass_path = self._maybe_generate_ass(video_duration_final)
 
         # 4.5 解析画中画配置
         pip_config = PiPConfig.from_dict((self.plan.config or {}).get("pip_config"))
@@ -257,7 +262,7 @@ class UnifiedRenderService:
             # 先尝试 stream copy 优化（无重编码，性能提升 10 倍+）
             # 条件不满足或失败时回退到带滤镜的直通渲染
             stream_copy_ok = self._try_render_stream_copy(
-                layers, output_path, ass_path=ass_path, video_duration=video_duration
+                layers, output_path, ass_path=ass_path, video_duration=video_duration_final
             )
             if stream_copy_ok:
                 used_stream_copy = True
@@ -271,7 +276,7 @@ class UnifiedRenderService:
                     layers,
                     output_path,
                     ass_path=ass_path,
-                    video_duration=video_duration,
+                    video_duration=video_duration_final,
                 )
         else:
             filter_complex, input_args = self._build_filter_complex(layers, ass_path=ass_path)
@@ -327,7 +332,7 @@ class UnifiedRenderService:
                         from video_processing.ffmpeg_utils import run_ffmpeg
 
                         run_ffmpeg(extract_cmd)
-                        final_audio = mix_bgm_with_main(ctx, main_audio_path, bgm_cfg, video_duration)
+                        final_audio = mix_bgm_with_main(ctx, main_audio_path, bgm_cfg, video_duration_final)
                         # 合并回视频
 
                         bgm_output = self.work_dir / f"rendered_{self.plan.id}_bgm.mp4"
@@ -353,7 +358,7 @@ class UnifiedRenderService:
             audio_path = mix_audio(
                 ctx,
                 layers,
-                video_duration,
+                video_duration_final,
                 bgm_path=self.bgm_path,
                 bgm_config=bgm_config,
                 audio_tracks_config=audio_tracks_config,
@@ -487,6 +492,147 @@ class UnifiedRenderService:
         """
         return _estimate_total_duration_pure(layers, self.transition_duration)
 
+    def _get_voice_audio_duration(self) -> float:
+        """获取配音音频文件的时长（秒）。
+
+        Returns:
+            配音音频时长，如果无配音或探测失败则返回 0.0
+        """
+        if not self.voiceover_audio_path:
+            return 0.0
+
+        audio_path = Path(self.voiceover_audio_path)
+        if not audio_path.exists() or audio_path.stat().st_size == 0:
+            return 0.0
+
+        try:
+            duration = probe_duration(audio_path)
+            logger.info("[voice-align] 配音音频时长: %.3fs path=%s", duration, self.voiceover_audio_path)
+            return duration
+        except Exception as e:
+            logger.warning("[voice-align] 探测配音音频时长失败: %s", e)
+            return 0.0
+
+    def _align_clips_to_voice_duration(
+        self,
+        layers: list[RenderLayer],
+        voice_duration: float,
+    ) -> None:
+        """调整片段时长以对齐配音时长。
+
+        核心逻辑：
+        - 计算片段总时长与配音时长的比例
+        - ±5% 以内不调整
+        - ratio < 1（片段比配音长）：按比例裁剪每段末尾
+        - ratio > 1（片段比配音短）：按比例慢放每段
+
+        Args:
+            layers: 渲染图层列表
+            voice_duration: 配音时长（秒）
+        """
+        if voice_duration <= 0:
+            return
+
+        # 只调整视频图层（main/broll/background），不调整音频图层
+        video_layers = [layer for layer in layers if layer.role in ("main", "broll", "background")]
+        if not video_layers:
+            return
+
+        # 计算所有视频图层的总时长
+        total_clips_duration = 0.0
+        for layer in video_layers:
+            for clip in layer.clips:
+                clip_dur = self._clip_adjusted_duration(clip)
+                total_clips_duration += clip_dur
+
+        if total_clips_duration <= 0:
+            return
+
+        ratio = voice_duration / total_clips_duration
+
+        # ±5% 以内不调整
+        if abs(ratio - 1.0) <= 0.05:
+            logger.info(
+                "[voice-align] 比例接近1:1，跳过调整: ratio=%.4f voice=%.3f clips=%.3f",
+                ratio,
+                voice_duration,
+                total_clips_duration,
+            )
+            return
+
+        logger.info(
+            "[voice-align] 开始调整片段时长: ratio=%.4f voice=%.3f clips=%.3f",
+            ratio,
+            voice_duration,
+            total_clips_duration,
+        )
+
+        # 收集所有视频 clip
+        all_clips: list[tuple[RenderLayer, ResolvedClip]] = []
+        for layer in video_layers:
+            for clip in layer.clips:
+                all_clips.append((layer, clip))
+
+        if not all_clips:
+            return
+
+        if ratio < 1.0:
+            # 片段比配音长，按比例裁剪每段末尾
+            # 减少每个 clip 的 duration
+            for _layer, clip in all_clips:
+                old_duration = clip.duration if clip.duration > 0 else clip.actual_duration
+                new_duration = old_duration * ratio
+
+                # 更新 duration
+                clip.duration = max(0.1, new_duration)  # 至少 0.1s
+
+                # 如果有 trim_config，也需要调整
+                if clip.trim_config is not None:
+                    new_trim_duration = clip.trim_config.duration * ratio
+                    clip.trim_config = TrimConfig(
+                        start_time=clip.trim_config.start_time,
+                        duration=max(0.1, new_trim_duration),
+                    )
+
+                logger.debug(
+                    "[voice-align] trim clip=%s: %.3f -> %.3f",
+                    clip.clip_id,
+                    old_duration,
+                    clip.duration,
+                )
+
+        else:
+            # ratio > 1.0: 片段比配音短，按比例慢放每段
+            # 降低 playback_speed
+            for _layer, clip in all_clips:
+                old_speed = clip.playback_speed if clip.playback_speed > 0 else 1.0
+                # speed = old_speed / ratio 会使视频变慢（ratio > 1 时）
+                new_speed = old_speed / ratio
+
+                # 下限 0.25x（避免过慢）
+                new_speed = max(0.25, round(new_speed, 4))
+                clip.playback_speed = new_speed
+
+                logger.debug(
+                    "[voice-align] slowdown clip=%s: speed %.4f -> %.4f",
+                    clip.clip_id,
+                    old_speed,
+                    new_speed,
+                )
+
+        # 调整后重新计算总时长用于日志
+        new_total = 0.0
+        for layer in video_layers:
+            for clip in layer.clips:
+                new_total += self._clip_adjusted_duration(clip)
+
+        logger.info(
+            "[voice-align] 调整完成: 新总时长=%.3fs (目标=%.3fs, 差异=%.3fs)",
+            new_total,
+            voice_duration,
+            abs(new_total - voice_duration),
+        )
+
     def _maybe_generate_ass(self, video_duration: float) -> Path | None:
         """根据 plan.config 生成 ASS 字幕文件。
 

packages/domain/ass_subtitle_builder.py

@@ -81,14 +81,14 @@ def position_to_ass_alignment(position: str) -> int:
         position: 位置字符串 top/center/bottom
 
     Returns:
-        ASS 对齐编号，默认 8（顶部居中）
+        ASS 对齐编号，默认 2（底部居中，与前端 DEFAULT_TITLE_SETTINGS.position="bottom" 对齐）
     """
     mapping = {
         "top": 8,
         "center": 5,
         "bottom": 2,
     }
-    return mapping.get(position, 8)
+    return mapping.get(position, 2)
 
 
 # ── Style 行构建 ──────────────────────────────────────────────────────────────
@@ -226,7 +226,6 @@ def _wrap_title_text(
 
     # 换行计算使用原始 font_size，与 CSS 预览一致；1.35x 补偿仅用于 ASS Fontsize 渲染
 
-
     # 先按已有 \N 分段，每段独立自动换行，最后用 \N 拼回
     segments = text.split("\\N")
     wrapped_segments: list[str] = []
@@ -386,8 +385,8 @@ def build_ass_content(
         # position → alignment 三档逻辑，现有输出保持一字节不变。
         title_pos = _parse_title_position(title_config, video_width, video_height)
 
-        title_alignment = 5 if title_pos is not None else position_to_ass_alignment(
-            title_config.get("position", "top")
+        title_alignment = (
+            5 if title_pos is not None else position_to_ass_alignment(title_config.get("position", "bottom"))
         )
 
         styles.append(

tests/unit/domain/test_ass_subtitle_builder.py

@@ -83,11 +83,11 @@ class TestPositionToAssAlignment:
     def test_bottom(self):
         assert position_to_ass_alignment("bottom") == 2
 
-    def test_unknown_defaults_top(self):
-        assert position_to_ass_alignment("unknown") == 8
+    def test_unknown_defaults_bottom(self):
+        assert position_to_ass_alignment("unknown") == 2
 
-    def test_empty_defaults_top(self):
-        assert position_to_ass_alignment("") == 8
+    def test_empty_defaults_bottom(self):
+        assert position_to_ass_alignment("") == 2
 
 
 # ============================================================
@@ -581,6 +581,7 @@ class TestConstants:
         assert isinstance(TITLE_MARGIN_BOTTOM, int)
         assert isinstance(TITLE_MARGIN_SIDE, int)
 
+
 # ============================================================
 # _wrap_title_text 换行逻辑验证
 # ============================================================

tests/unit/test_ass_subtitle_builder.py

@@ -65,11 +65,11 @@ class TestPositionToAssAlignment:
     def test_bottom(self):
         assert position_to_ass_alignment("bottom") == 2
 
-    def test_unknown_default_top(self):
-        assert position_to_ass_alignment("unknown") == 8
+    def test_unknown_default_bottom(self):
+        assert position_to_ass_alignment("unknown") == 2
 
-    def test_empty_default_top(self):
-        assert position_to_ass_alignment("") == 8
+    def test_empty_default_bottom(self):
+        assert position_to_ass_alignment("") == 2
 
 
 # ── Style 行构建 ─────────────────────────────────────────────────────────────
@@ -747,3 +747,45 @@ class TestTitleFreePosition:
             line for line in content.splitlines() if line.startswith("Dialogue:") and "SubtitleStyle" in line
         ][0]
         assert "\\pos(" not in sub_dialogue
+
+
+class TestDefaultPositionBottom:
+    """默认 position 应为 bottom（alignment=2），与前端 DEFAULT_TITLE_SETTINGS 对齐。"""
+
+    def _base_kwargs(self):
+        return dict(
+            video_width=1080,
+            video_height=1920,
+            video_duration=10.0,
+            title_text="测试标题",
+        )
+
+    def test_no_position_defaults_to_bottom_alignment(self):
+        """不传 position 时，Alignment 应为 2（bottom）。"""
+        content = build_ass_content(
+            **self._base_kwargs(),
+            title_config={"size": 36},
+        )
+        style_line = [line for line in content.splitlines() if line.startswith("Style: TitleStyle")][0]
+        fields = [f.strip() for f in style_line.split(",")]
+        assert fields[18] == "2", f"Expected alignment 2 (bottom), got {fields[18]}"
+
+    def test_no_position_no_coords_defaults_to_bottom(self):
+        """不传 position 也不传坐标时，走 bottom 三档逻辑。"""
+        content = build_ass_content(
+            **self._base_kwargs(),
+            title_config={},
+        )
+        style_line = [line for line in content.splitlines() if line.startswith("Style: TitleStyle")][0]
+        fields = [f.strip() for f in style_line.split(",")]
+        assert fields[18] == "2"
+
+    def test_explicit_top_still_works(self):
+        """显式传 position='top' 仍然得到 alignment=8。"""
+        content = build_ass_content(
+            **self._base_kwargs(),
+            title_config={"position": "top", "size": 36},
+        )
+        style_line = [line for line in content.splitlines() if line.startswith("Style: TitleStyle")][0]
+        fields = [f.strip() for f in style_line.split(",")]
+        assert fields[18] == "8"

tests/unit/test_render_subtitles.py

@@ -67,11 +67,11 @@ class TestPositionToAssAlignment:
     def test_bottom(self):
         assert _position_to_ass_alignment("bottom") == 2
 
-    def test_unknown_returns_top_default(self):
-        assert _position_to_ass_alignment("unknown") == 8
-        assert _position_to_ass_alignment("") == 8
-        assert _position_to_ass_alignment("left") == 8
-        assert _position_to_ass_alignment(None) == 8
+    def test_unknown_returns_bottom_default(self):
+        assert _position_to_ass_alignment("unknown") == 2
+        assert _position_to_ass_alignment("") == 2
+        assert _position_to_ass_alignment("left") == 2
+        assert _position_to_ass_alignment(None) == 2
 
 
 class TestBuildAssStyle:

tests/unit/test_render_subtitles_pure.py

@@ -66,15 +66,15 @@ class TestPositionToAssAlignment:
         """center → 居中(5)."""
         assert _position_to_ass_alignment("center") == 5
 
-    def test_unknown_defaults_to_top(self):
-        """未知位置默认顶部(8)."""
-        assert _position_to_ass_alignment("unknown") == 8
-        assert _position_to_ass_alignment("top_left") == 8
-        assert _position_to_ass_alignment("bottom_right") == 8
+    def test_unknown_defaults_to_bottom(self):
+        """未知位置默认底部(2)."""
+        assert _position_to_ass_alignment("unknown") == 2
+        assert _position_to_ass_alignment("top_left") == 2
+        assert _position_to_ass_alignment("bottom_right") == 2
 
-    def test_empty_string_defaults_to_top(self):
-        """空字符串默认顶部."""
-        assert _position_to_ass_alignment("") == 8
+    def test_empty_string_defaults_to_bottom(self):
+        """空字符串默认底部."""
+        assert _position_to_ass_alignment("") == 2
 
 
 class TestBuildAssStyle:

tests/unit/test_voice_duration_alignment.py

@@ -0,0 +1,205 @@
+"""Tests for voice duration alignment feature.
+
+Tests the _align_clips_to_voice_duration method in UnifiedRenderService.
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
+import pytest
+from video_processing.unified_render_service import RenderLayer, ResolvedClip, UnifiedRenderService
+
+
+class TestAlignClipsToVoiceDuration:
+    """Test clip duration alignment to voice audio."""
+
+    def _make_clip(
+        self,
+        clip_id: str,
+        duration: float,
+        actual_duration: float = 0.0,
+        playback_speed: float = 1.0,
+    ) -> ResolvedClip:
+        """Helper to create a ResolvedClip for testing."""
+        return ResolvedClip(
+            clip_id=clip_id,
+            asset_id=f"asset_{clip_id}",
+            local_path=Path(f"/tmp/{clip_id}.mp4"),
+            clip_type="main",
+            order=0,
+            duration=duration,
+            actual_duration=actual_duration or duration,
+            playback_speed=playback_speed,
+        )
+
+    def _make_layer(self, role: str, clips: list[ResolvedClip]) -> RenderLayer:
+        """Helper to create a RenderLayer for testing."""
+        return RenderLayer(role=role, clips=clips, z_index=0)
+
+    def _make_service(self, voiceover_path: str | None = None) -> UnifiedRenderService:
+        """Helper to create a mock UnifiedRenderService."""
+        plan = MagicMock()
+        plan.id = "test_plan"
+        plan.config = {}
+
+        with patch.object(UnifiedRenderService, "__init__", lambda self, **kwargs: None):
+            service = UnifiedRenderService.__new__(UnifiedRenderService)
+            service.plan = plan
+            service.voiceover_audio_path = voiceover_path
+            service.transition_duration = 0.0
+            return service
+
+    def test_no_voice_audio_no_adjustment(self):
+        """No voice audio → no adjustment."""
+        service = self._make_service(voiceover_path=None)
+        clips = [self._make_clip("c1", 10.0), self._make_clip("c2", 10.0)]
+        layers = [self._make_layer("main", clips)]
+
+        service._align_clips_to_voice_duration(layers, voice_duration=0.0)
+
+        # No change
+        assert clips[0].duration == 10.0
+        assert clips[1].duration == 10.0
+
+    def test_ratio_within_5_percent_no_adjustment(self):
+        """Ratio within ±5% → no adjustment."""
+        service = self._make_service()
+        clips = [self._make_clip("c1", 10.0)]
+        layers = [self._make_layer("main", clips)]
+
+        # Total clips = 10s, voice = 10.3s → ratio = 1.03 (within 5%)
+        service._align_clips_to_voice_duration(layers, voice_duration=10.3)
+
+        assert clips[0].duration == 10.0  # Unchanged
+
+    def test_ratio_less_than_1_trim_clips(self):
+        """Ratio < 1 (clips too long) → trim clips proportionally."""
+        service = self._make_service()
+        clips = [self._make_clip("c1", 10.0), self._make_clip("c2", 10.0)]
+        layers = [self._make_layer("main", clips)]
+
+        # Total clips = 20s, voice = 15s → ratio = 0.75
+        service._align_clips_to_voice_duration(layers, voice_duration=15.0)
+
+        # Each clip should be trimmed to 75%
+        assert abs(clips[0].duration - 7.5) < 0.01
+        assert abs(clips[1].duration - 7.5) < 0.01
+
+    def test_ratio_greater_than_1_slowdown_clips(self):
+        """Ratio > 1 (clips too short) → slow down clips."""
+        service = self._make_service()
+        clips = [self._make_clip("c1", 10.0), self._make_clip("c2", 10.0)]
+        layers = [self._make_layer("main", clips)]
+
+        # Total clips = 20s, voice = 25s → ratio = 1.25
+        service._align_clips_to_voice_duration(layers, voice_duration=25.0)
+
+        # Each clip's speed should be reduced: 1.0 / 1.25 = 0.8
+        assert abs(clips[0].playback_speed - 0.8) < 0.01
+        assert abs(clips[1].playback_speed - 0.8) < 0.01
+
+    def test_speed_lower_bound_025(self):
+        """Playback speed should not go below 0.25x."""
+        service = self._make_service()
+        clips = [self._make_clip("c1", 5.0)]
+        layers = [self._make_layer("main", clips)]
+
+        # Total clips = 5s, voice = 50s → ratio = 10.0
+        # Speed would be 1.0 / 10 = 0.1, but should be clamped to 0.25
+        service._align_clips_to_voice_duration(layers, voice_duration=50.0)
+
+        assert clips[0].playback_speed == 0.25
+
+    def test_only_video_layers_adjusted(self):
+        """Only main/broll/background layers are adjusted, not audio."""
+        service = self._make_service()
+
+        video_clips = [self._make_clip("v1", 10.0)]
+        audio_clips = [self._make_clip("a1", 10.0)]
+
+        layers = [
+            self._make_layer("main", video_clips),
+            self._make_layer("audio", audio_clips),
+        ]
+
+        # ratio = 0.5 → should trim video but not audio
+        service._align_clips_to_voice_duration(layers, voice_duration=5.0)
+
+        assert abs(video_clips[0].duration - 5.0) < 0.01  # Trimmed
+        assert audio_clips[0].duration == 10.0  # Unchanged
+
+    def test_multiple_video_layers_all_adjusted(self):
+        """All video layers (main, broll, background) are adjusted."""
+        service = self._make_service()
+
+        main_clips = [self._make_clip("m1", 10.0)]
+        broll_clips = [self._make_clip("b1", 10.0)]
+        bg_clips = [self._make_clip("bg1", 10.0)]
+
+        layers = [
+            self._make_layer("main", main_clips),
+            self._make_layer("broll", broll_clips),
+            self._make_layer("background", bg_clips),
+        ]
+
+        # Total video = 30s, voice = 15s → ratio = 0.5
+        service._align_clips_to_voice_duration(layers, voice_duration=15.0)
+
+        # All should be trimmed to 50%
+        assert abs(main_clips[0].duration - 5.0) < 0.01
+        assert abs(broll_clips[0].duration - 5.0) < 0.01
+        assert abs(bg_clips[0].duration - 5.0) < 0.01
+
+    def test_trim_config_also_adjusted(self):
+        """When clip has trim_config, it should also be adjusted."""
+        from video_processing.trim_engine import TrimConfig
+
+        service = self._make_service()
+
+        clip = self._make_clip("c1", 10.0)
+        clip.trim_config = TrimConfig(start_time=0.0, duration=10.0)
+
+        layers = [self._make_layer("main", [clip])]
+
+        # ratio = 0.5
+        service._align_clips_to_voice_duration(layers, voice_duration=5.0)
+
+        assert abs(clip.duration - 5.0) < 0.01
+        assert clip.trim_config is not None
+        assert abs(clip.trim_config.duration - 5.0) < 0.01
+
+
+class TestGetVoiceAudioDuration:
+    """Test voice audio duration probing."""
+
+    def test_no_voiceover_path_returns_zero(self):
+        """No voiceover path → return 0."""
+        with patch.object(UnifiedRenderService, "__init__", lambda self, **kwargs: None):
+            service = UnifiedRenderService.__new__(UnifiedRenderService)
+            service.voiceover_audio_path = None
+
+        assert service._get_voice_audio_duration() == 0.0
+
+    def test_nonexistent_file_returns_zero(self):
+        """Nonexistent file → return 0."""
+        with patch.object(UnifiedRenderService, "__init__", lambda self, **kwargs: None):
+            service = UnifiedRenderService.__new__(UnifiedRenderService)
+            service.voiceover_audio_path = "/nonexistent/path.mp3"
+
+        assert service._get_voice_audio_duration() == 0.0
+
+    @patch("video_processing.unified_render_service.probe_duration")
+    @patch("video_processing.unified_render_service.Path.exists", return_value=True)
+    @patch("video_processing.unified_render_service.Path.stat")
+    def test_probes_duration_from_file(self, mock_stat, mock_exists, mock_probe):
+        """Valid file → probe duration."""
+        mock_stat.return_value.st_size = 1000  # Non-empty file
+        mock_probe.return_value = 42.5
+
+        with patch.object(UnifiedRenderService, "__init__", lambda self, **kwargs: None):
+            service = UnifiedRenderService.__new__(UnifiedRenderService)
+            service.voiceover_audio_path = "/tmp/voice.mp3"
+
+        assert service._get_voice_audio_duration() == 42.5