chore: retry CI

chore: trigger CI rerun
fix: 修复上传音频弹窗 TS 类型错误和 prettier 格式问题
2026-08-07 15:57:31 +08:00 · 2026-08-07 15:54:17 +08:00 · 2026-08-07 15:50:43 +08:00 · 2026-08-07 15:50:43 +08:00 · 2026-08-07 15:50:32 +08:00 · 2026-08-07 15:16:44 +08:00
32 changed files with 1350 additions and 657 deletions
@@ -1845,27 +1845,6 @@
        "react": "^18 || ^19"
      }
    },
-    "node_modules/@testing-library/dom": {
-      "version": "10.4.1",
-      "resolved": "https://registry.npmjs.org/@testing-library/dom/-/dom-10.4.1.tgz",
-      "integrity": "sha512-o4PXJQidqJl82ckFaXUeoAW+XysPLauYI43Abki5hABd853iMhitooc6znOnczgbTYmEP6U6/y1ZyKAIsvMKGg==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true,
-      "dependencies": {
-        "@babel/code-frame": "^7.10.4",
-        "@babel/runtime": "^7.12.5",
-        "@types/aria-query": "^5.0.1",
-        "aria-query": "5.3.0",
-        "dom-accessibility-api": "^0.5.9",
-        "lz-string": "^1.5.0",
-        "picocolors": "1.1.1",
-        "pretty-format": "^27.0.2"
-      },
-      "engines": {
-        "node": ">=18"
-      }
-    },
    "node_modules/@testing-library/jest-dom": {
      "version": "6.9.1",
      "resolved": "https://registry.npmjs.org/@testing-library/jest-dom/-/jest-dom-6.9.1.tgz",
@@ -1935,14 +1914,6 @@
        "@testing-library/dom": ">=7.21.4"
      }
    },
-    "node_modules/@types/aria-query": {
-      "version": "5.0.4",
-      "resolved": "https://registry.npmjs.org/@types/aria-query/-/aria-query-5.0.4.tgz",
-      "integrity": "sha512-rfT93uj5s0PRL7EzccGMs3brplhcrghnDoV26NqKhCAS1hVo+WdNsPvE/yb6ilfr5hi2MEk6d5EWJTKdxg8jVw==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true
-    },
    "node_modules/@types/babel__core": {
      "version": "7.20.5",
      "resolved": "https://registry.npmjs.org/@types/babel__core/-/babel__core-7.20.5.tgz",
@@ -2009,14 +1980,14 @@
      "version": "15.7.15",
      "resolved": "https://registry.npmmirror.com/@types/prop-types/-/prop-types-15.7.15.tgz",
      "integrity": "sha512-F6bEyamV9jKGAFBEmlQnesRPGOQqS2+Uwi0Em15xenOxHaf2hv6L8YCVn3rPdPJOiJfPiCnLIRyvwVaqMY3MIw==",
-      "devOptional": true,
+      "dev": true,
      "license": "MIT"
    },
    "node_modules/@types/react": {
      "version": "18.3.31",
      "resolved": "https://registry.npmmirror.com/@types/react/-/react-18.3.31.tgz",
      "integrity": "sha512-vfEqpXTvwT91yhmwdfouStN2hSKwTvyRs8qpLfADyrq/kxDw0hZM7Wk9Ug1FELj8hIby+S/+kQCSRFF32nv2Qw==",
-      "devOptional": true,
+      "dev": true,
      "license": "MIT",
      "dependencies": {
        "@types/prop-types": "*",
@@ -3110,14 +3081,6 @@
        "node": ">=6.0.0"
      }
    },
-    "node_modules/dom-accessibility-api": {
-      "version": "0.5.16",
-      "resolved": "https://registry.npmjs.org/dom-accessibility-api/-/dom-accessibility-api-0.5.16.tgz",
-      "integrity": "sha512-X7BJ2yElsnOJ30pZF4uIIDfBEVgF4XEBxL9Bxhy6dnrm5hkzqmsWHGTiHqRiITNhMyFLyAiWndIJP7Z1NTteDg==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true
-    },
    "node_modules/dunder-proto": {
      "version": "1.0.1",
      "resolved": "https://registry.npmjs.org/dunder-proto/-/dunder-proto-1.0.1.tgz",
@@ -4028,18 +3991,6 @@
        "node": ">= 4"
      }
    },
-    "node_modules/immer": {
-      "version": "10.2.0",
-      "resolved": "https://registry.npmjs.org/immer/-/immer-10.2.0.tgz",
-      "integrity": "sha512-d/+XTN3zfODyjr89gM3mPq1WNX2B8pYsu7eORitdwyA2sBubnTl3laYlBk4sXY5FUa5qTZGBDPJICVbvqzjlbw==",
-      "license": "MIT",
-      "optional": true,
-      "peer": true,
-      "funding": {
-        "type": "opencollective",
-        "url": "https://opencollective.com/immer"
-      }
-    },
    "node_modules/import-fresh": {
      "version": "3.3.1",
      "resolved": "https://registry.npmjs.org/import-fresh/-/import-fresh-3.3.1.tgz",
@@ -4466,17 +4417,6 @@
        "yallist": "^3.0.2"
      }
    },
-    "node_modules/lz-string": {
-      "version": "1.5.0",
-      "resolved": "https://registry.npmjs.org/lz-string/-/lz-string-1.5.0.tgz",
-      "integrity": "sha512-h5bgJWpxJNswbU7qCrV0tIKQCaS3blPDrqKWx+QxzuzL1zGUzij9XCWLrSLsJPu5t+eWA/ycetzYAO5IOMcWAQ==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true,
-      "bin": {
-        "lz-string": "bin/bin.js"
-      }
-    },
    "node_modules/magic-string": {
      "version": "0.30.21",
      "resolved": "https://registry.npmjs.org/magic-string/-/magic-string-0.30.21.tgz",
@@ -5008,44 +4948,6 @@
        "url": "https://github.com/prettier/prettier?sponsor=1"
      }
    },
-    "node_modules/pretty-format": {
-      "version": "27.5.1",
-      "resolved": "https://registry.npmjs.org/pretty-format/-/pretty-format-27.5.1.tgz",
-      "integrity": "sha512-Qb1gy5OrP5+zDf2Bvnzdl3jsTf1qXVMazbvCoKhtKqVs4/YK4ozX4gKQJJVyNe+cajNPn0KoC0MC3FUmaHWEmQ==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true,
-      "dependencies": {
-        "ansi-regex": "^5.0.1",
-        "ansi-styles": "^5.0.0",
-        "react-is": "^17.0.1"
-      },
-      "engines": {
-        "node": "^10.13.0 || ^12.13.0 || ^14.15.0 || >=15.0.0"
-      }
-    },
-    "node_modules/pretty-format/node_modules/ansi-styles": {
-      "version": "5.2.0",
-      "resolved": "https://registry.npmjs.org/ansi-styles/-/ansi-styles-5.2.0.tgz",
-      "integrity": "sha512-Cxwpt2SfTzTtXcfOlzGEee8O+c+MmUgGrNiBcXnuWxuFJHe6a5Hz7qwhwe5OgaSYI0IJvkLqWX1ASG+cJOkEiA==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true,
-      "engines": {
-        "node": ">=10"
-      },
-      "funding": {
-        "url": "https://github.com/chalk/ansi-styles?sponsor=1"
-      }
-    },
-    "node_modules/pretty-format/node_modules/react-is": {
-      "version": "17.0.2",
-      "resolved": "https://registry.npmjs.org/react-is/-/react-is-17.0.2.tgz",
-      "integrity": "sha512-w2GsyukL62IJnlaff/nRegPQR94C/XXamvMWmSHRJ4y7Ts/4ocGRmTHvOs8PSE6pB3dWOrD/nueuU5sduBsQ4w==",
-      "dev": true,
-      "license": "MIT",
-      "peer": true
-    },
    "node_modules/proxy-from-env": {
      "version": "2.1.0",
      "resolved": "https://registry.npmjs.org/proxy-from-env/-/proxy-from-env-2.1.0.tgz",
@@ -60,7 +60,6 @@ export const useTtsPanel = ({ open, config, onChange, onClose }: UseTtsPanelOpti
    return () => {
      active = false
    }
-    // eslint-disable-next-line react-hooks/exhaustive-deps
  }, [open])

  /* ── 组件卸载时清理音频资源 ── */
@@ -224,21 +224,23 @@ const GeneratePage: React.FC = () => {
            />
          )}

-          {/* 正式生成结果 */}
-          <GenerateResultPanel
-            generated={generated}
-            generating={generating}
-            progress={progress}
-            generateError={generateError}
-            generatedVideos={generatedVideos}
-            onVideoPreview={(video) => {
-              setPreviewVideo(video)
-              setPreviewModalOpen(true)
-            }}
-            onDownload={handleDownload}
-            onShare={handleShare}
-            onGoToLibrary={() => navigate("/app/products")}
-          />
+          {/* 正式生成结果（Step4+ 才显示） */}
+          {currentStep >= 4 && (
+            <GenerateResultPanel
+              generated={generated}
+              generating={generating}
+              progress={progress}
+              generateError={generateError}
+              generatedVideos={generatedVideos}
+              onVideoPreview={(video) => {
+                setPreviewVideo(video)
+                setPreviewModalOpen(true)
+              }}
+              onDownload={handleDownload}
+              onShare={handleShare}
+              onGoToLibrary={() => navigate("/app/products")}
+            />
+          )}
        </div>
      </div>

@@ -94,14 +94,8 @@ export const GenerateStepContent: React.FC<GenerateStepContentProps> = (props) =
    selectedVoice,
    onSelectedVoiceChange,
    voiceMode,
-    onVoiceModeChange,
    selectedClonedVoice,
-    onSelectedClonedVoiceChange,
    clonedVoices,
-    addClone,
-    hasProcessing,
-    cloneModalOpen,
-    onCloneModalOpenChange,
    generateCount,
    onGenerateCountChange,
    generating,
@@ -170,16 +164,6 @@ export const GenerateStepContent: React.FC<GenerateStepContentProps> = (props) =
        <Step5VoiceSelect
          selectedVoice={selectedVoice}
          onSelectedVoiceChange={onSelectedVoiceChange}
-          voiceMode={voiceMode}
-          onVoiceModeChange={onVoiceModeChange}
-          selectedClonedVoice={selectedClonedVoice}
-          onSelectedClonedVoiceChange={onSelectedClonedVoiceChange}
-          clonedVoices={clonedVoices}
-          addClone={addClone}
-          hasProcessing={hasProcessing}
-          cloneModalOpen={cloneModalOpen}
-          onCloneModalOpenChange={onCloneModalOpenChange}
-          titleText={titleSettings.title}
        />
      )
    case 6:
@@ -1,9 +1,17 @@
 /**
 * 右侧预览视频面板
 * Step3 生成预览后常驻显示预览视频
- * Step4+ 显示标题实时预览
+ * Step4+ 用 Canvas 绘制标题预览（替代 CSS overlay，与 ASS 渲染行为一致）
+ *
+ * 设计说明：标题预览仅在有视频时显示（叠加在视频画面上方）。
+ * 无视频状态（idle/loading/error）下不再单独显示标题预览，这是有意为之的设计简化。
+ *
+ * Canvas 居中修复说明：
+ * Canvas 的 CSS 位置和尺寸直接匹配视频实际渲染区域（通过 getBoundingClientRect），
+ * 绘制坐标系基于 Canvas 自身尺寸，x = w/2 即可实现水平居中，
+ * 避免容器与视频尺寸不一致时浏览器拉伸 Canvas 导致居中偏移。
 */
-import React from "react"
+import React, { useRef, useEffect, useCallback } from "react"
 import { PlayCircleOutlined, LoadingOutlined } from "@ant-design/icons"
 import type { PreviewResult, PreviewStatus } from "../hooks/useStep3Preview"
 import type { TitleSettings } from "../types"
@@ -21,40 +29,131 @@ interface PreviewVideoPanelProps {
  titleSettings?: TitleSettings
 }

-/** 根据 position 返回 absolute 定位样式（叠加在视频上） */
-const positionOverlayStyle = (position: string): React.CSSProperties => {
+/* ── Canvas 绘制工具函数 ── */
+
+/**
+ * 将文本按 maxWidth 逐字换行，返回行数组。
+ * 与 ASS 字幕引擎的逐字换行行为一致。
+ */
+function wrapText(ctx: CanvasRenderingContext2D, text: string, maxWidth: number): string[] {
+  const lines: string[] = []
+  let currentLine = ""
+  for (const char of text) {
+    const testLine = currentLine + char
+    if (ctx.measureText(testLine).width > maxWidth && currentLine) {
+      lines.push(currentLine)
+      currentLine = char
+    } else {
+      currentLine = testLine
+    }
+  }
+  if (currentLine) lines.push(currentLine)
+  return lines
+}
+
+/**
+ * 在 canvas 上绘制标题文字（含描边/阴影/多行居中）
+ *
+ * Canvas 已通过 CSS 定位到视频实际渲染位置，
+ * 坐标系基于 Canvas 自身尺寸，居中直接使用 w/2。
+ *
+ * @param ctx canvas 上下文
+ * @param w canvas CSS 宽度（= 视频渲染宽度）
+ * @param h canvas CSS 高度（= 视频渲染高度）
+ * @param text 标题文字
+ * @param settings 标题样式
+ * @param paddingX 左右边距（px），与 ASS 的 MarginL/MarginR 对应
+ * @param position "top" | "center" | "bottom"
+ * @param topOffset 顶部/底部偏移量
+ */
+function drawTitleOnCanvas(
+  ctx: CanvasRenderingContext2D,
+  w: number,
+  h: number,
+  text: string,
+  settings: TitleSettings,
+  paddingX: number,
+  position: string,
+  topOffset: number,
+) {
+  const dpr = window.devicePixelRatio || 1
+
+  // 设置 canvas 物理像素尺寸（高清屏适配）
+  ctx.canvas.width = Math.round(w * dpr)
+  ctx.canvas.height = Math.round(h * dpr)
+  ctx.scale(dpr, dpr)
+
+  // 清除
+  ctx.clearRect(0, 0, w, h)
+
+  // 可用宽度 = 总宽 - 左右边距
+  const availableWidth = w - paddingX * 2
+  if (availableWidth <= 0) return
+
+  // 字体设置
+  const fontSize = Math.round(Math.min(settings.size, 36))
+  const fontWeight = settings.bold ? "bold" : "normal"
+  const fontStyle = settings.italic ? "italic" : "normal"
+  ctx.font = `${fontStyle} ${fontWeight} ${fontSize}px "${settings.font}"`
+
+  // 文字属性
+  ctx.textAlign = "center"
+  ctx.textBaseline = "middle"
+  ctx.fillStyle = settings.color
+
+  const lineHeight = fontSize * 1.4
+
+  // 描边 & 阴影
+  if (settings.stroke) {
+    ctx.strokeStyle = "rgba(0,0,0,0.6)"
+    ctx.lineWidth = 2
+    ctx.lineJoin = "round"
+  }
+  if (settings.shadow) {
+    ctx.shadowColor = "rgba(0,0,0,0.7)"
+    ctx.shadowBlur = 4
+    ctx.shadowOffsetX = 2
+    ctx.shadowOffsetY = 2
+  }
+
+  // 换行
+  const displayText = text && text.trim() ? text : "请选择或输入标题"
+  const lines = wrapText(ctx, displayText, availableWidth)
+
+  // 起始 Y：根据 position 计算
+  const totalTextHeight = lines.length * lineHeight
+  let startY: number
  switch (position) {
    case "top":
-      return { top: 16, left: 0, right: 0 }
+      startY = topOffset
+      break
    case "center":
-      return { top: "50%", left: 0, right: 0, transform: "translateY(-50%)" }
+      startY = (h - totalTextHeight) / 2 + lineHeight / 2
+      break
    case "bottom":
    default:
-      return { bottom: 56, left: 0, right: 0 } // 56px = 给视频控制条留空间
+      startY = h - topOffset - totalTextHeight + lineHeight / 2
+      break
+  }
+
+  // 居中 x = w/2（Canvas 已定位到视频位置，无需额外偏移）
+  const x = w / 2
+  lines.forEach((line, i) => {
+    const y = startY + i * lineHeight
+    if (settings.stroke) ctx.strokeText(line, x, y)
+    ctx.fillText(line, x, y)
+  })
+
+  // 重置 shadow（避免影响后续绘制）
+  if (settings.shadow) {
+    ctx.shadowColor = "transparent"
+    ctx.shadowBlur = 0
+    ctx.shadowOffsetX = 0
+    ctx.shadowOffsetY = 0
  }
 }

-/** 根据 TitleSettings 生成文字样式 */
-const buildTitleStyle = (settings: TitleSettings): React.CSSProperties => {
-  const style: React.CSSProperties = {
-    fontFamily: settings.font,
-    fontSize: Math.min(settings.size, 36), // 预览区域限制最大字号
-    color: settings.color,
-    fontWeight: settings.bold ? 700 : 400,
-    fontStyle: settings.italic ? "italic" : "normal",
-    textAlign: "center",
-    lineHeight: 1.4,
-    wordBreak: "break-word",
-    padding: "0 12px",
-  }
-  if (settings.stroke) {
-    style.WebkitTextStroke = "1px rgba(0,0,0,0.6)"
-  }
-  if (settings.shadow) {
-    style.textShadow = "2px 2px 4px rgba(0,0,0,0.7)"
-  }
-  return style
-}
+/* ── 组件 ── */

 export const PreviewVideoPanel: React.FC<PreviewVideoPanelProps> = ({
  previewStatus,
@@ -71,6 +170,125 @@ export const PreviewVideoPanel: React.FC<PreviewVideoPanelProps> = ({
  const isError = previewStatus === "error"
  const showTitlePreview = !!titleSettings

+  // video 模式 refs
+  const canvasRef = useRef<HTMLCanvasElement>(null)
+  const containerRef = useRef<HTMLDivElement>(null)
+  const videoRef = useRef<HTMLVideoElement>(null)
+
+  // 字体加载状态（ref 供 draw 回调同步读取，无需 state 避免触发不必要的重渲染）
+  const fontLoadedRef = useRef(false)
+
+  /** 在 video canvas 上绘制标题 */
+  const drawVideoTitle = useCallback(() => {
+    // 通过 ref 读取字体状态，避免 fontLoaded 进入依赖数组
+    if (!fontLoadedRef.current) return
+    const canvas = canvasRef.current
+    const container = containerRef.current
+    if (!canvas || !container || !titleSettings) return
+    const ctx = canvas.getContext("2d")
+    if (!ctx) return
+
+    const containerRect = container.getBoundingClientRect()
+    if (containerRect.width <= 0 || containerRect.height <= 0) return
+
+    // 使用 video 元素的 getBoundingClientRect 获取实际渲染尺寸和位置
+    const videoEl = videoRef.current
+    let drawW = containerRect.width
+    let drawH = containerRect.height
+    let offsetX = 0
+    let offsetY = 0
+
+    if (videoEl && videoEl.clientWidth > 0 && videoEl.clientHeight > 0) {
+      const videoRect = videoEl.getBoundingClientRect()
+      drawW = videoRect.width
+      drawH = videoRect.height
+      offsetX = videoRect.left - containerRect.left
+      offsetY = videoRect.top - containerRect.top
+    }
+
+    // 更新 Canvas CSS 位置和尺寸，使其与视频实际渲染区域完全对齐
+    canvas.style.left = `${offsetX}px`
+    canvas.style.top = `${offsetY}px`
+    canvas.style.width = `${drawW}px`
+    canvas.style.height = `${drawH}px`
+
+    // 绘制时，坐标系基于 Canvas 自身尺寸，无需额外偏移
+    drawTitleOnCanvas(
+      ctx,
+      drawW,
+      drawH,
+      titleText || "",
+      titleSettings,
+      40,
+      titleSettings.position,
+      60,
+    )
+  }, [titleText, titleSettings])
+
+  // video 模式：ResizeObserver 监听容器尺寸变化 → 重绘
+  useEffect(() => {
+    if (!showTitlePreview || !hasPreview) return
+    const container = containerRef.current
+    if (!container) return
+
+    const observer = new ResizeObserver(() => {
+      drawVideoTitle()
+    })
+    observer.observe(container)
+    requestAnimationFrame(drawVideoTitle)
+
+    return () => observer.disconnect()
+  }, [showTitlePreview, hasPreview, drawVideoTitle])
+
+  // 字体加载检测：字体变更时重新检测，确保 measureText 使用正确字体
+  useEffect(() => {
+    if (!showTitlePreview || !titleSettings) {
+      fontLoadedRef.current = false
+      return
+    }
+    let cancelled = false
+    fontLoadedRef.current = false
+
+    const fontWeight = titleSettings.bold ? "bold" : ""
+    const fontStyle = titleSettings.italic ? "italic" : ""
+    const fontSpec =
+      `${fontStyle} ${fontWeight} ${titleSettings.size}px "${titleSettings.font}"`.trim()
+
+    const onFontReady = () => {
+      if (cancelled) return
+      fontLoadedRef.current = true
+      // ref 已同步更新，显式触发重绘（draw 内部通过 ref 检查字体状态）
+      requestAnimationFrame(() => {
+        if (!cancelled) {
+          drawVideoTitle()
+        }
+      })
+    }
+
+    if (document.fonts.check(fontSpec)) {
+      onFontReady()
+      return
+    }
+
+    document.fonts
+      .load(fontSpec)
+      .then(() => onFontReady())
+      .catch(() => {
+        document.fonts.ready.then(() => onFontReady())
+      })
+
+    return () => {
+      cancelled = true
+    }
+  }, [showTitlePreview, titleSettings, drawVideoTitle])
+
+  // video 加载完成后重绘
+  const handleVideoLoaded = useCallback(() => {
+    if (showTitlePreview) {
+      requestAnimationFrame(drawVideoTitle)
+    }
+  }, [showTitlePreview, drawVideoTitle])
+
  return (
    <div className="xx-generate-preview">
      <div className="xx-preview-header">
@@ -121,50 +339,35 @@ export const PreviewVideoPanel: React.FC<PreviewVideoPanelProps> = ({
        </div>
      )}

-      {/* 预览成功 + 标题叠加 */}
+      {/* 预览成功 + Canvas 标题叠加 */}
      {hasPreview && (
-        <div style={{ position: "relative" }}>
+        <div ref={containerRef} style={{ position: "relative" }}>
          <div className="xx-preview-video">
-            <video src={previewResult.videoUrl} controls preload="metadata" />
+            <video
+              ref={videoRef}
+              src={previewResult.videoUrl}
+              controls
+              preload="metadata"
+              onLoadedMetadata={handleVideoLoaded}
+            />
          </div>
-          {/* 标题叠加在视频画面上 */}
          {showTitlePreview && (
-            <div
+            <canvas
+              ref={canvasRef}
              style={{
                position: "absolute",
-                ...positionOverlayStyle(titleSettings!.position),
+                top: 0,
+                left: 0,
                width: "100%",
+                height: "100%",
                zIndex: 1,
                pointerEvents: "none",
              }}
-            >
-              <span style={buildTitleStyle(titleSettings!)}>
-                {titleText && titleText.trim() ? titleText : "请选择或输入标题"}
-              </span>
-            </div>
+            />
          )}
        </div>
      )}

-      {/* 无视频时的标题预览（idle/loading/error 状态下仍然显示标题预览） */}
-      {!hasPreview && showTitlePreview && (
-        <div
-          style={{
-            background: "rgba(0,0,0,0.3)",
-            borderRadius: 8,
-            minHeight: 64,
-            marginTop: 12,
-            display: "flex",
-            alignItems: "center",
-            justifyContent: "center",
-          }}
-        >
-          <span style={buildTitleStyle(titleSettings!)}>
-            {titleText && titleText.trim() ? titleText : "请选择或输入标题"}
-          </span>
-        </div>
-      )}
-
      {/* 预览信息 */}
      {hasPreview && previewResult && (
        <div className="xx-preview-info">
@@ -27,9 +27,9 @@ interface Step3GeneratePreviewProps {
 }

 const Step3GeneratePreview: React.FC<Step3GeneratePreviewProps> = ({
-  templateName,
-  materialCount,
-  videoRatio,
+  templateName: _templateName,
+  materialCount: _materialCount,
+  videoRatio: _videoRatio,
  previewStatus,
  previewResult,
  previewError,
@@ -119,48 +119,6 @@ const Step3GeneratePreview: React.FC<Step3GeneratePreviewProps> = ({
              preload="metadata"
            />
          </div>
-
-          {/* 剪辑方案信息 */}
-          <div className="xx-preview-plan-card">
-            <div className="xx-preview-plan-title">剪辑方案信息</div>
-            <div className="xx-preview-plan-info">
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">模板</span>
-                <span className="xx-preview-plan-value">{templateName}</span>
-              </div>
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">素材数量</span>
-                <span className="xx-preview-plan-value">{materialCount}</span>
-              </div>
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">片段数</span>
-                <span className="xx-preview-plan-value">{previewResult.clipCount} 段</span>
-              </div>
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">转场次数</span>
-                <span className="xx-preview-plan-value">{previewResult.transitionCount} 次</span>
-              </div>
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">素材使用率</span>
-                <span className="xx-preview-plan-value">{previewResult.materialUsage}%</span>
-              </div>
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">实际时长</span>
-                <span className="xx-preview-plan-value">
-                  {(typeof previewResult.duration === "number"
-                    ? previewResult.duration
-                    : 0
-                  ).toFixed(1)}{" "}
-                  秒
-                </span>
-              </div>
-              <div className="xx-preview-plan-row">
-                <span className="xx-preview-plan-label">视频比例</span>
-                <span className="xx-preview-plan-value">{videoRatio}</span>
-              </div>
-            </div>
-            <div className="xx-preview-plan-hint">💡 这是 480p 预览版，正式生成将输出高清视频</div>
-          </div>
        </>
      )}
    </div>
@@ -2,7 +2,7 @@
 * Step 4 标题设置组件
 */
 import React from "react"
-import { Select } from "antd"
+import { AutoComplete } from "antd"
 import { POSITION_OPTIONS, FONT_OPTIONS } from "../constants"
 import type { TitleSettings } from "../types"
 import { useStep4Title } from "../hooks/useStep4Title"
@@ -75,10 +75,6 @@ const Step4TitleSettings: React.FC<Step4TitleSettingsProps> = (props) => {
            onRefresh={t.handleRefreshAiTitles}
          />

-          <div className="xx-divider">
-            <span>或手动选择</span>
-          </div>
-
          <div className="xx-title-ai-toggle">
            <span className="xx-toggle-label">AI 自动选择标题</span>
            <div className="xx-switch" onClick={t.toggleAiAutoSelect}>
@@ -87,11 +83,11 @@ const Step4TitleSettings: React.FC<Step4TitleSettingsProps> = (props) => {
          </div>

          <div className="xx-form-field">
-            <label>从标题库选择</label>
-            <Select
-              placeholder="请选择标题…"
+            <label>标题</label>
+            <AutoComplete
+              placeholder="输入或从标题库选择…"
              allowClear
-              showSearch
+              maxLength={50}
              style={{ width: "100%" }}
              value={t.titleSettings.title || undefined}
              onChange={(val) => t.updateTitle(val || "")}
@@ -99,9 +95,10 @@ const Step4TitleSettings: React.FC<Step4TitleSettingsProps> = (props) => {
                label: ut.content,
                value: ut.content,
              }))}
-              filterOption={(input, option) =>
-                ((option?.label as string) || "").toLowerCase().includes(input.toLowerCase())
-              }
+              filterOption={(inputValue, option) => {
+                const title = (option?.label || option?.value || "") as string
+                return title.toLowerCase().includes((inputValue || "").toLowerCase())
+              }}
              notFoundContent={
                t.userTitles.length === 0 ? (
                  <span style={{ color: "var(--text-tertiary)", fontSize: 13 }}>
@@ -111,15 +108,6 @@ const Step4TitleSettings: React.FC<Step4TitleSettingsProps> = (props) => {
              }
            />
          </div>
-          <div className="xx-form-field" style={{ marginTop: 14 }}>
-            <label>或手动输入</label>
-            <input
-              placeholder="输入自定义标题…"
-              value={t.titleSettings.title}
-              onChange={(e) => t.updateTitle(e.target.value)}
-              maxLength={50}
-            />
-          </div>
        </>
      )}

@@ -1,137 +1,250 @@
 /**
 * Step 5 配音选择组件
+ * 展示用户已上传的配音素材，支持选中、预览播放
 */
-import React from "react"
-import type { VoiceClone } from "@/api/voice-clone"
-import { useStep5Voice } from "../hooks/useStep5Voice"
-import VoiceRecommendSection from "./voice/VoiceRecommendSection"
-import VoiceChoiceCard from "./voice/VoiceChoiceCard"
-import PresetVoiceDetail from "./voice/PresetVoiceDetail"
-import CustomVoicePanel from "./voice/CustomVoicePanel"
-import SaveVoiceModal from "./voice/SaveVoiceModal"
-import CloneVoiceSection from "./voice/CloneVoiceSection"
+import React, { useState, useRef, useCallback } from "react"
+import { useQuery } from "@tanstack/react-query"
+import { AudioOutlined, SoundOutlined } from "@ant-design/icons"
+import { getAssetsByKind } from "@/api/assets"
+import type { AssetItem } from "@/api/assets"

 interface Step5VoiceSelectProps {
  selectedVoice: string
-  onSelectedVoiceChange: (voiceId: string) => void
-  voiceMode: "preset" | "custom" | "clone"
-  onVoiceModeChange: (mode: "preset" | "custom" | "clone") => void
-  selectedClonedVoice: string
-  onSelectedClonedVoiceChange: (voiceId: string) => void
-  clonedVoices: VoiceClone[]
-  addClone: (voice: VoiceClone) => void
-  hasProcessing: boolean
-  cloneModalOpen: boolean
-  onCloneModalOpenChange: (open: boolean) => void
-  titleText: string
+  onSelectedVoiceChange: (id: string) => void
 }

-const Step5VoiceSelect: React.FC<Step5VoiceSelectProps> = (props) => {
-  const v = useStep5Voice(props)
+/** 格式化时长 mm:ss */
+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"
+  if (bytes < 1024) return `${bytes} B`
+  if (bytes < 1024 * 1024) return `${(bytes / 1024).toFixed(1)} KB`
+  if (bytes < 1024 * 1024 * 1024) return `${(bytes / (1024 * 1024)).toFixed(1)} MB`
+  return `${(bytes / (1024 * 1024 * 1024)).toFixed(1)} GB`
+}
+
+const Step5VoiceSelect: React.FC<Step5VoiceSelectProps> = ({
+  selectedVoice,
+  onSelectedVoiceChange,
+}) => {
+  const [playingId, setPlayingId] = useState<string | null>(null)
+  const audioRef = useRef<HTMLAudioElement | null>(null)
+
+  // 获取用户上传的配音素材
+  const { data: materials = [], isLoading } = useQuery({
+    queryKey: ["assets", "voice"],
+    queryFn: () => getAssetsByKind("voice", { limit: 50 }),
+  })
+
+  /** 切换播放/暂停 */
+  const togglePlay = useCallback(
+    (material: AssetItem) => {
+      // 如果当前正在播放同一个素材，则暂停
+      if (playingId === material.id && audioRef.current) {
+        audioRef.current.pause()
+        setPlayingId(null)
+        return
+      }
+
+      // 停止之前的播放
+      if (audioRef.current) {
+        audioRef.current.pause()
+      }
+
+      // 创建新的 audio 元素播放
+      const audio = new Audio(material.file_url)
+      audioRef.current = audio
+      setPlayingId(material.id)
+
+      audio.onended = () => {
+        setPlayingId(null)
+        audioRef.current = null
+      }
+
+      audio.play().catch(() => {
+        setPlayingId(null)
+        audioRef.current = null
+      })
+    },
+    [playingId],
+  )
+
+  /** 选中素材 */
+  const handleSelect = useCallback(
+    (id: string) => {
+      onSelectedVoiceChange(id)
+    },
+    [onSelectedVoiceChange],
+  )
+
+  /** 跳转到配音库上传 */
+  const handleGoToUpload = useCallback(() => {
+    window.location.href = "/voices"
+  }, [])
+
+  // 加载中状态
+  if (isLoading) {
+    return (
+      <div className="xx-form-section">
+        <h3>🎙️ 选择配音</h3>
+        <div style={{ textAlign: "center", padding: "40px 0", color: "#999" }}>
+          加载配音素材中...
+        </div>
+      </div>
+    )
+  }
+
+  // 空状态
+  if (materials.length === 0) {
+    return (
+      <div className="xx-form-section">
+        <h3>🎙️ 选择配音</h3>
+        <div
+          style={{
+            textAlign: "center",
+            padding: "60px 0",
+            color: "#999",
+          }}
+        >
+          <AudioOutlined style={{ fontSize: 48, color: "#d9d9d9", marginBottom: 16 }} />
+          <p style={{ fontSize: 16, marginBottom: 16 }}>暂无配音素材</p>
+          <button
+            type="button"
+            onClick={handleGoToUpload}
+            style={{
+              padding: "8px 24px",
+              background: "#1677ff",
+              color: "#fff",
+              border: "none",
+              borderRadius: 6,
+              cursor: "pointer",
+              fontSize: 14,
+            }}
+          >
+            去配音库上传
+          </button>
+        </div>
+      </div>
+    )
+  }

  return (
    <div className="xx-form-section">
      <h3>🎙️ 选择配音</h3>
+      <p style={{ color: "#666", marginBottom: 16, fontSize: 14 }}>
+        从配音库中选择已上传的素材，点击卡片可预览播放
+      </p>
+      <div
+        style={{
+          display: "grid",
+          gridTemplateColumns: "repeat(auto-fill, minmax(220px, 1fr))",
+          gap: 12,
+        }}
+      >
+        {materials.map((item) => {
+          const isSelected = selectedVoice === item.id
+          const isPlaying = playingId === item.id

-      <VoiceRecommendSection
-        presetVoices={v.presetVoices}
-        voiceRecommendLoading={v.voiceRecommendLoading}
-        voiceRecommendations={v.voiceRecommendations}
-        hasVoiceRecommend={v.hasVoiceRecommend}
-        onRecommend={v.handleVoiceRecommend}
-        onSelectVoice={v.handleSelectRecommendedVoice}
-        selectedVoiceId={v.selectedVoice}
-        voiceMode={v.voiceMode}
-        VOICE_GENDER_ICON={v.VOICE_GENDER_ICON}
-      />
+          return (
+            <div
+              key={item.id}
+              onClick={() => handleSelect(item.id)}
+              style={{
+                padding: 16,
+                borderRadius: 8,
+                border: isSelected ? "2px solid #1677ff" : "1px solid #e8e8e8",
+                background: isSelected ? "#e6f4ff" : "#fff",
+                cursor: "pointer",
+                transition: "all 0.2s",
+                display: "flex",
+                flexDirection: "column",
+                gap: 8,
+              }}
+            >
+              {/* 顶部：图标 + 播放按钮 */}
+              <div
+                style={{ display: "flex", alignItems: "center", justifyContent: "space-between" }}
+              >
+                <div
+                  style={{
+                    width: 40,
+                    height: 40,
+                    borderRadius: 8,
+                    background: isSelected
+                      ? "linear-gradient(135deg, #1677ff, #4096ff)"
+                      : "linear-gradient(135deg, #f0f0f0, #e8e8e8)",
+                    display: "flex",
+                    alignItems: "center",
+                    justifyContent: "center",
+                  }}
+                >
+                  <AudioOutlined style={{ fontSize: 20, color: isSelected ? "#fff" : "#666" }} />
+                </div>
+                {item.file_url && (
+                  <button
+                    type="button"
+                    onClick={(e) => {
+                      e.stopPropagation()
+                      togglePlay(item)
+                    }}
+                    style={{
+                      width: 32,
+                      height: 32,
+                      borderRadius: "50%",
+                      border: "none",
+                      background: isPlaying ? "#ff4d4f" : "#1677ff",
+                      color: "#fff",
+                      cursor: "pointer",
+                      display: "flex",
+                      alignItems: "center",
+                      justifyContent: "center",
+                      fontSize: 14,
+                    }}
+                    title={isPlaying ? "暂停" : "播放"}
+                  >
+                    <SoundOutlined />
+                  </button>
+                )}
+              </div>

-      <div className="xx-divider">
-        <span>全部音色</span>
+              {/* 名称 */}
+              <div
+                style={{
+                  fontSize: 14,
+                  fontWeight: 500,
+                  color: "#333",
+                  overflow: "hidden",
+                  textOverflow: "ellipsis",
+                  whiteSpace: "nowrap",
+                }}
+                title={item.name}
+              >
+                {item.name}
+              </div>
+
+              {/* 时长 + 大小 */}
+              <div
+                style={{
+                  display: "flex",
+                  justifyContent: "space-between",
+                  fontSize: 12,
+                  color: "#999",
+                }}
+              >
+                <span>{formatDuration(item.duration)}</span>
+                <span>{formatFileSize(item.file_size)}</span>
+              </div>
+            </div>
+          )
+        })}
      </div>
-
-      <div className="xx-voice-choice-list" style={{ marginBottom: 16 }}>
-        {v.presetVoices.slice(0, 3).map((pv) => (
-          <VoiceChoiceCard
-            key={pv.voice_id}
-            selected={v.voiceMode === "preset" && v.selectedVoice === pv.voice_id}
-            onClick={() => v.handleSelectPresetVoice(pv.voice_id)}
-            avatar={v.VOICE_GENDER_ICON[pv.gender] ?? "✨"}
-            title={pv.name}
-            description={pv.description}
-          />
-        ))}
-        {v.presetVoicesLoading && (
-          <VoiceChoiceCard
-            selected={false}
-            onClick={() => {}}
-            avatar="⏳"
-            title="加载中…"
-            loading
-          />
-        )}
-        <VoiceChoiceCard
-          selected={v.voiceMode === "clone"}
-          onClick={v.handleSelectCloneVoice}
-          avatar="🎤"
-          title="克隆我的声音"
-          description="上传语音样本克隆"
-          avatarStyle={{ background: "linear-gradient(135deg, #10b981, #059669)" }}
-        />
-      </div>
-
-      {v.voiceMode === "preset" && (
-        <PresetVoiceDetail
-          presetVoices={v.presetVoices}
-          selectedVoice={v.selectedVoice}
-          onSelect={v.handleSelectPresetVoice}
-          playingVoice={v.playingVoice}
-          onTogglePlay={v.toggleVoicePlay}
-          presetVoicesLoading={v.presetVoicesLoading}
-        />
-      )}
-
-      {v.voiceMode === "custom" && (
-        <>
-          <CustomVoicePanel
-            customVoiceText={v.customVoiceText}
-            onTextChange={v.setCustomVoiceText}
-            synthesizePending={v.synthesizeMutation.isPending}
-            onSynthesize={v.handleSynthesizeVoice}
-            ttsError={v.ttsError}
-            customAudioUrl={v.customAudioUrl}
-            completedTtsJobId={v.completedTtsJobId}
-            onOpenSaveModal={v.handleOpenSaveModal}
-          />
-          <SaveVoiceModal
-            open={v.saveModalOpen}
-            onClose={() => v.setSaveModalOpen(false)}
-            saveName={v.saveName}
-            onNameChange={v.setSaveName}
-            saveTagIds={v.saveTagIds}
-            onTagIdsChange={v.setSaveTagIds}
-            saveNewTag={v.saveNewTag}
-            onNewTagChange={v.setSaveNewTag}
-            onAddTag={v.handleAddTagInModal}
-            allTags={v.allTags}
-            savePending={v.saveToLibraryMutation.isPending}
-            onConfirm={v.handleConfirmSave}
-          />
-        </>
-      )}
-
-      {v.voiceMode === "clone" && (
-        <CloneVoiceSection
-          clonedVoices={v.clonedVoices}
-          hasProcessing={v.hasProcessing}
-          selectedClonedVoice={v.selectedClonedVoice}
-          onSelect={v.handleSelectClonedVoice}
-          onOpenCloneModal={v.handleOpenCloneModal}
-          playingVoiceId={v.playingCloneVoice}
-          onPlaySample={v.handlePlayCloneSample}
-          CLONE_STATUS_CONFIG={v.CLONE_STATUS_CONFIG}
-          formatDuration={v.formatDuration}
-        />
-      )}
    </div>
  )
 }
@@ -91,7 +91,7 @@ const Step6CoverSettings: React.FC<Step6CoverSettingsProps> = (props) => {
                </span>
              </div>
            )}
-            <div className="xx-cover-preview-ratio">16:9</div>
+            <div className="xx-cover-preview-ratio">9:16</div>
          </div>
        </>
      )}
@@ -1,7 +1,7 @@
 /**
 * 手动选择素材列表
 */
-import React from "react"
+import React, { useRef, useCallback } from "react"
 import { Typography } from "antd"
 import type { AssetItem } from "@/api/assets"

@@ -20,6 +20,33 @@ const ManualMaterialList: React.FC<ManualMaterialListProps> = ({
  selectedMaterials,
  onToggle,
 }) => {
+  // 追踪当前正在播放的视频元素，确保同时只有一个视频播放
+  const activeVideoRef = useRef<HTMLVideoElement | null>(null)
+
+  const handleVideoMouseEnter = useCallback((e: React.MouseEvent<HTMLVideoElement>) => {
+    const video = e.currentTarget
+    // 暂停之前正在播放的视频（检查是否仍在 DOM 中）
+    if (
+      activeVideoRef.current &&
+      activeVideoRef.current !== video &&
+      document.body.contains(activeVideoRef.current)
+    ) {
+      activeVideoRef.current.pause()
+      activeVideoRef.current.currentTime = 0
+    }
+    activeVideoRef.current = video
+    video.play().catch(() => {})
+  }, [])
+
+  const handleVideoMouseLeave = useCallback((e: React.MouseEvent<HTMLVideoElement>) => {
+    const video = e.currentTarget
+    video.pause()
+    video.currentTime = 0
+    if (activeVideoRef.current === video) {
+      activeVideoRef.current = null
+    }
+  }, [])
+
  return (
    <div style={{ marginTop: 14 }}>
      {materialsLoading ? (
@@ -32,6 +59,8 @@ const ManualMaterialList: React.FC<ManualMaterialListProps> = ({
        <div style={{ display: "flex", flexDirection: "column", gap: 6 }}>
          {materials.items.map((m) => {
            const checked = selectedMaterials.includes(m.id)
+            const isVideo = m.mime_type?.startsWith("video/") ?? false
+            const thumbSrc = m.thumbnail_url || undefined
            return (
              <label
                key={m.id}
@@ -53,13 +82,103 @@ const ManualMaterialList: React.FC<ManualMaterialListProps> = ({
                  type="checkbox"
                  checked={checked}
                  onChange={() => onToggle(m.id)}
-                  style={{ accentColor: "var(--primary-color, #4f46e5)" }}
+                  style={{
+                    accentColor: "var(--primary-color, #4f46e5)",
+                    flexShrink: 0,
+                  }}
                />
+                {/* 缩略图预览 48×48 */}
+                <div
+                  style={{
+                    width: 48,
+                    height: 48,
+                    borderRadius: 6,
+                    overflow: "hidden",
+                    background: "#e2e8f0",
+                    flexShrink: 0,
+                    display: "flex",
+                    alignItems: "center",
+                    justifyContent: "center",
+                  }}
+                >
+                  {isVideo && m.file_url ? (
+                    <video
+                      src={m.file_url}
+                      poster={m.thumbnail_url || undefined}
+                      muted
+                      loop
+                      playsInline
+                      preload="none"
+                      onMouseEnter={handleVideoMouseEnter}
+                      onMouseLeave={handleVideoMouseLeave}
+                      style={{
+                        width: "100%",
+                        height: "100%",
+                        objectFit: "cover",
+                        cursor: "pointer",
+                      }}
+                    />
+                  ) : thumbSrc ? (
+                    <img
+                      src={thumbSrc}
+                      alt={m.name}
+                      style={{
+                        width: "100%",
+                        height: "100%",
+                        objectFit: "cover",
+                      }}
+                      onError={(e) => {
+                        const target = e.target as HTMLImageElement
+                        target.style.display = "none"
+                        const fallback = target.nextElementSibling as HTMLElement | null
+                        if (fallback) fallback.style.display = "flex"
+                      }}
+                    />
+                  ) : null}
+                  {!thumbSrc && !isVideo && (
+                    <span
+                      style={{
+                        fontSize: 20,
+                        opacity: 0.5,
+                        display: "flex",
+                      }}
+                    >
+                      🎵
+                    </span>
+                  )}
+                  {!thumbSrc && isVideo && !m.file_url && (
+                    <span
+                      style={{
+                        fontSize: 20,
+                        opacity: 0.5,
+                        display: "flex",
+                      }}
+                    >
+                      🎬
+                    </span>
+                  )}
+                  {/* img onError 时显示的 fallback（初始隐藏） */}
+                  {thumbSrc && !(isVideo && m.file_url) && (
+                    <span
+                      style={{
+                        fontSize: 20,
+                        opacity: 0.5,
+                        display: "none",
+                      }}
+                    >
+                      {isVideo ? "🎬" : "🎵"}
+                    </span>
+                  )}
+                </div>
                <span
                  style={{
                    fontSize: 13,
                    color: "var(--text-primary)",
                    flex: 1,
+                    minWidth: 0,
+                    overflow: "hidden",
+                    textOverflow: "ellipsis",
+                    whiteSpace: "nowrap",
                  }}
                >
                  {m.name}
@@ -68,9 +187,10 @@ const ManualMaterialList: React.FC<ManualMaterialListProps> = ({
                  style={{
                    fontSize: 11,
                    color: "var(--text-tertiary, #94a3b8)",
+                    flexShrink: 0,
                  }}
                >
-                  {m.mime_type.split("/")[1].toUpperCase()}
+                  {m.mime_type?.split("/")?.[1]?.toUpperCase() ?? "FILE"}
                </span>
              </label>
            )
@@ -3,11 +3,13 @@
 * 封装生成确认页的展示逻辑
 */
 import { useMemo } from "react"
+import { useQuery } from "@tanstack/react-query"
 import type { EditingTemplate } from "@/api/editing-planner"
 import type { GeneratedVideo } from "@/api/template-editor"
 import type { CoverConfig } from "../../editing-planner/types"
 import type { VoiceClone } from "@/api/voice-clone"
 import type { PresetVoiceItem } from "@/api/voices"
+import { getAssetsByKind } from "@/api/assets"
 import { COVER_MODE_LABELS } from "../constants"

 interface UseStep7GenerateProps {
@@ -39,11 +41,11 @@ export function useStep7Generate({
  selectedMaterials,
  smartSelectedIds,
  title,
-  voiceMode,
+  voiceMode: _voiceMode,
  selectedVoice,
-  selectedClonedVoice,
-  presetVoices,
-  clonedVoices,
+  selectedClonedVoice: _selectedClonedVoice,
+  presetVoices: _presetVoices,
+  clonedVoices: _clonedVoices,
  coverSettings,
  generateCount,
  onGenerateCountChange,
@@ -65,14 +67,16 @@ export function useStep7Generate({
    return `${selectedMaterials.length} 个素材`
  }, [materialMode, selectedMaterials.length, smartSelectedIds.length])

+  // 从配音素材库中查找 voiceName
+  const { data: voiceMaterials = [] } = useQuery({
+    queryKey: ["assets", "voice"],
+    queryFn: () => getAssetsByKind("voice", { limit: 50 }),
+  })
+
  const voiceName = useMemo(() => {
-    if (voiceMode === "clone") {
-      const cv = clonedVoices.find((v) => v.id === selectedClonedVoice)
-      return cv ? cv.name : "未选择"
-    }
-    const pv = presetVoices.find((v) => v.voice_id === selectedVoice)
-    return pv ? pv.name : "未选择"
-  }, [voiceMode, selectedVoice, selectedClonedVoice, presetVoices, clonedVoices])
+    const asset = voiceMaterials.find((v) => v.id === selectedVoice)
+    return asset ? asset.name : "未选择"
+  }, [voiceMaterials, selectedVoice])

  const coverSummary = useMemo(() => {
    if (!coverSettings.enabled) return "不使用"
@@ -117,11 +117,9 @@ const VoiceLibrary: React.FC = () => {
    uploadOpen,
    uploadFile,
    uploadName,
-    uploadGender,
    uploadDesc,
    uploadProgress,
    setUploadName,
-    setUploadGender,
    setUploadDesc,
    setUploadOpen,
    handleFileSelect,
@@ -252,14 +250,12 @@ const VoiceLibrary: React.FC = () => {
        uploadOpen={uploadOpen}
        uploadFile={uploadFile}
        uploadName={uploadName}
-        uploadGender={uploadGender}
        uploadDesc={uploadDesc}
        uploadProgress={uploadProgress}
        onUploadClose={handleUploadClose}
        onFileSelect={handleFileSelect}
        onFileRemove={handleFileRemove}
        onNameChange={setUploadName}
-        onGenderChange={setUploadGender}
        onDescChange={setUploadDesc}
        onUpload={handleUpload}
        ttsOpen={ttsOpen}
@@ -14,14 +14,12 @@ const UploadVoiceModal: React.FC<UploadVoiceModalProps> = ({
  open,
  uploadFile,
  uploadName,
-  uploadGender,
  uploadDesc,
  uploadProgress,
  onClose,
  onFileSelect,
  onFileRemove,
  onNameChange,
-  onGenderChange,
  onDescChange,
  onUpload,
 }) => {
@@ -30,7 +28,7 @@ const UploadVoiceModal: React.FC<UploadVoiceModalProps> = ({

  return (
    <Modal
-      title="上传音频"
+      title={<span style={{ fontSize: 16, fontWeight: 600 }}>上传配音素材</span>}
      open={open}
      onCancel={() => {
        if (uploading) return // 上传中不可关闭
@@ -45,7 +43,7 @@ const UploadVoiceModal: React.FC<UploadVoiceModalProps> = ({
          display: "flex",
          flexDirection: "column",
          gap: 16,
-          padding: "8px 0",
+          padding: "12px 0 4px",
        }}
      >
        {/* 拖拽上传区 */}
@@ -64,11 +62,9 @@ const UploadVoiceModal: React.FC<UploadVoiceModalProps> = ({
        {/* 表单字段 */}
        <FormFields
          name={uploadName}
-          gender={uploadGender}
          desc={uploadDesc}
          disabled={uploading}
          onNameChange={onNameChange}
-          onGenderChange={onGenderChange}
          onDescChange={onDescChange}
        />

@@ -2,7 +2,7 @@
 * VoiceLibrary 弹窗集合
 */
 import React from "react"
-import type { ClonedVoiceDisplay, PresetVoiceDisplay, VoiceGender } from "../types"
+import type { ClonedVoiceDisplay, PresetVoiceDisplay } from "../types"
 import type { VoiceClone } from "@/api/voice-clone"
 import type { TtsStatus } from "./TtsModal"
 import CloneModal from "@/components/voice/CloneModal"
@@ -27,14 +27,12 @@ export interface VoiceModalsProps {
  uploadOpen: boolean
  uploadFile: File | null
  uploadName: string
-  uploadGender: VoiceGender
  uploadDesc: string
  uploadProgress: number | null
  onUploadClose: () => void
  onFileSelect: (file: File) => void
  onFileRemove: () => void
  onNameChange: (name: string) => void
-  onGenderChange: (gender: VoiceGender) => void
  onDescChange: (desc: string) => void
  onUpload: () => void

@@ -67,14 +65,12 @@ export const VoiceModals: React.FC<VoiceModalsProps> = ({
  uploadOpen,
  uploadFile,
  uploadName,
-  uploadGender,
  uploadDesc,
  uploadProgress,
  onUploadClose,
  onFileSelect,
  onFileRemove,
  onNameChange,
-  onGenderChange,
  onDescChange,
  onUpload,
  ttsOpen,
@@ -113,14 +109,12 @@ export const VoiceModals: React.FC<VoiceModalsProps> = ({
        open={uploadOpen}
        uploadFile={uploadFile}
        uploadName={uploadName}
-        uploadGender={uploadGender}
        uploadDesc={uploadDesc}
        uploadProgress={uploadProgress}
        onClose={onUploadClose}
        onFileSelect={onFileSelect}
        onFileRemove={onFileRemove}
        onNameChange={onNameChange}
-        onGenderChange={onGenderChange}
        onDescChange={onDescChange}
        onUpload={onUpload}
      />
@@ -1,4 +1,5 @@
 import React from "react"
+import { Button } from "antd"

 interface ActionButtonsProps {
  uploading: boolean
@@ -13,12 +14,6 @@ const ActionButtons: React.FC<ActionButtonsProps> = ({
  onCancel,
  onUpload,
 }) => {
-  const disabled = uploading || !canUpload
-
-  const handleUpload = () => {
-    onUpload()
-  }
-
  return (
    <div
      style={{
@@ -28,39 +23,32 @@ const ActionButtons: React.FC<ActionButtonsProps> = ({
        paddingTop: 4,
      }}
    >
-      <button
-        type="button"
+      <Button
        onClick={onCancel}
        disabled={uploading}
        style={{
-          padding: "8px 20px",
+          height: 36,
          borderRadius: 8,
-          border: "1px solid var(--border-color)",
-          background: "transparent",
          fontSize: 13,
-          cursor: uploading ? "not-allowed" : "pointer",
-          color: "var(--text-secondary)",
        }}
      >
        取消
-      </button>
-      <button
-        type="button"
-        onClick={handleUpload}
-        disabled={disabled}
+      </Button>
+      <Button
+        type="primary"
+        onClick={onUpload}
+        disabled={!canUpload}
+        loading={uploading}
        style={{
-          padding: "8px 20px",
+          height: 36,
          borderRadius: 8,
-          border: "none",
-          background: disabled ? "var(--text-tertiary)" : "var(--primary-color)",
-          color: "#fff",
          fontSize: 13,
          fontWeight: 600,
-          cursor: disabled ? "not-allowed" : "pointer",
+          minWidth: 100,
        }}
      >
-        {uploading ? "上传中..." : "开始上传"}
-      </button>
+        {uploading ? "上传中" : "开始上传"}
+      </Button>
    </div>
  )
 }
@@ -10,28 +10,55 @@ const FileInfoCard: React.FC<FileInfoCardProps> = ({ file }) => {
  return (
    <div
      style={{
-        padding: "10px 12px",
+        padding: "10px 14px",
        background: "var(--bg-secondary)",
-        borderRadius: 8,
+        borderRadius: 10,
        display: "flex",
        alignItems: "center",
-        gap: 10,
+        gap: 12,
+        border: "1px solid var(--border-color)",
+        boxShadow: "0 1px 3px rgba(0, 0, 0, 0.04)",
      }}
    >
-      <SoundOutlined style={{ fontSize: 18, color: "var(--primary-color)" }} />
+      {/* 文件图标圆形底托 */}
+      <div
+        style={{
+          width: 36,
+          height: 36,
+          borderRadius: "50%",
+          background: "var(--primary-soft, rgba(22, 119, 255, 0.1))",
+          display: "flex",
+          alignItems: "center",
+          justifyContent: "center",
+          flexShrink: 0,
+        }}
+      >
+        <SoundOutlined style={{ fontSize: 16, color: "var(--primary-color)" }} />
+      </div>
+
+      {/* 文件信息 */}
      <div style={{ flex: 1, minWidth: 0 }}>
        <div
          style={{
            fontSize: 13,
            fontWeight: 500,
+            color: "var(--text-primary)",
            overflow: "hidden",
            textOverflow: "ellipsis",
            whiteSpace: "nowrap",
+            lineHeight: 1.4,
          }}
        >
          {file.name}
        </div>
-        <div style={{ fontSize: 11, color: "var(--text-secondary)" }}>
+        <div
+          style={{
+            fontSize: 12,
+            color: "var(--text-secondary)",
+            marginTop: 2,
+            lineHeight: 1.3,
+          }}
+        >
          {formatFileSize(file.size)}
        </div>
      </div>
@@ -14,6 +14,8 @@ const FileUploadZone: React.FC<FileUploadZoneProps> = ({
  onFileSelect,
  onFileRemove,
 }) => {
+  const [isHovered, setIsHovered] = React.useState(false)
+
  return (
    <Upload.Dragger
      accept={UPLOAD_CONFIG.accept}
@@ -28,25 +30,73 @@ const FileUploadZone: React.FC<FileUploadZoneProps> = ({
      showUploadList={false}
      disabled={disabled}
    >
-      <p
+      <div
        style={{
-          fontSize: 32,
-          color: "var(--primary-color)",
-          marginBottom: 8,
+          display: "flex",
+          flexDirection: "column",
+          alignItems: "center",
+          justifyContent: "center",
+          padding: "20px 16px",
+          borderRadius: 12,
+          background: isHovered
+            ? "var(--primary-soft, rgba(22, 119, 255, 0.08))"
+            : "var(--primary-soft, rgba(22, 119, 255, 0.03))",
+          border: `2px dashed ${isHovered ? "var(--primary-color)" : "var(--border-color)"}`,
+          transition: "all 0.25s ease",
+          cursor: disabled ? "not-allowed" : "pointer",
+          opacity: disabled ? 0.6 : 1,
        }}
+        onMouseEnter={() => setIsHovered(true)}
+        onMouseLeave={() => setIsHovered(false)}
      >
-        <UploadOutlined />
-      </p>
-      <p style={{ fontSize: 14, fontWeight: 500, margin: "0 0 4px" }}>点击或拖拽音频文件到此处</p>
-      <p
-        style={{
-          fontSize: 12,
-          color: "var(--text-secondary)",
-          margin: 0,
-        }}
-      >
-        支持 MP3、WAV、AAC、FLAC 等格式，最大 {UPLOAD_CONFIG.maxSizeMB}MB
-      </p>
+        {/* 图标圆形底托 */}
+        <div
+          style={{
+            width: 44,
+            height: 44,
+            borderRadius: "50%",
+            background: "var(--primary-soft, rgba(22, 119, 255, 0.1))",
+            display: "flex",
+            alignItems: "center",
+            justifyContent: "center",
+            marginBottom: 10,
+            transition: "transform 0.2s ease",
+            transform: isHovered ? "scale(1.08)" : "scale(1)",
+          }}
+        >
+          <UploadOutlined
+            style={{
+              fontSize: 20,
+              color: "var(--primary-color)",
+            }}
+          />
+        </div>
+
+        {/* 主文案 */}
+        <p
+          style={{
+            fontSize: 14,
+            fontWeight: 500,
+            color: "var(--text-primary)",
+            margin: "0 0 6px",
+            lineHeight: 1.4,
+          }}
+        >
+          点击或拖拽音频文件到此处
+        </p>
+
+        {/* 副文案 */}
+        <p
+          style={{
+            fontSize: 12,
+            color: "var(--text-secondary)",
+            margin: 0,
+            lineHeight: 1.4,
+          }}
+        >
+          支持 MP3、WAV、AAC、FLAC 等格式，最大 {UPLOAD_CONFIG.maxSizeMB}MB
+        </p>
+      </div>
    </Upload.Dragger>
  )
 }
@@ -1,101 +1,92 @@
 import React from "react"
-import type { VoiceGender } from "@/pages/voices/types"
-import { GENDER_OPTIONS, UPLOAD_CONFIG } from "./types"
+import { UPLOAD_CONFIG } from "./types"

 interface FormFieldsProps {
  name: string
-  gender: VoiceGender
  desc: string
  disabled: boolean
  onNameChange: (name: string) => void
-  onGenderChange: (gender: VoiceGender) => void
  onDescChange: (desc: string) => void
 }

 const labelStyle: React.CSSProperties = {
-  fontSize: 13,
-  color: "var(--text-secondary)",
+  fontSize: 12,
+  fontWeight: 600,
+  color: "var(--text-primary, #333)",
  marginBottom: 6,
+  letterSpacing: "0.02em",
 }

-const inputStyle: React.CSSProperties = {
+const baseInputStyle: React.CSSProperties = {
  width: "100%",
-  padding: "8px 12px",
+  padding: "10px 12px",
  border: "1px solid var(--border-color)",
  borderRadius: 8,
  fontSize: 13,
  background: "var(--bg-primary)",
  color: "var(--text-primary)",
  outline: "none",
+  transition: "border-color 0.2s ease, box-shadow 0.2s ease",
+  lineHeight: 1.5,
 }

+const focusStyle: React.CSSProperties = {
+  borderColor: "var(--primary-color)",
+  boxShadow: "0 0 0 2px var(--primary-soft, rgba(22, 119, 255, 0.12))",
+}
+
+
 const FormFields: React.FC<FormFieldsProps> = ({
  name,
-  gender,
  desc,
  disabled,
  onNameChange,
-  onGenderChange,
  onDescChange,
 }) => {
  return (
    <>
      {/* 名称 */}
      <div>
-        <div style={labelStyle}>素材名称</div>
+        <div style={labelStyle}>配音名称</div>
        <input
          value={name}
          onChange={(e) => onNameChange(e.target.value)}
-          placeholder="输入素材名称"
+          placeholder="输入配音名称"
          maxLength={UPLOAD_CONFIG.maxNameLength}
          disabled={disabled}
-          style={inputStyle}
+          style={baseInputStyle}
+          onFocus={(e) => {
+            Object.assign(e.target.style, focusStyle)
+          }}
+          onBlur={(e) => {
+            e.target.style.borderColor = "var(--border-color)"
+            e.target.style.boxShadow = "none"
+          }}
        />
      </div>

-      {/* 性别选择 */}
-      <div>
-        <div style={labelStyle}>音色性别</div>
-        <div style={{ display: "flex", gap: 8 }}>
-          {GENDER_OPTIONS.map((opt) => (
-            <button
-              key={opt.value}
-              type="button"
-              onClick={() => onGenderChange(opt.value)}
-              disabled={disabled}
-              style={{
-                flex: 1,
-                padding: "6px 0",
-                borderRadius: 8,
-                border: `1px solid ${gender === opt.value ? "var(--primary-color)" : "var(--border-color)"}`,
-                background: gender === opt.value ? "var(--primary-soft)" : "transparent",
-                color: gender === opt.value ? "var(--primary-color)" : "var(--text-secondary)",
-                fontSize: 13,
-                fontWeight: gender === opt.value ? 600 : 400,
-                cursor: disabled ? "not-allowed" : "pointer",
-                transition: "all 0.2s",
-              }}
-            >
-              {opt.label}
-            </button>
-          ))}
-        </div>
-      </div>
-
      {/* 描述 */}
      <div>
-        <div style={labelStyle}>音色描述（可选）</div>
+        <div style={labelStyle}>配音描述（可选）</div>
        <textarea
          value={desc}
          onChange={(e) => onDescChange(e.target.value)}
-          placeholder="描述这个音色的特点..."
+          placeholder="描述这段配音的内容..."
          maxLength={UPLOAD_CONFIG.maxDescLength}
          rows={2}
          disabled={disabled}
          style={{
-            ...inputStyle,
+            ...baseInputStyle,
            resize: "vertical",
            fontFamily: "inherit",
+            minHeight: 56,
+          }}
+          onFocus={(e) => {
+            Object.assign(e.target.style, focusStyle)
+          }}
+          onBlur={(e) => {
+            e.target.style.borderColor = "var(--border-color)"
+            e.target.style.boxShadow = "none"
          }}
        />
      </div>
@@ -6,25 +6,40 @@ interface UploadProgressProps {

 const UploadProgress: React.FC<UploadProgressProps> = ({ progress }) => {
  return (
-    <div style={{ textAlign: "center", padding: "8px 0" }}>
+    <div
+      style={{
+        padding: "12px 16px",
+        background: "var(--bg-secondary)",
+        borderRadius: 10,
+        textAlign: "center",
+        border: "1px solid var(--border-color)",
+      }}
+    >
      <div
        style={{
-          fontSize: 22,
+          fontSize: 20,
          fontWeight: 700,
          color: "var(--primary-color)",
+          lineHeight: 1.2,
        }}
      >
        {progress}%
      </div>
-      <div style={{ fontSize: 12, color: "var(--text-secondary)" }}>
+      <div
+        style={{
+          fontSize: 12,
+          color: "var(--text-secondary)",
+          marginTop: 2,
+          marginBottom: 10,
+        }}
+      >
        {progress < 100 ? "上传中..." : "处理中..."}
      </div>
      <div
        style={{
          height: 4,
-          background: "var(--bg-tertiary)",
+          background: "var(--bg-tertiary, #f0f0f0)",
          borderRadius: 2,
-          marginTop: 8,
          overflow: "hidden",
        }}
      >
@@ -32,7 +47,7 @@ const UploadProgress: React.FC<UploadProgressProps> = ({ progress }) => {
          style={{
            height: "100%",
            width: `${progress}%`,
-            background: "var(--primary-color)",
+            background: progress < 100 ? "var(--primary-color)" : "var(--success-color, #52c41a)",
            borderRadius: 2,
            transition: "width 0.3s ease",
          }}
@@ -1,27 +1,17 @@
-import type { VoiceGender } from "@/pages/voices/types"
-
 export interface UploadVoiceModalProps {
  open: boolean
  uploadFile: File | null
  uploadName: string
-  uploadGender: VoiceGender
  uploadDesc: string
  uploadProgress: number | null
  onClose: () => void
  onFileSelect: (file: File) => void
  onFileRemove: () => void
  onNameChange: (name: string) => void
-  onGenderChange: (gender: VoiceGender) => void
  onDescChange: (desc: string) => void
  onUpload: () => void
 }

-export const GENDER_OPTIONS: { value: VoiceGender; label: string }[] = [
-  { value: "female", label: "女声" },
-  { value: "male", label: "男声" },
-  { value: "child", label: "童声" },
-]
-
 export const UPLOAD_CONFIG = {
  maxSizeMB: 200,
  maxNameLength: 100,
@@ -1,6 +1,5 @@
 import { useState, useCallback } from "react"
 import { useMutation, useQueryClient } from "@tanstack/react-query"
-import { type VoiceGender } from "../types"
 import { uploadAssetDirect, getAssetLibraries, createAsset } from "@/api/assets"
 import { getAudioDuration } from "../utils/audio"
 import { buildVoiceMetadata } from "../types"
@@ -19,17 +18,11 @@ export function useVoiceUpload({ showToast }: UseVoiceUploadProps) {
  const [uploadOpen, setUploadOpen] = useState(false)
  const [uploadFile, setUploadFile] = useState<File | null>(null)
  const [uploadName, setUploadName] = useState("")
-  const [uploadGender, setUploadGender] = useState<VoiceGender>("female")
  const [uploadDesc, setUploadDesc] = useState("")
  const [uploadProgress, setUploadProgress] = useState<number | null>(null)

  const uploadMutation = useMutation({
-    mutationFn: async (data: {
-      file: File
-      name: string
-      gender: VoiceGender
-      description: string
-    }) => {
+    mutationFn: async (data: { file: File; name: string; description: string }) => {
      setUploadProgress(0)
      try {
        /* 获取或创建默认配音库 */
@@ -57,7 +50,6 @@ export function useVoiceUpload({ showToast }: UseVoiceUploadProps) {
          storage_key,
          mime_type: data.file.type || "audio/mpeg",
          metadata: buildVoiceMetadata({
-            gender: data.gender,
            description: data.description,
            duration,
          }),
@@ -71,8 +63,15 @@ export function useVoiceUpload({ showToast }: UseVoiceUploadProps) {
      showToast("上传成功", "success")
      handleUploadClose()
    },
-    onError: (err: Error) => {
-      showToast(err.message || "上传失败，请重试", "error")
+    onError: (err: Error & { response?: { status?: number } }) => {
+      const status = err?.response?.status
+      let msg = "上传失败，请重试"
+      if (status && status >= 500) {
+        msg = "服务器暂时不可用，请稍后重试"
+      } else if (err?.message) {
+        msg = err.message
+      }
+      showToast(msg, "error")
    },
  })

@@ -102,10 +101,9 @@ export function useVoiceUpload({ showToast }: UseVoiceUploadProps) {
    uploadMutation.mutate({
      file: uploadFile,
      name: uploadName.trim(),
-      gender: uploadGender,
      description: uploadDesc.trim(),
    })
-  }, [uploadFile, uploadName, uploadGender, uploadDesc, uploadMutation])
+  }, [uploadFile, uploadName, uploadDesc, uploadMutation])

  return {
    // 弹窗状态
@@ -114,13 +112,11 @@ export function useVoiceUpload({ showToast }: UseVoiceUploadProps) {
    // 表单状态
    uploadFile,
    uploadName,
-    uploadGender,
    uploadDesc,
    uploadProgress,
    isUploading: uploadMutation.isPending,
    // Setters
    setUploadName,
-    setUploadGender,
    setUploadDesc,
    // Handlers
    handleFileSelect,
@@ -3,7 +3,6 @@ import { Navigate } from "react-router-dom"
 import { useAuthStore } from "@/store/authStore"

 /** 受保护的路由组件 */
-// eslint-disable-next-line react-refresh/only-export-components
 export const ProtectedRoute = ({ children }: { children: React.ReactNode }) => {
  const isAuthenticated = useAuthStore((state) => state.isAuthenticated)
  const hasAccessToken = Boolean(localStorage.getItem("access_token"))
@@ -26,12 +26,10 @@ import { refreshAccessToken } from "@/api/auth"
 import apiClient from "@/api/client"

 // 从真实实例取出拦截器回调
-// eslint-disable-next-line @typescript-eslint/no-explicit-any
 const requestHandlers = (apiClient as any).interceptors.request.handlers as Array<{
  fulfilled: (config: unknown) => unknown
  rejected: (error: unknown) => unknown
 }>
-// eslint-disable-next-line @typescript-eslint/no-explicit-any
 const responseHandlers = (apiClient as any).interceptors.response.handlers as Array<{
  fulfilled: (response: unknown) => unknown
  rejected: (error: unknown) => Promise<unknown>
@@ -258,7 +256,6 @@ describe("apiClient - 401 token refresh", () => {
      isAuthenticated: false,
      clearAuth: mockClearAuth,
      setAuth: vi.fn(),
-      // eslint-disable-next-line @typescript-eslint/no-explicit-any
    } as any)

    const err = makeAxiosError(401, { detail: "Unauthorized" })
@@ -276,7 +273,6 @@ describe("apiClient - 401 token refresh", () => {
      isAuthenticated: true,
      clearAuth: vi.fn(),
      setAuth: mockSetAuth,
-      // eslint-disable-next-line @typescript-eslint/no-explicit-any
    } as any)
    vi.mocked(refreshAccessToken).mockResolvedValue({
      access_token: "new-access",
@@ -306,7 +302,6 @@ describe("apiClient - 401 token refresh", () => {
      isAuthenticated: true,
      clearAuth: mockClearAuth,
      setAuth: vi.fn(),
-      // eslint-disable-next-line @typescript-eslint/no-explicit-any
    } as any)
    vi.mocked(refreshAccessToken).mockRejectedValue(new Error("refresh failed") as never)

@@ -32,9 +32,7 @@ afterEach(() => {
 declare global {
  // eslint-disable-next-line @typescript-eslint/no-namespace
  namespace Vi {
-    // eslint-disable-next-line @typescript-eslint/no-explicit-any
    interface Assertion<T = any> extends jest.Matchers<void, T> {}
-    // eslint-disable-next-line @typescript-eslint/no-explicit-any
    interface AsymmetricMatchersContaining extends jest.Matchers<void, any> {}
  }
 }
@@ -82,14 +82,12 @@ def mix_audio(
    """音频后处理混音.

    处理逻辑：
-    1. 主音频源按优先级查找：main > broll（background 不参与主音频，通常是图片无音轨）
-    2. 主图层音频按顺序 concat 拼接
-    3. 独立音频轨（audio role）用 amix 混入
-    4. 输出时长截断到 video_duration
-    5. 无音频流的 clip 会被自动跳过，避免 FFmpeg 引用 [i:a] 失败
-    6. 如果提供了 bgm_path，则额外混入 BGM（支持淡入淡出、循环、人声闪避）
-    7. 如果配置了 audio_tracks，则混入多轨道音频（配音、音效等）
-    8. 如果配置了降噪，最后应用降噪
+    1. 丢弃主图层（main/broll/overlay/corner_voice）的原始音频，避免录入源视频杂音
+    2. 仅使用独立音频轨（audio role，TTS/配音）作为主音频
+    3. 如果提供了 bgm_path，则额外混入 BGM（支持淡入淡出、循环、人声闪避）
+    4. 如果配置了 audio_tracks，则混入多轨道音频（配音、音效等）
+    5. 输出时长截断到 video_duration
+    6. 如果配置了降噪，最后应用降噪

    Args:
        ctx: 渲染上下文
@@ -125,8 +123,10 @@ def mix_audio(
    if "audio" in layer_map:
        audio_clips = layer_map["audio"].clips

+    # ── 丢弃源视频的原始音频（避免录入杂音），成片仅保留 TTS 配音 + BGM ──
+    main_clips = []
+
    # ── 防御：过滤掉无音频流的 clip ──
-    main_clips = [c for c in main_clips if clip_has_audio(ctx, c)]
    audio_clips = [c for c in audio_clips if clip_has_audio(ctx, c)]

    if not main_clips and not audio_clips:
@@ -144,12 +144,21 @@ def mix_audio(
    # 构建音频处理命令
    output_path = ctx.work_dir / f"audio_{ctx.plan_id}.aac"

-    # 简单场景：只有主图层 + 无独立音频 → 直接从视频提取音频并拼接
-    if main_clips and not audio_clips:
-        concat_main_audio(ctx, main_clips, output_path, video_duration)
+    # 源视频原始音频已被丢弃（main_clips = []），最终音频完全由独立音频轨 + BGM + 多轨配置组成。
+    # 当无 main_clips 时，将独立音频轨作为主音频走 concat 拼接；当二者均有则走 amix 混音。
+    if main_clips:
+        effective_main = main_clips
+        effective_audio = audio_clips
+    else:
+        effective_main = audio_clips
+        effective_audio = []
+
+    # 简单场景：只有主音频 + 无独立音频 → 直接拼接
+    if effective_main and not effective_audio:
+        concat_main_audio(ctx, effective_main, output_path, video_duration)
    else:
        # 有独立音频轨 → amix 混音
-        mix_with_independent_audio(ctx, main_clips, audio_clips, output_path, video_duration)
+        mix_with_independent_audio(ctx, effective_main, effective_audio, output_path, video_duration)

    # ── BGM 混音 ──
    if bgm_path and bgm_config and bgm_config.get("enabled", False):
@@ -163,6 +163,36 @@ def _compose_with_unified_engine(task, job_service, job, plan_id: str, db) -> di
        job_service.fail_job(job_id, error_msg[:500])
        raise RuntimeError(result.error_message)

+    # 生成封面（如果配置启用）
+    cover_url = None
+    try:
+        from video_processing.cover_generator import generate_cover_from_plan
+
+        from packages.adapters.sqlalchemy_impl.edit_plan_repository import (
+            SQLAlchemyEditPlanRepository as EditPlanRepository,
+        )
+
+        # 获取 plan 对象
+        plan_repo = EditPlanRepository(db)
+        plan = plan_repo.get(plan_id)
+
+        if plan and result.output_path:
+            # 检查 cover_config
+            cover_config = (plan.config or {}).get("cover_config")
+            if cover_config and cover_config.get("enabled", False):
+                from pathlib import Path
+
+                output_dir = Path(result.output_path).parent
+                cover_path = generate_cover_from_plan(plan, result.output_path, output_dir)
+                if cover_path:
+                    # 生成 cover_url（相对路径或上传到存储）
+                    cover_url = f"/covers/{plan_id}.jpg"
+                    logger.info("封面生成成功: plan_id=%s cover_path=%s", plan_id, cover_path)
+            else:
+                logger.info("封面生成未启用: plan_id=%s", plan_id)
+    except Exception as e:
+        logger.warning("封面生成失败（不影响视频合成）: plan_id=%s error=%s", plan_id, e)
+
    # 更新 Job 状态为完成
    result_data = {
        "plan_id": plan_id,
@@ -175,6 +205,7 @@ def _compose_with_unified_engine(task, job_service, job, plan_id: str, db) -> di
        "width": result.width,
        "height": result.height,
        "file_size": result.file_size,
+        "cover_url": cover_url,
    }
    job_service.complete_job(job_id, result=result_data)

@@ -1,9 +1,10 @@
 """Quota system with registry pattern.

-Three subscription tiers with different limits:
+Four subscription tiers with different limits:
 - free:    2GB storage, 5 videos/month, 3 concurrent, 3 templates, 50 titles, 10 voiceovers, no AI voice
 - basic:   20GB storage, 30 videos/month, 10 concurrent, 15 templates, 500 titles, 100 voiceovers, AI voice
 - premium: 100GB storage, 100 videos/month, 20 concurrent, unlimited templates, 500 titles, 100 voiceovers, AI voice
+- pro:     Same as premium (alias for premium tier)

 Quota dimensions are registered by modules via the ModuleRegistry,
 and checked against the user's subscription plan.
@@ -100,6 +101,8 @@ QUOTA_TIERS: Dict[str, QuotaTier] = {
        },
    ),
 }
+# pro 套餐与 premium 配额相同，使用别名引用避免重复维护
+QUOTA_TIERS["pro"] = QUOTA_TIERS["premium"]


 class QuotaWarningLevel:
@@ -354,13 +354,13 @@ class TestQuotaRegistry:
        assert len(reg.list_dimensions()) == len(QuotaDimension)

    def test_list_tiers(self):
-        """三个套餐等级."""
+        """四个套餐等级."""
        reg = QuotaRegistry()
        tiers = reg.list_tiers()
+        assert "pro" in tiers
        assert "free" in tiers
        assert "basic" in tiers
-        assert "premium" in tiers
-        assert len(tiers) == 3
+        assert len(tiers) == 4

    def test_get_tier_existing(self):
        """获取已有的套餐."""
@@ -370,7 +370,7 @@ class TestQuotaRegistry:
        assert tier.name == "free"

    def test_get_tier_nonexistent(self):
-        """获取不存在的套餐返回 None."""
+        """不存在的套餐返回 None"""
        reg = QuotaRegistry()
        assert reg.get_tier("enterprise") is None

@@ -380,7 +380,7 @@ class TestQuotaRegistry:
        assert reg.get_limit("free", QuotaDimension.STORAGE_GB) == 2

    def test_get_limit_nonexistent_plan(self):
-        """不存在的套餐返回 0."""
+        """不存在的套餐 fallback 到 free 配额"""
        reg = QuotaRegistry()
        assert reg.get_limit("enterprise", QuotaDimension.STORAGE_GB) == 0

@@ -0,0 +1,232 @@
+"""测试 compose_video 任务中封面生成集成."""
+
+from __future__ import annotations
+
+from pathlib import Path
+from unittest.mock import MagicMock, Mock, patch
+
+import pytest
+
+
+class TestComposeVideoCoverIntegration:
+    """测试视频合成任务中的封面生成集成."""
+
+    @pytest.fixture
+    def mock_job_service(self):
+        """模拟 JobService."""
+        service = MagicMock()
+        service.get_job.return_value = MagicMock(
+            id="job_123",
+            payload={"plan_id": "plan_456"},
+        )
+        return service
+
+    @pytest.fixture
+    def mock_db(self):
+        """模拟数据库会话."""
+        return MagicMock()
+
+    @pytest.fixture
+    def mock_render_result(self):
+        """模拟渲染结果."""
+        result = MagicMock()
+        result.success = True
+        result.output_path = Path("/tmp/output/video_123.mp4")
+        result.output_url = "https://example.com/video_123.mp4"
+        result.duration = 30.0
+        result.clip_count = 5
+        result.width = 1080
+        result.height = 1920
+        result.file_size = 1024000
+        return result
+
+    @pytest.fixture
+    def mock_plan_with_cover_enabled(self):
+        """模拟启用封面的 plan."""
+        plan = MagicMock()
+        plan.id = "plan_456"
+        plan.config = {
+            "cover_config": {
+                "enabled": True,
+                "mode": "smart",
+            }
+        }
+        return plan
+
+    @pytest.fixture
+    def mock_plan_with_cover_disabled(self):
+        """模拟禁用封面的 plan."""
+        plan = MagicMock()
+        plan.id = "plan_456"
+        plan.config = {
+            "cover_config": {
+                "enabled": False,
+            }
+        }
+        return plan
+
+    @pytest.fixture
+    def mock_plan_without_cover_config(self):
+        """模拟没有 cover_config 的 plan."""
+        plan = MagicMock()
+        plan.id = "plan_456"
+        plan.config = {}
+        return plan
+
+    def test_cover_generation_called_when_enabled(
+        self,
+        mock_job_service,
+        mock_db,
+        mock_render_result,
+        mock_plan_with_cover_enabled,
+    ):
+        """测试封面生成在启用时被调用."""
+        from worker_app.tasks.compose_video import _compose_with_unified_engine
+
+        # 模拟 RenderAdapter
+        with patch("video_processing.render_adapter.RenderAdapter") as MockAdapter:
+            adapter_instance = MagicMock()
+            adapter_instance.render_plan.return_value = mock_render_result
+            adapter_instance.validate_plan.return_value = (True, [], [], 5, 5)
+            MockAdapter.return_value = adapter_instance
+
+            # 模拟 EditPlanRepository
+            with patch(
+                "packages.adapters.sqlalchemy_impl.edit_plan_repository.SQLAlchemyEditPlanRepository"
+            ) as MockPlanRepo:
+                plan_repo_instance = MagicMock()
+                plan_repo_instance.get.return_value = mock_plan_with_cover_enabled
+                MockPlanRepo.return_value = plan_repo_instance
+
+                # 模拟 generate_cover_from_plan
+                with patch("video_processing.cover_generator.generate_cover_from_plan") as mock_gen_cover:
+                    mock_gen_cover.return_value = Path("/tmp/output/cover_plan_456.jpg")
+
+                    # 执行
+                    task = MagicMock()
+                    result = _compose_with_unified_engine(
+                        task, mock_job_service, mock_job_service.get_job(), "plan_456", mock_db
+                    )
+
+                    # 验证封面生成被调用
+                    mock_gen_cover.assert_called_once()
+                    call_args = mock_gen_cover.call_args
+                    assert call_args[0][0] == mock_plan_with_cover_enabled  # plan
+                    assert call_args[0][1] == mock_render_result.output_path  # video_path
+                    assert call_args[0][2] == mock_render_result.output_path.parent  # output_dir
+
+                    # 验证结果包含 cover_url
+                    assert "cover_url" in result["result"]
+                    assert result["result"]["cover_url"] == "/covers/plan_456.jpg"
+
+    def test_cover_generation_skipped_when_disabled(
+        self,
+        mock_job_service,
+        mock_db,
+        mock_render_result,
+        mock_plan_with_cover_disabled,
+    ):
+        """测试封面生成在禁用时被跳过."""
+        from worker_app.tasks.compose_video import _compose_with_unified_engine
+
+        with patch("video_processing.render_adapter.RenderAdapter") as MockAdapter:
+            adapter_instance = MagicMock()
+            adapter_instance.render_plan.return_value = mock_render_result
+            adapter_instance.validate_plan.return_value = (True, [], [], 5, 5)
+            MockAdapter.return_value = adapter_instance
+
+            with patch(
+                "packages.adapters.sqlalchemy_impl.edit_plan_repository.SQLAlchemyEditPlanRepository"
+            ) as MockPlanRepo:
+                plan_repo_instance = MagicMock()
+                plan_repo_instance.get.return_value = mock_plan_with_cover_disabled
+                MockPlanRepo.return_value = plan_repo_instance
+
+                with patch("video_processing.cover_generator.generate_cover_from_plan") as mock_gen_cover:
+                    task = MagicMock()
+                    result = _compose_with_unified_engine(
+                        task, mock_job_service, mock_job_service.get_job(), "plan_456", mock_db
+                    )
+
+                    # 验证封面生成未被调用
+                    mock_gen_cover.assert_not_called()
+
+                    # 验证结果中 cover_url 为 None
+                    assert "cover_url" in result["result"]
+                    assert result["result"]["cover_url"] is None
+
+    def test_cover_generation_skipped_when_no_config(
+        self,
+        mock_job_service,
+        mock_db,
+        mock_render_result,
+        mock_plan_without_cover_config,
+    ):
+        """测试没有 cover_config 时封面生成被跳过."""
+        from worker_app.tasks.compose_video import _compose_with_unified_engine
+
+        with patch("video_processing.render_adapter.RenderAdapter") as MockAdapter:
+            adapter_instance = MagicMock()
+            adapter_instance.render_plan.return_value = mock_render_result
+            adapter_instance.validate_plan.return_value = (True, [], [], 5, 5)
+            MockAdapter.return_value = adapter_instance
+
+            with patch(
+                "packages.adapters.sqlalchemy_impl.edit_plan_repository.SQLAlchemyEditPlanRepository"
+            ) as MockPlanRepo:
+                plan_repo_instance = MagicMock()
+                plan_repo_instance.get.return_value = mock_plan_without_cover_config
+                MockPlanRepo.return_value = plan_repo_instance
+
+                with patch("video_processing.cover_generator.generate_cover_from_plan") as mock_gen_cover:
+                    task = MagicMock()
+                    result = _compose_with_unified_engine(
+                        task, mock_job_service, mock_job_service.get_job(), "plan_456", mock_db
+                    )
+
+                    # 验证封面生成未被调用
+                    mock_gen_cover.assert_not_called()
+
+                    # 验证结果中 cover_url 为 None
+                    assert "cover_url" in result["result"]
+                    assert result["result"]["cover_url"] is None
+
+    def test_cover_generation_failure_does_not_break_video(
+        self,
+        mock_job_service,
+        mock_db,
+        mock_render_result,
+        mock_plan_with_cover_enabled,
+    ):
+        """测试封面生成失败不影响视频合成."""
+        from worker_app.tasks.compose_video import _compose_with_unified_engine
+
+        with patch("video_processing.render_adapter.RenderAdapter") as MockAdapter:
+            adapter_instance = MagicMock()
+            adapter_instance.render_plan.return_value = mock_render_result
+            adapter_instance.validate_plan.return_value = (True, [], [], 5, 5)
+            MockAdapter.return_value = adapter_instance
+
+            with patch(
+                "packages.adapters.sqlalchemy_impl.edit_plan_repository.SQLAlchemyEditPlanRepository"
+            ) as MockPlanRepo:
+                plan_repo_instance = MagicMock()
+                plan_repo_instance.get.return_value = mock_plan_with_cover_enabled
+                MockPlanRepo.return_value = plan_repo_instance
+
+                with patch("video_processing.cover_generator.generate_cover_from_plan") as mock_gen_cover:
+                    # 模拟封面生成抛出异常
+                    mock_gen_cover.side_effect = Exception("FFmpeg failed")
+
+                    task = MagicMock()
+                    result = _compose_with_unified_engine(
+                        task, mock_job_service, mock_job_service.get_job(), "plan_456", mock_db
+                    )
+
+                    # 验证视频合成仍然成功
+                    assert result["status"] == "completed"
+                    assert "result" in result
+                    assert result["result"]["output_url"] == mock_render_result.output_url
+
+                    # 验证结果中 cover_url 为 None
+                    assert result["result"]["cover_url"] is None
@@ -231,6 +231,7 @@ class TestQuotaRegistry:
        assert tier.name == "free"

    def test_get_tier_unknown_returns_none(self):
+        """未知套餐返回 None"""
        reg = QuotaRegistry()
        assert reg.get_tier("nonexistent") is None

@@ -300,8 +301,8 @@ class TestQuotaChecker:
    def test_check_unknown_plan(self):
        checker = QuotaChecker()
        result = checker.check("unknown", "storage_gb", 1.0)
-        assert result.allowed is False
        assert result.limit == 0
+        assert not result.allowed

    def test_check_multiple(self):
        checker = QuotaChecker()
@@ -412,3 +413,30 @@ class TestGlobalSingletons:
        result = quota_checker.check("free", "storage_gb", 1.0)
        assert result.allowed is True
        assert result.limit == 2
+
+
+class TestProTier:
+    """Pro 套餐专项测试"""
+
+    def test_pro_tier_exists(self):
+        """pro 套餐存在于 QUOTA_TIERS"""
+        from packages.domain.quota import QUOTA_TIERS
+
+        assert "pro" in QUOTA_TIERS
+
+    def test_pro_tier_same_as_premium(self):
+        """pro 套餐配额与 premium 完全一致"""
+        from packages.domain.quota import QUOTA_TIERS
+
+        pro = QUOTA_TIERS["pro"]
+        premium = QUOTA_TIERS["premium"]
+        assert pro.limits == premium.limits
+
+    def test_pro_tier_get_limit(self):
+        """pro 套餐各维度配额正确"""
+        reg = QuotaRegistry()
+        assert reg.get_limit("pro", "storage_gb") == 100
+        assert reg.get_limit("pro", "videos_per_month") == 100
+        assert reg.get_limit("pro", "max_concurrent") == 20
+        assert reg.get_limit("pro", "max_titles") == 500
+        assert reg.get_limit("pro", "ai_voice_enabled") == 1
@@ -185,9 +185,10 @@ class TestQuotaRegistry:
        reg = QuotaRegistry()
        tiers = reg.list_tiers()
        assert "free" in tiers
+        assert "pro" in tiers
        assert "basic" in tiers
        assert "premium" in tiers
-        assert len(tiers) == 3
+        assert len(tiers) == 4

    def test_get_tier_existing(self):
        reg = QuotaRegistry()
@@ -976,7 +976,7 @@ class TestAudioMixing:
        assert UnifiedRenderService._clip_effective_duration(clip) == 0.0

    def test_mix_audio_single_main_clip(self):
-        """单主clip时直接提取音频。"""
+        """只有 main clip（无独立音频轨）→ 源视频音频被丢弃，返回 None。"""
        clips = [_make_clip("c1", "main", order=0, duration=5.0)]
        asset_paths = {"asset_c1.mp4": Path("/tmp/asset_c1.mp4")}
        svc = _make_service(clips, asset_paths)
@@ -990,23 +990,21 @@ class TestAudioMixing:
            ctx = _make_ctx()
            result = mix_audio(ctx, layers, 5.0)

-        assert result is not None
-        assert result.name == "audio_plan_001.aac"
-        mock_run.assert_called_once()
-        # 验证命令包含 -vn（无视频）和 aac 编码
-        cmd = mock_run.call_args[0][0]
-        assert "-vn" in cmd
-        assert "aac" in cmd
+        # 源视频音频被丢弃，没有独立音频轨 → 无音频
+        assert result is None
+        mock_run.assert_not_called()

    def test_mix_audio_multi_main_clips(self):
-        """多主clip时用concat拼接音频。"""
+        """多个独立音频轨用 concat 拼接（main 图层音频被丢弃）。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=3.0),
-            _make_clip("c2", "main", order=1, duration=2.0),
+            _make_clip("c1", "main", order=0, duration=3.0),  # 源视频音频被丢弃
+            _make_clip("tts1", "main", order=0, duration=3.0, config={"role": "audio"}),
+            _make_clip("tts2", "main", order=1, duration=2.0, config={"role": "audio"}),
        ]
        asset_paths = {
            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
-            "asset_c2.mp4": Path("/tmp/asset_c2.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+            "asset_tts2.mp4": Path("/tmp/asset_tts2.mp4"),
        }
        svc = _make_service(clips, asset_paths)

@@ -1022,15 +1020,16 @@ class TestAudioMixing:
        assert result is not None
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
-        # 验证有 filter_complex 和 concat
-        assert "-filter_complex" in cmd
        cmd_str = " ".join(cmd)
+        # 2 个独立音频轨 concat
        assert "concat=n=2:v=0:a=1" in cmd_str
+        # main 的 c1 不参与音频（源视频杂音被丢弃）
+        assert "asset_c1.mp4" not in cmd_str

    def test_mix_audio_with_independent_audio_track(self):
-        """有独立音频轨时用amix混音。"""
+        """独立音频轨生效；main 图层源视频音频被丢弃。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=5.0),
+            _make_clip("c1", "main", order=0, duration=5.0),  # 源视频音频被丢弃
            _make_clip(
                "bgm1",
                "main",
@@ -1055,11 +1054,52 @@ class TestAudioMixing:
            result = mix_audio(ctx, layers, 5.0)

        assert result is not None
+        mock_run.assert_called_once()
+        cmd = mock_run.call_args[0][0]
+        cmd_str = " ".join(cmd)
+        # main 的源视频 c1 不参与音频
+        assert "asset_c1.mp4" not in cmd_str
+        # 独立音频轨 bgm1 作为最终音频生效
+        assert "asset_bgm1.mp4" in cmd_str
+
+    def test_mix_with_independent_audio_amix(self):
+        """直接验证 mix_with_independent_audio：main 音频 + 独立音频轨走 amix 混音。"""
+        from video_processing.render_audio import mix_with_independent_audio
+        from video_processing.unified_render_service import ResolvedClip
+
+        main = ResolvedClip(
+            clip_id="c1",
+            asset_id="asset_c1",
+            clip_type="main",
+            order=0,
+            local_path=Path("/tmp/asset_c1.mp4"),
+            duration=5.0,
+            actual_duration=5.0,
+            config={},
+        )
+        aud = ResolvedClip(
+            clip_id="bgm1",
+            asset_id="asset_bgm1",
+            clip_type="main",
+            order=1,
+            local_path=Path("/tmp/asset_bgm1.mp4"),
+            duration=5.0,
+            actual_duration=5.0,
+            config={"volume": 0.5},
+        )
+        ctx = _make_ctx()
+        output_path = ctx.work_dir / "audio_plan_001.aac"
+
+        with patch("video_processing.render_audio.run_ffmpeg") as mock_run:
+            mix_with_independent_audio(ctx, [main], [aud], output_path, 5.0)
+
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
        cmd_str = " ".join(cmd)
        assert "amix" in cmd_str
        assert "volume=0.5" in cmd_str
+        assert "asset_c1.mp4" in cmd_str
+        assert "asset_bgm1.mp4" in cmd_str

    def test_mix_audio_no_audio_returns_none(self):
        """没有音频素材时返回None。"""
@@ -1082,14 +1122,16 @@ class TestAudioMixing:
        assert result is None

    def test_mix_audio_background_not_used_as_main(self):
-        """background 图层不参与主音频，main 优先级更高。"""
+        """main/background 图层音频都被丢弃，只有独立音频轨参与混音。"""
        clips = [
            _make_clip("bg1", "background", order=0, duration=5.0),
            _make_clip("c1", "main", order=0, duration=5.0),
+            _make_clip("tts1", "main", order=0, duration=5.0, config={"role": "audio"}),
        ]
        asset_paths = {
            "asset_bg1.mp4": Path("/tmp/asset_bg1.mp4"),
            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
        }
        svc = _make_service(clips, asset_paths)

@@ -1105,21 +1147,24 @@ class TestAudioMixing:
        assert result is not None
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
-        # 验证主音频源是 main 的 c1，不是 background 的 bg1
-        # 单 main clip 走直接提取路径，输入文件应该只有 c1
        cmd_str = " ".join(cmd)
-        assert "asset_c1.mp4" in cmd_str
+        # 源视频（background + main）音频都被丢弃
        assert "asset_bg1.mp4" not in cmd_str
+        assert "asset_c1.mp4" not in cmd_str
+        # 只有独立音频轨
+        assert "asset_tts1.mp4" in cmd_str

    def test_mix_audio_main_priority_over_broll(self):
-        """main 图层优先级高于 broll。"""
+        """main/broll 图层的源视频音频都被丢弃，只使用独立音频轨。"""
        clips = [
            _make_clip("b1", "b_roll", order=0, duration=5.0),
            _make_clip("c1", "main", order=0, duration=5.0),
+            _make_clip("tts1", "main", order=0, duration=5.0, config={"role": "audio"}),
        ]
        asset_paths = {
            "asset_b1.mp4": Path("/tmp/asset_b1.mp4"),
            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
        }
        svc = _make_service(clips, asset_paths)

@@ -1136,12 +1181,13 @@ class TestAudioMixing:
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
        cmd_str = " ".join(cmd)
-        # 主音频源应该是 main 的 c1，不是 broll 的 b1
-        assert "asset_c1.mp4" in cmd_str
+        # main 和 broll 的源视频音频都被丢弃
+        assert "asset_c1.mp4" not in cmd_str
        assert "asset_b1.mp4" not in cmd_str
+        assert "asset_tts1.mp4" in cmd_str

    def test_mix_audio_broll_used_when_no_main(self):
-        """没有 main 时，broll 作为主音频源。"""
+        """broll 图层源视频音频也被丢弃；无独立音频轨 → 返回 None。"""
        clips = [_make_clip("b1", "b_roll", order=0, duration=5.0)]
        asset_paths = {"asset_b1.mp4": Path("/tmp/asset_b1.mp4")}
        svc = _make_service(clips, asset_paths)
@@ -1155,16 +1201,20 @@ class TestAudioMixing:
            ctx = _make_ctx()
            result = mix_audio(ctx, layers, 5.0)

-        assert result is not None
-        mock_run.assert_called_once()
-        cmd = mock_run.call_args[0][0]
-        assert "-vn" in cmd
-        assert "asset_b1.mp4" in " ".join(cmd)
+        # 源视频音频被丢弃，无独立音频轨 → 无音频
+        assert result is None
+        mock_run.assert_not_called()

    def test_mix_audio_single_clip_truncated_to_video_duration(self):
-        """单clip音频截断到 video_duration（video_duration < clip有效时长）。"""
-        clips = [_make_clip("c1", "main", order=0, duration=10.0)]
-        asset_paths = {"asset_c1.mp4": Path("/tmp/asset_c1.mp4")}
+        """单独立音频轨截断到 video_duration（video_duration < clip有效时长）。"""
+        clips = [
+            _make_clip("c1", "main", order=0, duration=10.0),  # 源视频音频被丢弃
+            _make_clip("tts1", "main", order=0, duration=10.0, config={"role": "audio"}),
+        ]
+        asset_paths = {
+            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+        }
        svc = _make_service(clips, asset_paths)

        with (
@@ -1174,17 +1224,17 @@ class TestAudioMixing:
        ):
            layers = svc._group_clips_into_layers(svc._resolve_clips())
            ctx = _make_ctx()
-            # video_duration 只有 3.0，小于 clip 的 10.0
+            # video_duration 只有 3.0，小于独立音频轨的 10.0
            result = mix_audio(ctx, layers, 3.0)

        assert result is not None
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
-        # 验证 -t 参数是 3.0 不是 10.0
-        t_index = cmd.index("-t")
-        assert t_index >= 0
-        t_value = float(cmd[t_index + 1])
-        assert t_value == 3.0
+        # 验证截断到 3.0（-t 3.0 或 atrim=0:3.000）
+        cmd_str = " ".join(cmd)
+        assert "3.000" in cmd_str or "3.0" in cmd_str
+        # 源视频不参与
+        assert "asset_c1.mp4" not in cmd_str

    def test_merge_audio_video(self):
        """合并音视频命令正确。"""
@@ -1341,25 +1391,22 @@ class TestAudioMixing:
        mock_run.assert_not_called()

    def test_mix_audio_partial_clips_no_audio_filtered(self):
-        """部分主图层clip无音频流时，过滤掉无音轨的，剩余有音频的正常concat。"""
+        """main 图层音频全部丢弃；独立音频轨有/无音频时按预期过滤。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=3.0),  # 无音频
-            _make_clip("c2", "main", order=1, duration=2.0),  # 有音频
+            _make_clip("c1", "main", order=0, duration=3.0),  # 源视频音频被丢弃
+            _make_clip("c2", "main", order=1, duration=2.0),  # 源视频音频被丢弃
+            _make_clip("tts1", "main", order=0, duration=2.0, config={"role": "audio"}),  # 独立音频轨
        ]
        asset_paths = {
            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
            "asset_c2.mp4": Path("/tmp/asset_c2.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
        }
        svc = _make_service(clips, asset_paths)

-        # 模拟：c1 无音频，c2 有音频
-        def fake_has_audio(path):
-            return "c2" in str(path)
-
        with (
            _patch_path_exists(),
            patch("video_processing.unified_render_service.probe_duration", return_value=5.0),
-            patch("video_processing.render_audio.probe_has_audio", side_effect=fake_has_audio),
            patch("video_processing.render_audio.run_ffmpeg") as mock_run,
        ):
            layers = svc._group_clips_into_layers(svc._resolve_clips())
@@ -1370,21 +1417,23 @@ class TestAudioMixing:
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
        cmd_str = " ".join(cmd)
-        # 只剩 1 个有效音频 clip，走单clip路径（-vn），不走 filter_complex concat
-        assert "-vn" in cmd
-        assert "concat=n=2" not in cmd_str
+        # 只有独立音频轨 tts1 参与
+        assert "asset_tts1.mp4" in cmd_str
+        # main 的 c1/c2 源视频音频被丢弃
+        assert "asset_c1.mp4" not in cmd_str
+        assert "asset_c2.mp4" not in cmd_str

    def test_mix_audio_all_main_no_audio_but_independent_track(self):
-        """主图层全部无音频，但有独立音频轨时，正常走amix混音。"""
+        """main 图层源视频音频全部丢弃；仅独立音频轨生效，走 concat 单轨路径。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=5.0),  # 无音频
+            _make_clip("c1", "main", order=0, duration=5.0),  # 源视频音频被丢弃
            _make_clip(
                "bgm1",
                "main",
                order=0,
                duration=5.0,
                config={"role": "audio", "volume": 0.5},
-            ),  # 独立音频轨（有音频）
+            ),  # 独立音频轨
        ]
        asset_paths = {
            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
@@ -1392,13 +1441,9 @@ class TestAudioMixing:
        }
        svc = _make_service(clips, asset_paths)

-        def fake_has_audio(path):
-            return "bgm" in str(path)
-
        with (
            _patch_path_exists(),
            patch("video_processing.unified_render_service.probe_duration", return_value=5.0),
-            patch("video_processing.render_audio.probe_has_audio", side_effect=fake_has_audio),
            patch("video_processing.render_audio.run_ffmpeg") as mock_run,
        ):
            layers = svc._group_clips_into_layers(svc._resolve_clips())
@@ -1409,8 +1454,10 @@ class TestAudioMixing:
        mock_run.assert_called_once()
        cmd = mock_run.call_args[0][0]
        cmd_str = " ".join(cmd)
-        # 只有独立音频轨参与混音，amix 输入数=1
-        assert "amix=inputs=1" in cmd_str
+        # main 的源视频 c1 不参与音频
+        assert "asset_c1.mp4" not in cmd_str
+        # 只有独立音频轨 bgm1 作为主音频走单轨拼接
+        assert "asset_bgm1.mp4" in cmd_str

    def test_mix_audio_both_no_audio_returns_none(self):
        """主图层和独立音频轨都无音频时，返回None。"""
@@ -2069,14 +2116,16 @@ class TestConcatNormalizeAudioFormat:
    """

    def test_multi_clip_concat_has_aformat(self):
-        """多 clip 音频 concat 前，每个 clip 都有 aformat 归一化。"""
+        """多独立音频轨 concat 前，每个轨都有 aformat 归一化（main 图层源视频音频被丢弃）。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=3.0),
-            _make_clip("c2", "main", order=1, duration=2.0),
+            _make_clip("c1", "main", order=0, duration=3.0),  # 源视频音频被丢弃
+            _make_clip("tts1", "main", order=0, duration=3.0, config={"role": "audio"}),
+            _make_clip("tts2", "main", order=1, duration=2.0, config={"role": "audio"}),
        ]
        asset_paths = {
            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
-            "asset_c2.mp4": Path("/tmp/asset_c2.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+            "asset_tts2.mp4": Path("/tmp/asset_tts2.mp4"),
        }
        svc = _make_service(clips, asset_paths)

@@ -2099,22 +2148,24 @@ class TestConcatNormalizeAudioFormat:
        assert "channel_layouts=stereo" in cmd_str, "声道应统一为 stereo"
        assert "sample_fmts=fltp" in cmd_str, "采样格式应统一为 fltp"

-        # 两个 clip 都应该有 aformat
+        # 2 个独立音频轨都应有 aformat
        aformat_count = cmd_str.count("aformat=")
-        assert aformat_count >= 2, f"每个 clip 都应有 aformat，实际 {aformat_count} 个"
+        assert aformat_count >= 2, f"每个独立音频轨都应有 aformat，实际 {aformat_count} 个"

        # 有 concat
        assert "concat=n=2:v=0:a=1" in cmd_str
+        # main 源视频不参与
+        assert "asset_c1.mp4" not in cmd_str

    def test_aformat_before_concat(self):
-        """aformat 应在 concat 之前（每个 clip 处理链中 aformat 在 concat 之前）。"""
+        """aformat 应在 concat 之前（每个独立音频轨处理链中 aformat 在 concat 之前）。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=3.0),
-            _make_clip("c2", "main", order=1, duration=2.0),
+            _make_clip("tts1", "main", order=0, duration=3.0, config={"role": "audio"}),
+            _make_clip("tts2", "main", order=1, duration=2.0, config={"role": "audio"}),
        ]
        asset_paths = {
-            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
-            "asset_c2.mp4": Path("/tmp/asset_c2.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+            "asset_tts2.mp4": Path("/tmp/asset_tts2.mp4"),
        }
        svc = _make_service(clips, asset_paths)

@@ -2131,19 +2182,23 @@ class TestConcatNormalizeAudioFormat:
        fc_idx = cmd.index("-filter_complex")
        fc_str = cmd[fc_idx + 1]

-        # 找到 concat 的位置
        concat_pos = fc_str.find("concat=")
        assert concat_pos > 0, "应找到 concat filter"

-        # 在 concat 之前的部分应该有 aformat
        before_concat = fc_str[:concat_pos]
        aformat_before_count = before_concat.count("aformat=")
-        assert aformat_before_count >= 2, f"concat 之前每个 clip 都应有 aformat，实际 {aformat_before_count} 个"
+        assert aformat_before_count >= 2, f"concat 之前每个独立音频轨都应有 aformat，实际 {aformat_before_count} 个"

    def test_single_clip_audio_has_normalized_output(self):
-        """单 clip 音频输出也应统一格式（一致性保障）。"""
-        clips = [_make_clip("c1", "main", order=0, duration=5.0)]
-        asset_paths = {"asset_c1.mp4": Path("/tmp/asset_c1.mp4")}
+        """单独立音频轨输出也应统一格式（一致性保障）。"""
+        clips = [
+            _make_clip("c1", "main", order=0, duration=5.0),  # 源视频音频被丢弃
+            _make_clip("tts1", "main", order=0, duration=5.0, config={"role": "audio"}),
+        ]
+        asset_paths = {
+            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+        }
        svc = _make_service(clips, asset_paths)

        with (
@@ -2158,18 +2213,20 @@ class TestConcatNormalizeAudioFormat:
        assert mock_run.called
        cmd = mock_run.call_args[0][0]

-        # 单 clip 简单路径应该有 -ar 48000 和 -ac 2
-        assert "-ar" in cmd, "单 clip 音频应指定采样率"
+        # 单独立音频轨简单路径应有 -ar 48000 和 -ac 2
+        assert "-ar" in cmd, "单独立音频轨应指定采样率"
        ar_idx = cmd.index("-ar")
        assert cmd[ar_idx + 1] == "48000", "采样率应为 48000"
-        assert "-ac" in cmd, "单 clip 音频应指定声道数"
+        assert "-ac" in cmd, "单独立音频轨应指定声道数"
        ac_idx = cmd.index("-ac")
        assert cmd[ac_idx + 1] == "2", "声道数应为 2（stereo）"
+        cmd_str = " ".join(cmd)
+        assert "asset_c1.mp4" not in cmd_str

    def test_independent_audio_track_has_aformat(self):
-        """独立音频轨在 amix 前也有 aformat 归一化。"""
+        """独立音频轨输出也应统一格式（48000Hz + stereo + aac）。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=5.0),
+            _make_clip("c1", "main", order=0, duration=5.0),  # 源视频音频被丢弃
            _make_clip(
                "audio1",
                "main",
@@ -2197,12 +2254,16 @@ class TestConcatNormalizeAudioFormat:
        cmd = mock_run.call_args[0][0]
        cmd_str = " ".join(cmd)

-        # 应该有 aformat
-        assert "aformat=" in cmd_str
-        assert "sample_rates=48000" in cmd_str
-        assert "channel_layouts=stereo" in cmd_str
-        # amix 应该存在
-        assert "amix" in cmd_str
+        # 单独立音频轨走 concat 单轨简单路径：-ar 48000 -ac 2
+        assert "-ar" in cmd
+        ar_idx = cmd.index("-ar")
+        assert cmd[ar_idx + 1] == "48000"
+        assert "-ac" in cmd
+        ac_idx = cmd.index("-ac")
+        assert cmd[ac_idx + 1] == "2"
+        # main 源视频 c1 不参与音频
+        assert "asset_c1.mp4" not in cmd_str
+        assert "asset_audio1.mp4" in cmd_str


 class TestConcatNormalizeAudioCodec:
@@ -2212,14 +2273,14 @@ class TestConcatNormalizeAudioCodec:
    """

    def test_multi_clip_output_is_aac(self):
-        """多 clip concat 后输出编码为 aac。"""
+        """多独立音频轨 concat 后输出编码为 aac。"""
        clips = [
-            _make_clip("c1", "main", order=0, duration=3.0),
-            _make_clip("c2", "main", order=1, duration=2.0),
+            _make_clip("tts1", "main", order=0, duration=3.0, config={"role": "audio"}),
+            _make_clip("tts2", "main", order=1, duration=2.0, config={"role": "audio"}),
        ]
        asset_paths = {
-            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
-            "asset_c2.mp4": Path("/tmp/asset_c2.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+            "asset_tts2.mp4": Path("/tmp/asset_tts2.mp4"),
        }
        svc = _make_service(clips, asset_paths)

@@ -2238,9 +2299,15 @@ class TestConcatNormalizeAudioCodec:
        assert "-b:a" in cmd, "应指定音频码率"

    def test_single_clip_output_is_aac(self):
-        """单 clip 音频输出编码为 aac。"""
-        clips = [_make_clip("c1", "main", order=0, duration=5.0)]
-        asset_paths = {"asset_c1.mp4": Path("/tmp/asset_c1.mp4")}
+        """单独立音频轨输出编码为 aac。"""
+        clips = [
+            _make_clip("c1", "main", order=0, duration=5.0),  # 源视频音频被丢弃
+            _make_clip("tts1", "main", order=0, duration=5.0, config={"role": "audio"}),
+        ]
+        asset_paths = {
+            "asset_c1.mp4": Path("/tmp/asset_c1.mp4"),
+            "asset_tts1.mp4": Path("/tmp/asset_tts1.mp4"),
+        }
        svc = _make_service(clips, asset_paths)

        with (
@@ -2255,6 +2322,7 @@ class TestConcatNormalizeAudioCodec:
        assert mock_run.called
        cmd = mock_run.call_args[0][0]
        assert "aac" in cmd
+        assert "asset_c1.mp4" not in " ".join(cmd)


 class TestConcatNormalizeFourItemsComplete:
@@ -2304,9 +2372,14 @@ class TestConcatNormalizeFourItemsComplete:
        assert fps_count >= 3, f"3个 clip 都应有 fps=30，实际 {fps_count} 个"

    def test_audio_format_normalized(self):
-        """[3/4] 音频格式：所有 clip concat 前都有 aformat 归一化。"""
-        clips = self._make_one_take_clips()
+        """[3/4] 音频格式：3 个独立音频轨 concat 前都有 aformat 归一化（main 图层源视频音频被丢弃）。"""
+        clips = self._make_one_take_clips() + [
+            _make_clip("tts1", "main", order=10, duration=5.0, config={"role": "audio"}),
+            _make_clip("tts2", "main", order=11, duration=4.0, config={"role": "audio"}),
+            _make_clip("tts3", "main", order=12, duration=3.0, config={"role": "audio"}),
+        ]
        asset_paths = {f"asset_c{i}.mp4": Path(f"/tmp/asset_c{i}.mp4") for i in range(1, 4)}
+        asset_paths.update({f"asset_tts{i}.mp4": Path(f"/tmp/asset_tts{i}.mp4") for i in range(1, 4)})
        svc = _make_service(clips, asset_paths)

        with (
@@ -2323,15 +2396,23 @@ class TestConcatNormalizeFourItemsComplete:
        cmd_str = " ".join(cmd)

        aformat_count = cmd_str.count("aformat=")
-        assert aformat_count >= 3, f"3个 clip 都应有 aformat，实际 {aformat_count} 个"
+        assert aformat_count >= 3, f"3个独立音频轨都应有 aformat，实际 {aformat_count} 个"
        assert "sample_rates=48000" in cmd_str
        assert "channel_layouts=stereo" in cmd_str
        assert "sample_fmts=fltp" in cmd_str
+        # main 图层源视频不参与
+        for i in range(1, 4):
+            assert f"asset_c{i}.mp4" not in cmd_str

    def test_audio_codec_aac(self):
-        """[4/4] 音频编码：输出为 aac。"""
-        clips = self._make_one_take_clips()
+        """[4/4] 音频编码：输出为 aac（仅独立音频轨参与）。"""
+        clips = self._make_one_take_clips() + [
+            _make_clip("tts1", "main", order=10, duration=5.0, config={"role": "audio"}),
+            _make_clip("tts2", "main", order=11, duration=4.0, config={"role": "audio"}),
+            _make_clip("tts3", "main", order=12, duration=3.0, config={"role": "audio"}),
+        ]
        asset_paths = {f"asset_c{i}.mp4": Path(f"/tmp/asset_c{i}.mp4") for i in range(1, 4)}
+        asset_paths.update({f"asset_tts{i}.mp4": Path(f"/tmp/asset_tts{i}.mp4") for i in range(1, 4)})
        svc = _make_service(clips, asset_paths)

        with (