Merge branch 'master' into 'main'

Master See merge request root/sionrui!3
2026-02-03 18:31:39 +00:00
parent fc52be834b 7edce5494d
commit 998547ff29
33 changed files with 1981 additions and 1910 deletions
--- a/.claude/settings.local.json
+++ b/.claude/settings.local.json
@@ -82,7 +82,8 @@
      "Skill(pptx:*)",
      "Bash(pdftoppm:*)",
      "Bash(pip install:*)",
-      "Bash(where:*)"
+      "Bash(where:*)",
+      "mcp__web-reader__webReader"
    ],
    "deny": [],
    "ask": []
--- a/frontend/app/web-gold/package.json
+++ b/frontend/app/web-gold/package.json
@@ -23,6 +23,7 @@
    "@tailwindcss/vite": "^4.1.14",
    "ai": "^6.0.39",
    "ant-design-vue": "^4.2.6",
+    "aplayer": "^1.10.1",
    "dayjs": "^1.11.18",
    "markdown-it": "^14.1.0",
    "path-to-regexp": "^6.3.0",
--- a/frontend/app/web-gold/src/api/kling.js
+++ b/frontend/app/web-gold/src/api/kling.js
@@ -2,18 +2,66 @@
 * 可灵数字人 API
 */
 import request from './http'
-import { message } from "ant-design-vue"
 import { MaterialService } from './material'

-/**
- * 显示加载提示
- */
-const showLoading = (text) => message.loading(text, 0)
+// ========== 辅助函数 ==========

 /**
- * 销毁加载提示
+ * 从视频中提取封面（可选操作）
 */
-const hideLoading = () => message.destroy()
+async function extractVideoCoverOptional(file) {
+  try {
+    const { extractVideoCover } = await import('@/utils/video-cover')
+    const cover = await extractVideoCover(file, {
+      maxWidth: 800,
+      quality: 0.8
+    })
+    return cover.base64
+  } catch {
+    return null
+  }
+}
+
+/**
+ * 执行人脸识别并返回结果
+ */
+async function performFaceIdentification(videoUrl) {
+  const identifyRes = await identifyFace({ video_url: videoUrl })
+  if (identifyRes.code !== 0) {
+    throw new Error(identifyRes.msg || '识别失败')
+  }
+
+  const faceData = identifyRes.data.data?.face_data?.[0]
+  return {
+    sessionId: identifyRes.data.sessionId,
+    faceId: faceData?.face_id || null,
+    startTime: faceData?.start_time || 0,
+    endTime: faceData?.end_time || 0
+  }
+}
+
+/**
+ * 构建标准响应格式
+ */
+function buildIdentifyResponse(fileId, videoUrl, identifyData, isUploadedFile = false) {
+  return {
+    success: true,
+    data: {
+      fileId,
+      videoUrl,
+      sessionId: identifyData.sessionId,
+      faceId: identifyData.faceId,
+      startTime: isUploadedFile
+        ? Math.round(identifyData.startTime * 1000)
+        : identifyData.startTime,
+      endTime: isUploadedFile
+        ? Math.round(identifyData.endTime * 1000)
+        : identifyData.endTime
+    }
+  }
+}
+
+// ========== API 方法 ==========

 export function identifyFace(data) {
  return request({
@@ -38,93 +86,46 @@ export function getLipSyncTask(taskId) {
  })
 }

+/**
+ * 识别已上传的视频
+ */
 export async function identifyUploadedVideo(videoFile) {
  try {
-    showLoading('正在识别视频中的人脸...')
-    const identifyRes = await identifyFace({ video_url: videoFile.fileUrl })
-    hideLoading()
-
-    if (identifyRes.code !== 0) {
-      throw new Error(identifyRes.msg || '识别失败')
+    const urlRes = await MaterialService.getVideoPlayUrl(videoFile.fileId)
+    if (urlRes.code !== 0 || !urlRes.data) {
+      throw new Error(urlRes.msg || '获取播放链接失败')
    }

-    return {
-      success: true,
-      data: {
-        fileId: videoFile.id,
-        videoUrl: videoFile.fileUrl,
-        sessionId: identifyRes.data.sessionId,
-        faceId: identifyRes.data.data.face_data[0].face_id || null,
-        startTime: identifyRes.data.data.face_data[0].start_time || 0,
-        endTime: identifyRes.data.data.face_data[0].end_time || 0
-      }
-    }
+    const identifyData = await performFaceIdentification(urlRes.data)
+    return buildIdentifyResponse(videoFile.id, urlRes.data, identifyData, false)
  } catch (error) {
-    hideLoading()
    throw error
  }
 }

+/**
+ * 上传视频并识别
+ */
 export async function uploadAndIdentifyVideo(file) {
+  const coverBase64 = await extractVideoCoverOptional(file)
+
  try {
-    showLoading('正在提取视频封面...')
-    let coverBase64 = null
-    try {
-      const { extractVideoCover } = await import('@/utils/video-cover')
-      const cover = await extractVideoCover(file, {
-        maxWidth: 800,
-        quality: 0.8
-      })
-      coverBase64 = cover.base64
-    } catch (coverError) {
-      // 封面提取失败不影响主流程
-    }
-    hideLoading()
-
-    showLoading('正在上传视频...')
-
-    // 使用useUpload Hook（注意：这里需要在组件中使用，这里先用MaterialService）
-    // TODO: 在组件中集成useUpload Hook
    const uploadRes = await MaterialService.uploadFile(file, 'video', coverBase64, null, null)
-    hideLoading()
-
    if (uploadRes.code !== 0) {
      throw new Error(uploadRes.msg || '上传失败')
    }

    const fileId = uploadRes.data

-    showLoading('正在生成播放链接...')
    const urlRes = await MaterialService.getVideoPlayUrl(fileId)
-    hideLoading()
-
    if (urlRes.code !== 0) {
      throw new Error(urlRes.msg || '获取播放链接失败')
    }

-    const videoUrl = urlRes.data
-
-    showLoading('正在识别视频中的人脸...')
-    const identifyRes = await identifyFace({ video_url: videoUrl })
-    hideLoading()
-
-    if (identifyRes.code !== 0) {
-      throw new Error(identifyRes.msg || '识别失败')
-    }
-
-    return {
-      success: true,
-      data: {
-        fileId,
-        videoUrl,
-        sessionId: identifyRes.data.sessionId,
-        faceId: identifyRes.data.data.face_data[0].face_id || null,
-        startTime: identifyRes.data.data.face_data[0].start_time || 0,
-        endTime: identifyRes.data.data.face_data[0].end_time || 0
-      }
-    }
+    const identifyData = await performFaceIdentification(urlRes.data)
+    return buildIdentifyResponse(fileId, urlRes.data, identifyData, true)
  } catch (error) {
-    hideLoading()
    throw error
  }
 }
+
--- a/frontend/app/web-gold/src/api/voice.js
+++ b/frontend/app/web-gold/src/api/voice.js
@@ -22,6 +22,7 @@ export const VoiceService = {
   * @param {string} data.language - 语言（可选）
   * @param {string} data.gender - 音色类型（可选）
   * @param {string} data.note - 备注（可选）
+   * @param {string} data.text - 音频文本（用于语音复刻，前端通过音频识别获取）
   * @param {string} data.providerType - 供应商类型（可选）：cosyvoice-阿里云，siliconflow-硅基流动
   * @returns {Promise}
   */
--- a/frontend/app/web-gold/src/components/PipelineProgress.vue
+++ b/frontend/app/web-gold/src/components/PipelineProgress.vue
@@ -0,0 +1,112 @@
+<template>
+  <div class="pipeline-progress">
+    <!-- 状态和进度 -->
+    <div class="progress-header">
+      <span class="state-text">{{ stateLabel }}</span>
+      <span v-if="stateDescription" class="state-desc">{{ stateDescription }}</span>
+    </div>
+
+    <a-progress
+      :percent="displayProgress"
+      :status="progressStatus"
+      :stroke-color="progressColor"
+    />
+
+    <!-- 错误时显示 -->
+    <div v-if="isFailed && error" class="error-section">
+      <span class="error-text">{{ error }}</span>
+      <a-button size="small" @click="handleRetry">重试</a-button>
+    </div>
+  </div>
+</template>
+
+<script setup lang="ts">
+import { computed } from 'vue'
+import type { PipelineState } from '@/views/kling/hooks/pipeline/types'
+import { STATE_CONFIG } from '@/views/kling/hooks/pipeline/states'
+
+interface Props {
+  state: PipelineState | string
+  progress: number
+  isBusy: boolean
+  isReady: boolean
+  isFailed: boolean
+  isCompleted: boolean
+  error: string | null
+}
+
+const props = defineProps<Props>()
+
+const emit = defineEmits<{
+  retry: []
+  reset: []
+}>()
+
+const stateConfig = computed(() => STATE_CONFIG[props.state as PipelineState])
+const stateLabel = computed(() => stateConfig.value.label)
+const stateDescription = computed(() => stateConfig.value.description)
+
+const progressStatus = computed(() => {
+  if (props.isFailed) return 'exception'
+  if (props.isCompleted) return 'success'
+  return 'active'
+})
+
+const progressColor = computed(() => {
+  if (props.isFailed) return '#ff4d4f'
+  if (props.isCompleted) return '#52c41a'
+  return '#1890ff'
+})
+
+const displayProgress = computed(() => {
+  if (props.isFailed) return 0
+  return props.progress
+})
+
+function handleRetry() {
+  emit('retry')
+}
+</script>
+
+<style scoped lang="less">
+.pipeline-progress {
+  padding: 12px 16px;
+  background: var(--bg-primary);
+  border-radius: 8px;
+  border: 1px solid var(--border-light);
+  margin-bottom: 16px;
+}
+
+.progress-header {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  margin-bottom: 8px;
+  font-size: 13px;
+
+  .state-text {
+    font-weight: 600;
+    color: var(--text-primary);
+  }
+
+  .state-desc {
+    color: var(--text-secondary);
+  }
+}
+
+.error-section {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  margin-top: 12px;
+  padding: 8px 12px;
+  background: rgba(255, 77, 79, 0.1);
+  border-radius: 6px;
+
+  .error-text {
+    flex: 1;
+    color: #ff4d4f;
+    font-size: 13px;
+  }
+}
+</style>
--- a/frontend/app/web-gold/src/components/VoiceSelector.vue
+++ b/frontend/app/web-gold/src/components/VoiceSelector.vue
@@ -1,6 +1,6 @@
 <template>
  <div class="voice-selector">
-    <div v-if="displayedVoices.length === 0" class="empty-voices">
+    <div v-if="userVoiceCards.length === 0" class="empty-voices">
      还没有配音，可先在"配音管理"中上传
    </div>

@@ -19,50 +19,110 @@
        size="small"
        :disabled="!selectedVoiceId"
        :loading="previewLoadingVoiceId === selectedVoiceId"
-        @click="handlePreviewCurrentVoice"
+        @click="handleSynthesize"
      >
        <template #icon>
          <SoundOutlined />
        </template>
-        试听
+        合成
      </a-button>
    </div>
+
+    <!-- APlayer 播放器容器 -->
+    <div v-if="audioUrl" ref="playerContainer" class="aplayer-container"></div>
+
+    <!-- 下载按钮 -->
+    <a-button
+      v-if="audioUrl"
+      type="link"
+      size="small"
+      @click="downloadAudio"
+      class="download-link"
+    >
+      下载音频
+    </a-button>
  </div>
 </template>

 <script setup>
-import { ref, computed, onMounted } from 'vue'
+import { ref, computed, onMounted, onBeforeUnmount, nextTick, watch } from 'vue'
 import { useVoiceCopyStore } from '@/stores/voiceCopy'
 import { useTTS, TTS_PROVIDERS } from '@/composables/useTTS'
+import APlayer from 'aplayer'
+
+const props = defineProps({
+  synthText: {
+    type: String,
+    default: ''
+  },
+  speechRate: {
+    type: Number,
+    default: 1.0
+  }
+})

 const voiceStore = useVoiceCopyStore()

 const emit = defineEmits(['select'])

-// 使用TTS Hook，默认使用Qwen供应商
+let player = null
+const playerContainer = ref(null)
+const audioUrl = ref('')
+const currentVoiceName = ref('')
+
+// 默认封面图片（音频波形图标）
+const defaultCover = `data:image/svg+xml;base64,${btoa(`
+<svg width="100" height="100" viewBox="0 0 100 100" xmlns="http://www.w3.org/2000/svg">
+  <rect width="100" height="100" fill="#1f2937" rx="8"/>
+  <g fill="#60a5fa">
+    <rect x="20" y="35" width="4" height="30" rx="2">
+      <animate attributeName="height" values="30;20;30" dur="0.8s" repeatCount="indefinite"/>
+      <animate attributeName="y" values="35;40;35" dur="0.8s" repeatCount="indefinite"/>
+    </rect>
+    <rect x="30" y="30" width="4" height="40" rx="2">
+      <animate attributeName="height" values="40;25;40" dur="0.6s" repeatCount="indefinite"/>
+      <animate attributeName="y" values="30;37.5;30" dur="0.6s" repeatCount="indefinite"/>
+    </rect>
+    <rect x="40" y="25" width="4" height="50" rx="2">
+      <animate attributeName="height" values="50;30;50" dur="0.7s" repeatCount="indefinite"/>
+      <animate attributeName="y" values="25;35;25" dur="0.7s" repeatCount="indefinite"/>
+    </rect>
+    <rect x="50" y="28" width="4" height="44" rx="2">
+      <animate attributeName="height" values="44;28;44" dur="0.9s" repeatCount="indefinite"/>
+      <animate attributeName="y" values="28;36;28" dur="0.9s" repeatCount="indefinite"/>
+    </rect>
+    <rect x="60" y="32" width="4" height="36" rx="2">
+      <animate attributeName="height" values="36;22;36" dur="0.5s" repeatCount="indefinite"/>
+      <animate attributeName="y" values="32;39;32" dur="0.5s" repeatCount="indefinite"/>
+    </rect>
+    <rect x="70" y="38" width="4" height="24" rx="2">
+      <animate attributeName="height" values="24;15;24" dur="0.7s" repeatCount="indefinite"/>
+      <animate attributeName="y" values="38;42.5;38" dur="0.7s" repeatCount="indefinite"/>
+    </rect>
+  </g>
+  <circle cx="50" cy="50" r="18" fill="none" stroke="#60a5fa" stroke-width="2" opacity="0.3"/>
+  <path d="M44 44 L44 56 L56 50 Z" fill="#60a5fa" opacity="0.5"/>
+</svg>
+`.trim())}`
+
+// 使用TTS Hook
 const {
  previewLoadingVoiceId,
-  playingPreviewVoiceId,
-  ttsText,
-  speechRate,
  playVoiceSample,
  setText,
-  setSpeechRate,
-  resetPreviewState
+  setSpeechRate
 } = useTTS({
  provider: TTS_PROVIDERS.SILICONFLOW
 })

-// 当前选中的音色ID
 const selectedVoiceId = ref('')

-// 从store数据构建音色列表
 const userVoiceCards = computed(() =>
  (voiceStore.profiles || []).map(profile => ({
    id: `user-${profile.id}`,
    rawId: profile.id,
    name: profile.name || '未命名',
-    category:'',
+    category: '',
    gender: profile.gender || 'female',
    description: profile.note || '我的配音',
    fileUrl: profile.fileUrl,
@@ -72,76 +132,139 @@ const userVoiceCards = computed(() =>
  }))
 )

-const displayedVoices = computed(() => userVoiceCards.value)
-
-// 转换为下拉框选项格式
 const voiceOptions = computed(() =>
-  displayedVoices.value.map(voice => ({
+  userVoiceCards.value.map(voice => ({
    value: voice.id,
    label: voice.name,
-    data: voice // 保存完整数据
+    data: voice
  }))
 )

-// 音色选择变化处理
 const handleVoiceChange = (value, option) => {
  const voice = option.data
  selectedVoiceId.value = value
+  currentVoiceName.value = voice.name
  emit('select', voice)
 }

-// 试听当前选中的音色
-const handlePreviewCurrentVoice = () => {
+const handleSynthesize = () => {
  if (!selectedVoiceId.value) return

-  const voice = displayedVoices.value.find(v => v.id === selectedVoiceId.value)
+  const voice = userVoiceCards.value.find(v => v.id === selectedVoiceId.value)
  if (!voice) return

+  currentVoiceName.value = voice.name
  handlePlayVoiceSample(voice)
 }

+// 监听 prop 变化，更新 TTS 参数
+watch(() => props.synthText, (newText) => {
+  setText(newText || '')
+}, { immediate: true })
+
+watch(() => props.speechRate, (newRate) => {
+  setSpeechRate(newRate)
+}, { immediate: true })
+
 /**
- * 处理音色试听
- * 使用Hook提供的playVoiceSample方法
+ * 处理音色
 */
 const handlePlayVoiceSample = (voice) => {
+  currentVoiceName.value = voice.name
  playVoiceSample(
    voice,
-    (audioData) => {
-      // 成功回调
-      console.log('音频播放成功', audioData)
+    (data) => {
+      const url = data.audioUrl || data.objectUrl
+      if (!url) {
+        console.error('无效的音频数据格式', data)
+        return
+      }
+      initPlayer(url)
    },
    (error) => {
-      // 错误回调
      console.error('音频播放失败', error)
    }
  )
 }

 /**
- * 设置要试听的文本（供父组件调用）
- * @param {string} text 要试听的文本
+ * 初始化 APlayer
 */
-const setPreviewText = (text) => {
-  setText(text)
+const initPlayer = (url) => {
+  destroyPlayer()
+  audioUrl.value = url
+
+  nextTick(() => {
+    if (!playerContainer.value) return
+
+    player = new APlayer({
+      container: playerContainer.value,
+      autoplay: true,
+      theme: '#3b82f6',
+      preload: 'auto',
+      volume: 0.7,
+      audio: [{
+        name: currentVoiceName.value || '语音合成',
+        artist: '合成',
+        url: url,
+        cover: defaultCover
+      }]
+    })
+
+    player.on('ended', () => {
+      if (audioUrl.value?.startsWith('blob:')) {
+        URL.revokeObjectURL(audioUrl.value)
+      }
+      audioUrl.value = ''
+    })
+
+    player.on('error', (e) => {
+      console.error('APlayer 播放错误:', e)
+    })
+  })
 }

 /**
- * 设置语速（供父组件调用）
- * @param {number} rate 语速倍率
+ * 下载音频
 */
-const setPreviewSpeechRate = (rate) => {
-  setSpeechRate(rate)
+const downloadAudio = () => {
+  if (!audioUrl.value) return
+
+  const link = document.createElement('a')
+  link.href = audioUrl.value
+  link.download = `${currentVoiceName.value || '语音合成'}.mp3`
+  document.body.appendChild(link)
+  link.click()
+  document.body.removeChild(link)
 }

-defineExpose({
-  setPreviewText,
-  setPreviewSpeechRate
-})
+/**
+ * 销毁播放器
+ */
+const destroyPlayer = () => {
+  if (player) {
+    try {
+      player.destroy()
+    } catch (e) {
+      console.error('销毁播放器失败:', e)
+    }
+    player = null
+  }
+  if (audioUrl.value) {
+    URL.revokeObjectURL(audioUrl.value)
+    audioUrl.value = ''
+  }
+}
+
+defineExpose({})

 onMounted(async () => {
  await voiceStore.refresh()
 })
+
+onBeforeUnmount(() => {
+  destroyPlayer()
+})
 </script>

 <style scoped>
@@ -176,4 +299,16 @@ onMounted(async () => {
  height: 32px;
  white-space: nowrap;
 }
+
+/* APlayer 容器样式 */
+.aplayer-container {
+  margin-top: 12px;
+}
+
+/* 下载链接样式 */
+.download-link {
+  margin-top: 8px;
+  padding: 0;
+  font-size: 12px;
+}
 </style>
--- a/frontend/app/web-gold/src/composables/useTTS.js
+++ b/frontend/app/web-gold/src/composables/useTTS.js
@@ -7,72 +7,97 @@ import { message } from 'ant-design-vue'
 import { VoiceService } from '@/api/voice'
 import { normalizeProviderType, VOICE_PROVIDER_TYPES } from '@/config/voiceConfig'

-// 兼容旧代码的导出
+// ========== 常量 ==========
+
+/** 兼容旧代码的导出 */
 const TTS_PROVIDERS = VOICE_PROVIDER_TYPES

-const DEFAULT_CONFIG =  {
-    apiEndpoint: '/api/tik/voice/tts',
-    audioFormat: 'mp3',
-    supportedFormats: ['mp3', 'wav']
+/** 默认配置 */
+const DEFAULT_CONFIG = {
+  apiEndpoint: '/api/tik/voice/tts',
+  audioFormat: 'mp3',
+  supportedFormats: ['mp3', 'wav']
 }

+/** 最大预览缓存数量 */
+const MAX_PREVIEW_CACHE_SIZE = 50
+
+// ========== 类型定义 ==========
+
+/**
+ * @typedef {Object} AudioData
+ * @property {Blob} blob - 音频 Blob
+ * @property {string} objectUrl - 对象 URL
+ * @property {string} format - 音频格式
+ */
+
 export function useTTS(options = {}) {
  const {
    provider = VOICE_PROVIDER_TYPES.SILICONFLOW,
    customConfig = {}
  } = options

-  // 状态管理
+  // 状态管理（移到函数内部，避免模块级状态污染）
  const previewAudioCache = new Map()
-  const MAX_PREVIEW_CACHE_SIZE = 50
  const previewLoadingVoiceId = ref(null)
  const playingPreviewVoiceId = ref(null)
  const ttsText = ref('')
  const speechRate = ref(1.0)

-  // 音频实例
+  // 音频实例（移到函数内部）
  let previewAudio = null
  let previewObjectUrl = ''

-  // 获取当前供应商配置
-  const getProviderConfig = () => {
+  // ========== 辅助函数 ==========
+
+  function getProviderConfig() {
    return DEFAULT_CONFIG
  }

  /**
   * 播放音频预览
-   * @param {string} url 音频URL
-   * @param {Object} options 播放选项
+   * @param {string} url - 音频 URL
+   * @param {Object} playOptions - 播放选项
+   * @param {boolean} [playOptions.revokeOnEnd=false] - 播放结束后是否释放 URL
+   * @param {Function} [playOptions.onEnded] - 播放结束回调
   */
-  const playAudioPreview = (url, options = {}) => {
-    if (!url) return message.warning('暂无可试听的音频')
+  function playAudioPreview(url, playOptions = {}) {
+    if (!url) {
+      message.warning('暂无可试听的音频')
+      return
+    }

+    // 停止当前播放
    try {
      previewAudio?.pause?.()
      previewAudio = null
-    } catch (error) {
+    } catch {
+      // 忽略停止播放的错误
    }

    const audio = new Audio(url)
-    const cleanup = () => {
-      if (options.revokeOnEnd && url.startsWith('blob:')) {
+
+    function cleanup() {
+      if (playOptions.revokeOnEnd && url.startsWith('blob:')) {
        URL.revokeObjectURL(url)
-        previewObjectUrl === url && (previewObjectUrl = '')
+        if (previewObjectUrl === url) {
+          previewObjectUrl = ''
+        }
      }
      previewAudio = null
-      options.onEnded && options.onEnded()
+      playOptions.onEnded?.()
    }

    audio.play()
-      .then(() => {
+      .then(function() {
        previewAudio = audio
        audio.onended = cleanup
-        audio.onerror = () => {
+        audio.onerror = function() {
          cleanup()
          message.error('播放失败')
        }
      })
-      .catch(err => {
+      .catch(function() {
        cleanup()
        message.error('播放失败')
      })
@@ -80,10 +105,10 @@ export function useTTS(options = {}) {

  /**
   * 生成预览缓存键
-   * @param {Object} voice 音色对象
+   * @param {Object} voice - 音色对象
   * @returns {string} 缓存键
   */
-  const generatePreviewCacheKey = (voice) => {
+  function generatePreviewCacheKey(voice) {
    const voiceId = voice.voiceId || voice.rawId || voice.id
    const text = ttsText.value.trim()
    const rate = speechRate.value
@@ -92,12 +117,12 @@ export function useTTS(options = {}) {

  /**
   * 解码并缓存Base64音频
-   * @param {string} audioBase64 Base64编码的音频数据
-   * @param {string} format 音频格式
-   * @param {string} cacheKey 缓存键
+   * @param {string} audioBase64 - Base64 编码的音频数据
+   * @param {string} [format='mp3'] - 音频格式
+   * @param {string} cacheKey - 缓存键
   * @returns {Promise<Object>} 音频数据
   */
-  const decodeAndCacheBase64 = async (audioBase64, format = 'mp3', cacheKey) => {
+  async function decodeAndCacheBase64(audioBase64, format = 'mp3', cacheKey) {
    const byteCharacters = window.atob(audioBase64)
    const byteNumbers = new Uint8Array(byteCharacters.length)
    for (let i = 0; i < byteCharacters.length; i++) {
@@ -123,10 +148,10 @@ export function useTTS(options = {}) {

  /**
   * 播放缓存的音频
-   * @param {Object} audioData 音频数据
-   * @param {Function} onEnded 播放结束回调
+   * @param {Object} audioData - 音频数据
+   * @param {Function} onEnded - 播放结束回调
   */
-  const playCachedAudio = (audioData, onEnded) => {
+  function playCachedAudio(audioData, onEnded) {
    if (previewObjectUrl && previewObjectUrl !== audioData.objectUrl) {
      URL.revokeObjectURL(previewObjectUrl)
    }
@@ -134,7 +159,10 @@ export function useTTS(options = {}) {

    playAudioPreview(previewObjectUrl, {
      revokeOnEnd: false,
-      onEnded: () => {
+      onEnded: function() {
+        if (audioData.objectUrl?.startsWith('blob:')) {
+          URL.revokeObjectURL(audioData.objectUrl)
+        }
        onEnded && onEnded()
      }
    })
@@ -143,17 +171,17 @@ export function useTTS(options = {}) {
  /**
   * 重置预览状态
   */
-  const resetPreviewState = () => {
+  function resetPreviewState() {
    previewLoadingVoiceId.value = null
    playingPreviewVoiceId.value = null
  }

  /**
   * 提取ID从字符串
-   * @param {string} idStr 包含前缀的ID字符串
+   * @param {string} idStr - 包含前缀的ID字符串
   * @returns {number|null} 提取的ID
   */
-  const extractIdFromString = (idStr) => {
+  function extractIdFromString(idStr) {
    if (typeof idStr !== 'string' || !idStr.startsWith('user-')) return null
    const extractedId = parseInt(idStr.replace('user-', ''))
    return Number.isNaN(extractedId) ? null : extractedId
@@ -161,10 +189,10 @@ export function useTTS(options = {}) {

  /**
   * 构建预览参数
-   * @param {Object} voice 音色对象
+   * @param {Object} voice - 音色对象
   * @returns {Object|null} 预览参数
   */
-  const buildPreviewParams = (voice) => {
+  function buildPreviewParams(voice) {
    const configId = voice.rawId || extractIdFromString(voice.id)
    if (!configId) {
      message.error('配音配置无效')
@@ -184,11 +212,11 @@ export function useTTS(options = {}) {

  /**
   * 播放音色试听
-   * @param {Object} voice 音色对象
-   * @param {Function} onSuccess 成功回调
-   * @param {Function} onError 错误回调
+   * @param {Object} voice - 音色对象
+   * @param {Function} onSuccess - 成功回调
+   * @param {Function} onError - 错误回调
   */
-  const playVoiceSample = async (voice, onSuccess, onError) => {
+  async function playVoiceSample(voice, onSuccess, onError) {
    if (!voice) return
    if (previewLoadingVoiceId.value === voice.id || playingPreviewVoiceId.value === voice.id) {
      return
@@ -197,7 +225,8 @@ export function useTTS(options = {}) {
      try {
        previewAudio?.pause?.()
        previewAudio = null
-      } catch (error) {
+      } catch {
+        // 忽略错误
      }
    }

@@ -225,21 +254,30 @@ export function useTTS(options = {}) {
      if (res.code !== 0) {
        message.error(res.msg || '试听失败')
        resetPreviewState()
-        onError && onError(new Error(res.msg || '试听失败'))
+        onError?.(new Error(res.msg || '试听失败'))
        return
      }

      if (res.data?.audioUrl) {
-        playAudioPreview(res.data.audioUrl, { onEnded: resetPreviewState })
-        onSuccess && onSuccess(res.data)
+        resetPreviewState()
+        playAudioPreview(res.data.audioUrl, {
+          revokeOnEnd: true,
+          onEnded: function() {
+            URL.revokeObjectURL(res.data.audioUrl)
+          }
+        })
+        onSuccess?.(res.data)
      } else if (res.data?.audioBase64) {
        const audioData = await decodeAndCacheBase64(res.data.audioBase64, res.data.format, cacheKey)
-        playCachedAudio(audioData, resetPreviewState)
-        onSuccess && onSuccess(audioData)
+        resetPreviewState()
+        playCachedAudio(audioData, function() {
+          URL.revokeObjectURL(audioData.objectUrl)
+        })
+        onSuccess?.(audioData)
      } else {
        message.error('试听失败')
        resetPreviewState()
-        onError && onError(new Error('未收到音频数据'))
+        onError?.(new Error('未收到音频数据'))
      }
    } catch (error) {
      message.error('试听失败')
@@ -250,10 +288,10 @@ export function useTTS(options = {}) {

  /**
   * TTS文本转语音
-   * @param {Object} params TTS参数
-   * @returns {Promise<Object>} TTS结果
+   * @param {Object} params - TTS 参数
+   * @returns {Promise<Object>} TTS 结果
   */
-  const synthesize = async (params) => {
+  async function synthesize(params) {
    const providerConfig = getProviderConfig()

    const ttsParams = {
@@ -269,25 +307,25 @@ export function useTTS(options = {}) {

  /**
   * 设置文本
-   * @param {string} text 要设置的文本
+   * @param {string} text - 要设置的文本
   */
-  const setText = (text) => {
+  function setText(text) {
    ttsText.value = text
  }

  /**
   * 设置语速
-   * @param {number} rate 语速倍率
+   * @param {number} rate - 语速倍率
   */
-  const setSpeechRate = (rate) => {
+  function setSpeechRate(rate) {
    speechRate.value = rate
  }

  /**
   * 清除音频缓存
   */
-  const clearAudioCache = () => {
-    previewAudioCache.forEach((audioData) => {
+  function clearAudioCache() {
+    previewAudioCache.forEach(function(audioData) {
      URL.revokeObjectURL(audioData.objectUrl)
    })
    previewAudioCache.clear()
@@ -296,14 +334,17 @@ export function useTTS(options = {}) {
  /**
   * 停止当前播放
   */
-  const stopCurrentPlayback = () => {
+  function stopCurrentPlayback() {
    try {
      previewAudio?.pause?.()
      previewAudio = null
-    } catch (error) {
+    } catch {
+      // 忽略错误
    }
  }

+  // ========== 返回接口 ==========
+
  return {
    // 状态
    previewLoadingVoiceId,
--- a/frontend/app/web-gold/src/composables/useUpload.js
+++ b/frontend/app/web-gold/src/composables/useUpload.js
@@ -133,28 +133,23 @@ export function useUpload() {
        fileType: file.type,
        groupId,
        coverBase64,
-        duration: file.type.startsWith('video/') ? null : undefined // 视频时长由后端处理或前端可选传递
+        duration: file.type.startsWith('video/') ? null : undefined
      })

-      // 设置成功状态
      state.uploading = false
      state.status = 'success'
      state.progress = 100

-      // 通知成功
      const fileId = completeData.data?.infraFileId || completeData.data?.userFileId
-      onSuccess && onSuccess(fileId)
+      const fileUrl = presignedData.data.presignedUrl
+      onSuccess && onSuccess(fileId, fileUrl)

      return fileId
    } catch (error) {
-      // 设置错误状态
      state.uploading = false
      state.status = 'error'
      state.error = error.message || '上传失败'
-
-      // 通知错误
      onError && onError(error)
-
      throw error
    }
  }
--- a/frontend/app/web-gold/src/main.js
+++ b/frontend/app/web-gold/src/main.js
@@ -3,6 +3,7 @@ import { createPinia } from 'pinia'
 import Antd from 'ant-design-vue'
 import 'normalize.css'
 import 'ant-design-vue/dist/reset.css'
+import 'aplayer/dist/APlayer.min.css'
 import piniaPluginPersistedstate from 'pinia-plugin-persistedstate';
 import 'dayjs/locale/zh-cn';

--- a/frontend/app/web-gold/src/views/dh/VoiceCopy.vue
+++ b/frontend/app/web-gold/src/views/dh/VoiceCopy.vue
@@ -105,14 +105,15 @@
 </template>

 <script setup>
-import { ref, reactive, computed, onMounted, nextTick } from 'vue'
+import { ref, reactive, computed, onMounted } from 'vue'
 import { message, Modal } from 'ant-design-vue'
 import { PlusOutlined, SearchOutlined, UploadOutlined, PlayCircleOutlined } from '@ant-design/icons-vue'
-import { VoiceService } from '@/api/voice'
-import { MaterialService } from '@/api/material'
-import { useUpload } from '@/composables/useUpload'
 import dayjs from 'dayjs'
 import BasicLayout from '@/layouts/components/BasicLayout.vue'
+import { MaterialService } from '@/api/material'
+import { VoiceService } from '@/api/voice'
+import { useUpload } from '@/composables/useUpload'
+import useVoiceText from '@gold/hooks/web/useVoiceText'

 // ========== 常量 ==========

@@ -123,9 +124,15 @@ const DEFAULT_FORM_DATA = {
  autoTranscribe: true,
  language: 'zh-CN',
  gender: 'female',
-  note: ''
+  note: '',
+  text: '',
+  fileUrl: ''
 }

+const MAX_FILE_SIZE = 50 * 1024 * 1024
+const VALID_AUDIO_TYPES = ['audio/mpeg', 'audio/wav', 'audio/wave', 'audio/x-wav', 'audio/aac', 'audio/mp4', 'audio/flac', 'audio/ogg']
+const VALID_AUDIO_EXTENSIONS = ['.mp3', '.wav', '.aac', '.m4a', '.flac', '.ogg']
+
 // ========== 响应式数据 ==========
 const loading = ref(false)
 const submitting = ref(false)
@@ -155,6 +162,9 @@ const formData = reactive({ ...DEFAULT_FORM_DATA })
 // ========== Upload Hook ==========
 const { state: uploadState, upload } = useUpload()

+// ========== VoiceText Hook ==========
+const { getVoiceText } = useVoiceText()
+
 // ========== 计算属性 ==========
 const isCreateMode = computed(() => formMode.value === 'create')

@@ -210,31 +220,31 @@ const loadVoiceList = async () => {
 }

 // ========== 搜索和分页 ==========
-const handleSearch = () => {
+function handleSearch() {
  pagination.current = 1
  loadVoiceList()
 }

-const handleReset = () => {
+function handleReset() {
  searchParams.name = ''
  pagination.current = 1
  loadVoiceList()
 }

-const handleTableChange = (pag) => {
+function handleTableChange(pag) {
  pagination.current = pag.current
  pagination.pageSize = pag.pageSize
  loadVoiceList()
 }

 // ========== CRUD 操作 ==========
-const handleCreate = () => {
+function handleCreate() {
  formMode.value = 'create'
  resetForm()
  modalVisible.value = true
 }

-const handleEdit = async (record) => {
+async function handleEdit(record) {
  formMode.value = 'edit'
  try {
    const res = await VoiceService.get(record.id)
@@ -246,13 +256,13 @@ const handleEdit = async (record) => {
  modalVisible.value = true
 }

-const handleDelete = (record) => {
+function handleDelete(record) {
  Modal.confirm({
    title: '确认删除',
    content: `确定要删除配音「${record.name}」吗？此操作不可恢复。`,
    okButtonProps: { danger: true },
    centered: true,
-    onOk: async () => {
+    onOk: async function() {
      try {
        const res = await VoiceService.delete(record.id)
        if (res.code !== 0) return message.error(res.msg || '删除失败')
@@ -268,7 +278,7 @@ const handleDelete = (record) => {
 }

 // ========== 音频播放 ==========
-const handlePlayAudio = (record) => {
+function handlePlayAudio(record) {
  if (record.fileUrl && audioPlayer.value) {
    audioPlayer.value.src = record.fileUrl
    audioPlayer.value.play()
@@ -278,20 +288,20 @@ const handlePlayAudio = (record) => {
 }

 // ========== 文件上传 ==========
-const handleBeforeUpload = (file) => {
-  const MAX_FILE_SIZE = 50 * 1024 * 1024
+function handleBeforeUpload(file) {
  if (file.size > MAX_FILE_SIZE) {
    message.error('文件大小不能超过 50MB')
    return false
  }

-  const validTypes = ['audio/mpeg', 'audio/wav', 'audio/wave', 'audio/x-wav', 'audio/aac', 'audio/mp4', 'audio/flac', 'audio/ogg']
-  const validExtensions = ['.mp3', '.wav', '.aac', '.m4a', '.flac', '.ogg']
  const fileName = file.name.toLowerCase()
  const fileType = file.type.toLowerCase()

-  const isValidType = validTypes.some(type => fileType.includes(type)) ||
-                      validExtensions.some(ext => fileName.endsWith(ext))
+  const isValidType = VALID_AUDIO_TYPES.some(function(type) {
+    return fileType.includes(type)
+  }) || VALID_AUDIO_EXTENSIONS.some(function(ext) {
+    return fileName.endsWith(ext)
+  })

  if (!isValidType) {
    message.error('请上传音频文件（MP3、WAV、AAC、M4A、FLAC、OGG）')
@@ -301,22 +311,23 @@ const handleBeforeUpload = (file) => {
  return true
 }

-const handleCustomUpload = async (options) => {
+// ========== 文件上传相关 ==========
+async function handleCustomUpload(options) {
  const { file, onSuccess, onError } = options

  try {
    const fileId = await upload(file, {
      fileCategory: 'voice',
-      groupId: null, // 配音模块不使用groupId
+      groupId: null,
      coverBase64: null,
-      onStart: () => {},
-      onProgress: () => {},
-      onSuccess: (id) => {
+      onSuccess: async function(id, fileUrl) {
        formData.fileId = id
+        formData.fileUrl = fileUrl
        message.success('文件上传成功')
+        await fetchAudioTextById(id)
        onSuccess?.({ code: 0, data: id }, file)
      },
-      onError: (error) => {
+      onError: function(error) {
        const errorMsg = error.message || '上传失败，请稍后重试'
        message.error(errorMsg)
        onError?.(error)
@@ -330,24 +341,43 @@ const handleCustomUpload = async (options) => {
  }
 }

-const handleFileListChange = (info) => {
-  // 处理文件列表变化，避免直接修改导致 DOM 错误
-  const { fileList: newFileList } = info
-  
-  // 只更新文件列表，不直接修改文件项的状态
-  // 让组件自己管理状态
-  if (newFileList) {
-    fileList.value = newFileList.filter(item => item.status !== 'removed')
+// 通过fileId获取音频文本
+async function fetchAudioTextById(fileId) {
+  if (!fileId) return
+  try {
+    const res = await MaterialService.getAudioPlayUrl(fileId)
+    if (res.code === 0 && res.data) {
+      const rawFileUrl = res.data
+      const results = await getVoiceText([{ audio_url: rawFileUrl }])
+      if (results && results.length > 0) {
+        const text = results[0].value
+        formData.text = text
+        if (text) {
+          message.success('音频文本获取成功')
+        }
+      }
+    }
+  } catch (error) {
+    console.error('获取音频文本失败:', error)
  }
 }

-const handleRemoveFile = () => {
+function handleFileListChange(info) {
+  const { fileList: newFileList } = info
+  if (newFileList) {
+    fileList.value = newFileList.filter(function(item) {
+      return item.status !== 'removed'
+    })
+  }
+}
+
+function handleRemoveFile() {
  formData.fileId = null
  fileList.value = []
 }

 // ========== 表单操作 ==========
-const handleSubmit = async () => {
+async function handleSubmit() {
  try {
    await formRef.value.validate()
  } catch {
@@ -363,7 +393,8 @@ const handleSubmit = async () => {
        autoTranscribe: formData.autoTranscribe,
        language: formData.language,
        gender: formData.gender,
-        note: formData.note
+        note: formData.note,
+        text: formData.text
      }
    : {
        id: formData.id,
@@ -394,19 +425,19 @@ const handleSubmit = async () => {
  }
 }

-const handleCancel = () => {
+function handleCancel() {
  modalVisible.value = false
  resetForm()
 }

-const resetForm = () => {
+function resetForm() {
  Object.assign(formData, { ...DEFAULT_FORM_DATA })
  fileList.value = []
  formRef.value?.resetFields()
 }

 // ========== 生命周期 ==========
-onMounted(() => {
+onMounted(function() {
  loadVoiceList()
 })
 </script>
--- a/frontend/app/web-gold/src/views/kling/IdentifyFace.vue
+++ b/frontend/app/web-gold/src/views/kling/IdentifyFace.vue
@@ -15,16 +15,16 @@
              :show-count="true"
              class="tts-textarea"
            />
-            <div v-if="identifyState.identified && faceDuration > 0" class="text-hint">
-              <span class="hint-icon">💡</span>
-              <span>视频中人脸出现时长约 {{ (faceDuration / 1000).toFixed(1) }} 秒，建议文案不超过 {{ suggestedMaxChars }} 字</span>
-            </div>
          </div>

          <!-- 音色选择 -->
          <div class="section">
            <h3>音色</h3>
-            <VoiceSelector ref="voiceSelectorRef" @select="handleVoiceSelect" />
+            <VoiceSelector
+              :synth-text="ttsText"
+              :speech-rate="speechRate"
+              @select="handleVoiceSelect"
+            />
          </div>

          <!-- TTS 控制 -->
@@ -137,143 +137,74 @@
            </div>
          </div>

-          <!-- 素材校验结果 -->
-          <div v-if="materialValidation.videoDuration > 0 && materialValidation.audioDuration > 0" class="section">
-            <h3>素材校验</h3>
-            <div class="validation-result" :class="{ 'validation-passed': materialValidation.isValid, 'validation-failed': !materialValidation.isValid }">
-              <div class="validation-status">
-                <span class="status-icon">{{ materialValidation.isValid ? '✅' : '❌' }}</span>
-                <span class="status-text">{{ materialValidation.isValid ? '校验通过' : '校验失败' }}</span>
-              </div>
-
-              <!-- 时长对比进度条 -->
-              <div class="duration-comparison">
-                <div class="duration-bar">
-                  <div class="duration-label">
-                    <span>音频时长</span>
-                    <span class="duration-value">{{ (materialValidation.audioDuration / 1000).toFixed(1) }}s</span>
-                  </div>
-                  <div class="progress-bar audio-bar">
-                    <div
-                      class="progress-fill"
-                      :style="{ width: `${(materialValidation.audioDuration / Math.max(materialValidation.videoDuration, materialValidation.audioDuration)) * 100}%` }"
-                    ></div>
-                  </div>
-                </div>
-
-                <div class="duration-bar">
-                  <div class="duration-label">
-                    <span>视频时长</span>
-                    <span class="duration-value">{{ (materialValidation.videoDuration / 1000).toFixed(1) }}s</span>
-                  </div>
-                  <div class="progress-bar video-bar">
-                    <div
-                      class="progress-fill"
-                      :class="{ 'success': materialValidation.isValid, 'error': !materialValidation.isValid }"
-                      :style="{ width: `${(materialValidation.videoDuration / Math.max(materialValidation.videoDuration, materialValidation.audioDuration)) * 100}%` }"
-                    ></div>
-                  </div>
-                </div>
-              </div>
-
-              <!-- 失败提示和建议 -->
-              <div v-if="!materialValidation.isValid" class="validation-error">
-                <p class="error-message">
-                  视频时长必须大于音频时长才能生成数字人视频
-                </p>
-                <div class="quick-actions">
-                  <a-button size="small" @click="replaceVideo">更换视频</a-button>
-                  <a-button size="small" @click="handleSimplifyScript">精简文案</a-button>
-                </div>
-              </div>
-            </div>
-          </div>
-
-          <!-- 配音生成与校验（仅在识别后显示） -->
-          <div v-if="identifyState.identified" class="section audio-generation-section">
-            <h3>配音生成与校验</h3>
-
-            <!-- 生成配音按钮 -->
-            <div class="generate-audio-row">
-              <a-button
-                type="default"
-                size="large"
-                :disabled="!canGenerateAudio"
-                :loading="audioState.generating"
-                block
-                @click="generateAudio"
-              >
-                {{ audioState.generating ? '生成中...' : '生成配音（用于校验时长）' }}
-              </a-button>
-            </div>
-
-            <!-- 音频预览（生成后显示） -->
-            <div v-if="audioState.generated" class="audio-preview">
-              <div class="audio-info">
-                <h4>生成的配音</h4>
-                <div class="duration-info">
-                  <span class="label">音频时长：</span>
-                  <span class="value">{{ (audioState.durationMs / 1000).toFixed(1) }} 秒</span>
-                </div>
-                <div class="duration-info">
-                  <span class="label">人脸区间：</span>
-                  <span class="value">{{ (faceDuration / 1000).toFixed(1) }} 秒</span>
-                </div>
-                <div class="duration-info" :class="{ 'validation-passed': audioState.validationPassed, 'validation-failed': !audioState.validationPassed }">
-                  <span class="label">校验结果：</span>
-                  <span class="value">
-                    {{ audioState.validationPassed ? '✅ 通过' : '❌ 不通过（需至少2秒重合）' }}
-                  </span>
-                </div>
+          <!-- 配音生成（仅在 Pipeline 到达 ready 状态后显示） -->
+          <div v-if="isPipelineReady" class="section audio-section">
+            <!-- 已生成音频 -->
+            <div v-if="audioState.generated" class="audio-generated">
+              <div class="audio-header">
+                <span class="audio-title">配音</span>
+                <span class="audio-duration">{{ audioDurationSec }}秒</span>
              </div>

              <!-- 音频播放器 -->
-              <div class="audio-player">
-                <audio
-                  v-if="audioState.generated.audioBase64"
-                  :src="`data:audio/mp3;base64,${audioState.generated.audioBase64}`"
-                  controls
-                  class="audio-element"
-                />
-                <audio
-                  v-else-if="audioState.generated.audioUrl"
-                  :src="audioState.generated.audioUrl"
-                  controls
-                  class="audio-element"
-                />
+              <div v-if="audioUrl" class="audio-player-wrapper">
+                <audio :src="audioUrl" controls class="audio-player" />
              </div>

-              <!-- 重新生成按钮 -->
-              <div class="regenerate-row">
-                <a-button
-                  type="link"
-                  size="small"
-                  @click="generateAudio"
-                  :loading="audioState.generating"
-                >
-                  重新生成
-                </a-button>
+              <!-- 校验失败提示 -->
+              <div v-if="!validationPassed" class="validation-warning">
+                <span class="warning-icon">⚠️</span>
+                <span class="warning-text">音频时长({{ audioDurationSec }}秒)超过视频人脸区间({{ faceDurationSec }}秒)，请缩短文案或调整语速</span>
              </div>
+
+              <!-- 重新生成 -->
+              <a-button type="link" size="small" :loading="audioState.generating" @click="generateAudio">
+                重新生成
+              </a-button>
            </div>
          </div>

+          <!-- Pipeline 进度条 -->
+          <PipelineProgress
+            v-if="isPipelineBusy || isPipelineReady || isPipelineFailed || isPipelineCompleted"
+            :state="pipelineState"
+            :progress="pipelineProgress"
+            :is-busy="isPipelineBusy"
+            :is-ready="isPipelineReady"
+            :is-failed="isPipelineFailed"
+            :is-completed="isPipelineCompleted"
+            :error="pipelineError"
+            @retry="retryPipeline"
+            @reset="resetPipeline"
+          />
+
          <!-- 按钮组 -->
          <div class="action-buttons">
+            <!-- 准备阶段：先运行到 ready -->
            <a-button
+              v-if="!isPipelineReady"
              type="primary"
              size="large"
              :disabled="!canGenerate"
+              :loading="isPipelineBusy"
+              block
+              @click="generateAudio"
+            >
+              {{ isPipelineBusy ? '处理中...' : '生成配音并验证' }}
+            </a-button>
+
+            <!-- Ready 后：生成数字人视频 -->
+            <a-button
+              v-else
+              type="primary"
+              size="large"
+              :loading="isPipelineBusy"
              block
              @click="generateDigitalHuman"
            >
-              生成数字人视频
+              {{ isPipelineBusy ? '处理中...' : '生成数字人视频' }}
            </a-button>

-            <!-- 添加提示信息 -->
-            <div v-if="canGenerate && !audioState.validationPassed" class="generate-hint">
-              <span class="hint-icon">⚠️</span>
-              <span>请先生成配音并通过时长校验</span>
-            </div>
          </div>
        </div>

@@ -295,13 +226,12 @@ import VideoSelector from '@/components/VideoSelector.vue'
 import VoiceSelector from '@/components/VoiceSelector.vue'
 import ResultPanel from '@/components/ResultPanel.vue'
 import FullWidthLayout from '@/layouts/components/FullWidthLayout.vue'
+import PipelineProgress from '@/components/PipelineProgress.vue'

 // Controller Hook
 import { useIdentifyFaceController } from './hooks/useIdentifyFaceController'

 const voiceStore = useVoiceCopyStore()
-const voiceSelectorRef: any = ref(null)
-
 const dragOver = ref(false)

 // ==================== 初始化 Controller ====================
@@ -315,15 +245,10 @@ const {
  ttsText,
  speechRate,
  audioState,
-  canGenerateAudio,
-  suggestedMaxChars,
  generateAudio,

  // 数字人生成相关
  videoState,
-  identifyState,
-  materialValidation,
-  faceDuration,
  getVideoPreviewUrl,

  // 计算属性
@@ -332,6 +257,21 @@ const {
  textareaPlaceholder,
  speechRateMarks,
  speechRateDisplay,
+  faceDurationSec,
+  audioDurationSec,
+  audioUrl,
+  validationPassed,
+
+  // Pipeline 状态（单一状态源）
+  pipelineState,
+  isPipelineBusy,
+  isPipelineReady,
+  isPipelineFailed,
+  isPipelineCompleted,
+  pipelineProgress,
+  pipelineError,
+  retryPipeline,
+  resetPipeline,

  // 事件处理方法
  handleVoiceSelect,
@@ -341,7 +281,6 @@ const {
  handleSelectUpload,
  handleSelectFromLibrary,
  handleVideoSelect,
-  handleSimplifyScript,
  handleVideoLoaded,
  replaceVideo,
  generateDigitalHuman,
@@ -355,12 +294,6 @@ const {

 onMounted(async () => {
  await voiceStore.refresh()
-
-  // 设置VoiceSelector的试听文本和语速
-  if (voiceSelectorRef.value) {
-    voiceSelectorRef.value.setPreviewText(ttsText.value)
-    voiceSelectorRef.value.setPreviewSpeechRate(speechRate.value)
-  }
 })
 </script>

@@ -423,20 +356,15 @@ onMounted(async () => {
  font-weight: 600;
 }

-.card-content h4,
-.audio-info h4 {
+.card-content h4 {
  color: var(--text-primary);
  font-size: 14px;
  margin-bottom: 12px;
 }

-.card-content p,
-.duration-label span:first-child {
+.card-content p {
  color: var(--text-secondary);
  font-size: 13px;
-}
-
-.card-content p {
  margin: 0;
 }

@@ -454,24 +382,6 @@ onMounted(async () => {
  }
 }

-.text-hint {
-  display: flex;
-  align-items: center;
-  gap: 8px;
-  margin-top: 12px;
-  padding: 12px 16px;
-  background: rgba(var(--color-primary), 0.1);
-  border: 1px solid rgba(var(--color-primary), 0.3);
-  border-radius: 8px;
-  font-size: 13px;
-  color: var(--text-secondary);
-}
-
-.hint-icon {
-  font-size: 16px;
-}
-
-/* ========== 控制面板 ========== */
 .control-group {
  margin-bottom: 16px;
 }
@@ -688,175 +598,76 @@ onMounted(async () => {
  border-radius: 8px;
 }

-/* ========== 验证结果 ========== */
-.validation-result {
-  padding: 16px;
-  background: var(--bg-primary);
-  border-radius: 8px;
-  border: 1px solid var(--border-light);
+/* ========== 音频区域 ========== */
+.audio-section {
+  margin-bottom: 24px;
 }

-.validation-result.validation-passed {
-  border-color: var(--color-success);
-  background: rgba(var(--color-success), 0.05);
+.audio-generated {
+  display: flex;
+  flex-direction: column;
+  gap: 12px;
 }

-.validation-result.validation-failed {
-  border-color: var(--color-error);
-  background: rgba(var(--color-error), 0.05);
-}
-
-.validation-status {
+.audio-header {
  display: flex;
  align-items: center;
-  gap: 8px;
-  margin-bottom: 12px;
-  padding: 12px;
-  background: var(--bg-secondary);
-  border-radius: 6px;
+  justify-content: space-between;
 }

-.status-icon {
-  font-size: 18px;
-}
-
-.status-text {
-  color: var(--text-primary);
+.audio-title {
  font-size: 14px;
  font-weight: 600;
-}
-
-/* ========== 时长对比进度条 ========== */
-.duration-comparison {
-  margin-bottom: 16px;
-  padding: 12px;
-  background: var(--bg-secondary);
-  border-radius: 6px;
-}
-
-.duration-bar {
-  margin-bottom: 12px;
-
-  &:last-child {
-    margin-bottom: 0;
-  }
-}
-
-.duration-label {
-  display: flex;
-  justify-content: space-between;
-  align-items: center;
-  margin-bottom: 8px;
-  font-size: 13px;
-}
-
-.duration-value {
  color: var(--text-primary);
-  font-weight: 600;
-  font-size: 13px;
-  padding: 4px 8px;
-  background: var(--bg-primary);
-  border-radius: 4px;
 }

-.progress-bar {
-  height: 8px;
-  background: var(--bg-primary);
-  border-radius: 4px;
-  overflow: hidden;
-}
-
-.progress-fill {
-  height: 100%;
-  border-radius: 4px;
-  transition: width 0.3s;
-}
-
-.audio-bar .progress-fill {
-  background: var(--color-primary);
-}
-
-.video-bar .progress-fill.success {
-  background: var(--color-success);
-}
-
-.video-bar .progress-fill.error {
-  background: var(--color-error);
-}
-
-/* ========== 错误提示 ========== */
-.validation-error {
-  padding: 12px;
-  background: var(--bg-secondary);
-  border: 1px solid var(--border-light);
-  border-radius: 6px;
-}
-
-.error-message {
-  color: var(--color-error);
-  font-size: 13px;
-  margin: 0 0 12px 0;
-}
-
-.quick-actions {
-  display: flex;
-  gap: 8px;
-}
-
-/* ========== 音频生成 ========== */
-.audio-generation-section {
-  margin-bottom: 24px;
-  padding: 16px;
-  background: var(--bg-secondary);
-  border-radius: 8px;
-  border: 1px solid var(--border-light);
-}
-
-.generate-audio-row {
-  margin-bottom: 16px;
-}
-
-.audio-preview {
-  padding: 16px;
-  background: var(--bg-primary);
-  border-radius: 8px;
-}
-
-.duration-info {
-  display: flex;
-  justify-content: space-between;
-  margin-bottom: 8px;
-  font-size: 13px;
-}
-
-.duration-info .label {
+.audio-duration {
+  font-size: 12px;
  color: var(--text-secondary);
 }

-.duration-info .value {
-  color: var(--text-primary);
-  font-weight: 600;
-}
-
-.duration-info.validation-passed .value {
-  color: var(--color-success);
-}
-
-.duration-info.validation-failed .value {
-  color: var(--color-error);
-}
-
-.audio-player {
-  margin: 16px 0;
-}
-
-.audio-element {
+.audio-player-wrapper {
  width: 100%;
 }

-.regenerate-row {
+.audio-player {
+  width: 100%;
+  height: 36px;
+}
+
+.validation-warning {
+  display: flex;
+  align-items: flex-start;
+  gap: 8px;
+  padding: 10px 12px;
+  background: rgba(var(--color-warning), 0.1);
+  border: 1px solid rgba(var(--color-warning), 0.3);
+  border-radius: 6px;
+  font-size: 13px;
+}
+
+.warning-icon {
+  flex-shrink: 0;
+  font-size: 14px;
+}
+
+.warning-text {
+  color: var(--text-secondary);
+  line-height: 1.4;
+}
+
+.audio-prompt {
  text-align: center;
-  margin-top: 12px;
+  padding: 20px;
+  background: var(--bg-secondary);
+  border-radius: 8px;
+  border: 1px dashed var(--border-light);
+}
+
+.audio-prompt p {
+  margin: 0 0 16px 0;
+  font-size: 14px;
+  color: var(--text-secondary);
 }

 /* ========== 操作按钮 ========== */
@@ -876,18 +687,6 @@ onMounted(async () => {
  border-radius: 8px;
 }

-.generate-hint {
-  display: flex;
-  align-items: center;
-  gap: 8px;
-  padding: 12px 16px;
-  background: rgba(var(--color-warning), 0.1);
-  border: 1px solid rgba(var(--color-warning), 0.3);
-  border-radius: 6px;
-  font-size: 13px;
-  color: var(--color-warning);
-}
-
 /* ========== 响应式 ========== */
@media (max-width: 1024px) {
  .kling-content {
--- a/frontend/app/web-gold/src/views/kling/hooks/pipeline/states.ts
+++ b/frontend/app/web-gold/src/views/kling/hooks/pipeline/states.ts
@@ -0,0 +1,124 @@
+/**
+ * @fileoverview 状态机配置 - 状态定义和配置
+ */
+
+import type { PipelineState, StateConfig } from './types'
+
+/**
+ * 状态配置映射表
+ */
+export const STATE_CONFIG: Record<PipelineState, StateConfig> = {
+  idle: {
+    label: '等待开始',
+    progress: 0,
+    description: '请先选择视频并输入文案',
+  },
+  uploading: {
+    label: '上传视频中',
+    progress: 15,
+    description: '正在上传视频文件...',
+  },
+  recognizing: {
+    label: '识别人脸中',
+    progress: 35,
+    description: '正在分析视频中的人脸信息...',
+  },
+  generating: {
+    label: '生成配音中',
+    progress: 55,
+    description: '正在合成语音...',
+  },
+  validating: {
+    label: '校验时长中',
+    progress: 70,
+    description: '正在校验音频与视频时长...',
+  },
+  ready: {
+    label: '准备就绪',
+    progress: 80,
+    description: '校验通过，可以创建数字人视频',
+  },
+  creating: {
+    label: '创建任务中',
+    progress: 95,
+    description: '正在提交数字人视频生成任务...',
+  },
+  completed: {
+    label: '已完成',
+    progress: 100,
+    description: '任务已提交成功',
+  },
+  failed: {
+    label: '失败',
+    progress: 0,
+    description: '操作失败，请重试',
+  },
+}
+
+/**
+ * 状态顺序（用于步骤条显示）
+ */
+export const STATE_ORDER: PipelineState[] = [
+  'idle',
+  'uploading',
+  'recognizing',
+  'generating',
+  'validating',
+  'ready',
+  'creating',
+  'completed',
+]
+
+/**
+ * 忙碌状态（正在执行中的状态）
+ */
+export const BUSY_STATES: PipelineState[] = [
+  'uploading',
+  'recognizing',
+  'generating',
+  'validating',
+  'creating',
+]
+
+/**
+ * 终态（不能再转换的状态）
+ */
+export const TERMINAL_STATES: PipelineState[] = [
+  'completed',
+  'failed',
+]
+
+/**
+ * 获取状态在步骤条中的索引
+ */
+export function getStateIndex(state: PipelineState): number {
+  return STATE_ORDER.indexOf(state)
+}
+
+/**
+ * 获取状态的进度百分比
+ */
+export function getStateProgress(state: PipelineState): number {
+  return STATE_CONFIG[state].progress
+}
+
+/**
+ * 判断是否为忙碌状态
+ */
+export function isBusyState(state: PipelineState): boolean {
+  return BUSY_STATES.includes(state)
+}
+
+/**
+ * 判断是否为终态
+ */
+export function isTerminalState(state: PipelineState): boolean {
+  return TERMINAL_STATES.includes(state)
+}
+
+/**
+ * 判断状态是否可以重试
+ */
+export function canRetryFrom(state: PipelineState): boolean {
+  return state === 'failed'
+}
--- a/frontend/app/web-gold/src/views/kling/hooks/pipeline/types.ts
+++ b/frontend/app/web-gold/src/views/kling/hooks/pipeline/types.ts
@@ -0,0 +1,126 @@
+/**
+ * @fileoverview 数字人生成流程状态机 - 类型定义
+ */
+
+/**
+ * 状态机所有可能的状态
+ */
+export type PipelineState =
+  | 'idle'          // 空闲
+  | 'uploading'     // 上传视频中
+  | 'recognizing'   // 人脸识别中
+  | 'generating'    // 生成配音中
+  | 'validating'    // 校验时长中
+  | 'ready'         // 准备就绪
+  | 'creating'      // 创建任务中
+  | 'completed'     // 已完成
+  | 'failed'        // 失败
+
+/**
+ * 状态配置
+ */
+export interface StateConfig {
+  /** 状态标签 */
+  label: string
+  /** 进度百分比 */
+  progress: number
+  /** 描述 */
+  description: string
+  /** 图标（可选） */
+  icon?: string
+}
+
+/**
+ * 步骤执行结果
+ */
+export interface StepResult<T = any> {
+  /** 是否成功 */
+  success: boolean
+  /** 返回数据 */
+  data?: T
+  /** 错误信息 */
+  error?: Error
+}
+
+/**
+ * Pipeline 上下文数据
+ */
+export interface PipelineContext {
+  /** 视频文件 */
+  videoFile: File | null
+  /** 已选择的视频 */
+  selectedVideo: any
+  /** 文案内容 */
+  text: string
+  /** 音色 */
+  voice: any
+  /** 语速 */
+  speechRate: number
+  /** 视频文件ID */
+  videoFileId: string | number | null
+  /** 会话ID */
+  sessionId: string
+  /** 人脸ID */
+  faceId: string
+  /** 人脸开始时间 */
+  faceStartTime: number
+  /** 人脸结束时间 */
+  faceEndTime: number
+  /** 音频 Base64 */
+  audioBase64: string
+  /** 音频格式 */
+  audioFormat: string
+  /** 音频时长（毫秒） */
+  audioDurationMs: number
+  /** 视频时长（毫秒） */
+  videoDurationMs: number
+  /** 校验是否通过 */
+  validationPassed: boolean
+}
+
+/**
+ * Pipeline 执行参数
+ */
+export interface PipelineParams {
+  videoFile: File | null
+  selectedVideo: any
+  text: string
+  voice: any
+  speechRate: number
+}
+
+/**
+ * Pipeline 选项配置
+ */
+export interface PipelineOptions {
+  /** 上传视频 */
+  uploadVideo: (file: File) => Promise<string | number>
+  /** 从库中识别 */
+  recognizeFromLibrary: (video: any) => Promise<any>
+  /** 识别已上传视频 */
+  recognizeUploaded: (fileId: string | number) => Promise<any>
+  /** 生成音频 */
+  generateAudio: (text: string, voice: any, speechRate: number) => Promise<{
+    audioBase64: string
+    format?: string
+    durationMs?: number
+  }>
+  /** 创建任务 */
+  createTask: (data: any) => Promise<void>
+}
+
+/**
+ * 状态机执行状态
+ */
+export interface ExecutionState {
+  /** 当前状态 */
+  current: PipelineState
+  /** 历史状态 */
+  history: PipelineState[]
+  /** 上下文数据 */
+  context: Partial<PipelineContext>
+  /** 是否可以继续下一步 */
+  canNext: boolean
+  /** 是否可以重试 */
+  canRetry: boolean
+}
--- a/frontend/app/web-gold/src/views/kling/hooks/pipeline/useSimplePipeline.ts
+++ b/frontend/app/web-gold/src/views/kling/hooks/pipeline/useSimplePipeline.ts
@@ -0,0 +1,287 @@
+/**
+ * @fileoverview 极简状态机 Hook - 数字人生成流程
+ *
+ * 设计理念：
+ * 1. 简单直观 - 用普通 JS/TS 代码，无需学习复杂概念
+ * 2. 易于调试 - 打断点即可查看状态
+ * 3. 功能完整 - 支持状态管理、进度显示、错误处理、重试
+ */
+
+import { ref, computed } from 'vue'
+import { message } from 'ant-design-vue'
+import type { LipSyncTaskData } from '../../types/identify-face'
+import { createLipSyncTask } from '@/api/kling'
+import type {
+  PipelineState,
+  PipelineContext,
+  PipelineParams,
+  PipelineOptions,
+  ExecutionState,
+} from './types'
+import {
+  STATE_CONFIG,
+  getStateIndex,
+  isBusyState,
+  isTerminalState,
+  canRetryFrom,
+} from './states'
+
+/**
+ * 初始上下文
+ */
+const INITIAL_CONTEXT: Partial<PipelineContext> = {
+  videoFile: null,
+  selectedVideo: null,
+  text: '',
+  voice: null,
+  speechRate: 1,
+  videoFileId: null,
+  sessionId: '',
+  faceId: '',
+  faceStartTime: 0,
+  faceEndTime: 0,
+  audioBase64: '',
+  audioFormat: 'mp3',
+  audioDurationMs: 0,
+  videoDurationMs: 0,
+  validationPassed: false,
+}
+
+/**
+ * 极简状态机 Hook
+ */
+export function useSimplePipeline(options: PipelineOptions) {
+  // ========== 状态管理 ==========
+  const state = ref<PipelineState>('idle')
+  const context = ref<Partial<PipelineContext>>({ ...INITIAL_CONTEXT })
+  const error = ref<string | null>(null)
+  const history = ref<PipelineState[]>(['idle'])
+
+  // ========== 计算属性 ==========
+  const stateLabel = computed(() => STATE_CONFIG[state.value].label)
+  const stateDescription = computed(() => STATE_CONFIG[state.value].description)
+  const progress = computed(() => STATE_CONFIG[state.value].progress)
+  const currentStepIndex = computed(() => getStateIndex(state.value))
+  const isBusy = computed(() => isBusyState(state.value))
+  const isReady = computed(() => state.value === 'ready')
+  const isFailed = computed(() => state.value === 'failed')
+  const isCompleted = computed(() => state.value === 'completed')
+  const isTerminal = computed(() => isTerminalState(state.value))
+  const canRetry = computed(() => canRetryFrom(state.value))
+
+  // ========== 内部方法 ==========
+
+  /**
+   * 更新状态
+   */
+  function setState(newState: PipelineState) {
+    const oldState = state.value
+    state.value = newState
+    history.value.push(newState)
+    console.log(`[Pipeline] ${oldState} -> ${newState}`)
+  }
+
+  /**
+   * 设置错误状态
+   */
+  function setError(err: Error | string) {
+    const errorMsg = typeof err === 'string' ? err : err.message
+    error.value = errorMsg
+    setState('failed')
+    message.error(errorMsg)
+  }
+
+  /**
+   * 执行步骤（带错误处理）
+   */
+  async function executeStep<T>(
+    newState: PipelineState,
+    fn: () => Promise<T>
+  ): Promise<T> {
+    setState(newState)
+    try {
+      return await fn()
+    } catch (err) {
+      setError(err as Error)
+      throw err
+    }
+  }
+
+  // ========== 公开方法 ==========
+
+  /**
+   * 运行完整流程（到 ready 状态）
+   */
+  async function run(params: PipelineParams): Promise<void> {
+    // 重置状态
+    reset()
+
+    try {
+      // 保存参数到上下文
+      context.value.videoFile = params.videoFile
+      context.value.selectedVideo = params.selectedVideo
+      context.value.text = params.text
+      context.value.voice = params.voice
+      context.value.speechRate = params.speechRate
+
+      // 步骤1: 上传视频（如果是上传模式）
+      if (params.videoFile && !params.selectedVideo) {
+        const fileId = await executeStep('uploading', () =>
+          options.uploadVideo(params.videoFile!)
+        )
+        context.value.videoFileId = fileId
+      } else if (params.selectedVideo) {
+        context.value.videoFileId = params.selectedVideo.fileId
+      }
+
+      // 步骤2: 识别人脸
+      const recognizeData = params.selectedVideo
+        ? await options.recognizeFromLibrary(params.selectedVideo)
+        : await options.recognizeUploaded(context.value.videoFileId!)
+
+      await executeStep('recognizing', async () => recognizeData)
+
+      context.value.sessionId = recognizeData.sessionId
+      context.value.faceId = recognizeData.faceId
+      context.value.faceStartTime = recognizeData.startTime || 0
+      context.value.faceEndTime = recognizeData.endTime || 0
+      context.value.videoDurationMs = recognizeData.duration || 0
+
+      // 步骤3: 生成音频
+      const audioData = await executeStep('generating', () =>
+        options.generateAudio(params.text, params.voice, params.speechRate)
+      )
+
+      context.value.audioBase64 = audioData.audioBase64
+      context.value.audioFormat = audioData.format || 'mp3'
+      context.value.audioDurationMs = audioData.durationMs || 0
+
+      // 步骤4: 校验时长
+      setState('validating')
+      const videoDurationMs = context.value.videoDurationMs ?? 0
+      if (context.value.audioDurationMs > videoDurationMs) {
+        throw new Error(
+          `校验失败：音频时长(${(context.value.audioDurationMs / 1000).toFixed(1)}秒) 超过人脸时长(${(videoDurationMs / 1000).toFixed(1)}秒)`
+        )
+      }
+      context.value.validationPassed = true
+
+      // 到达 ready 状态
+      setState('ready')
+
+    } catch (err) {
+      // 错误已在 executeStep 中处理
+    }
+  }
+
+  /**
+   * 创建数字人任务（从 ready 状态）
+   */
+  async function createTask(): Promise<void> {
+    if (state.value !== 'ready') {
+      message.warning('请先完成视频识别和音频生成')
+      return
+    }
+
+    try {
+      setState('creating')
+
+      const taskData: LipSyncTaskData = {
+        taskName: `数字人任务_${Date.now()}`,
+        videoFileId: context.value.videoFileId!,
+        inputText: context.value.text!,
+        speechRate: context.value.speechRate!,
+        volume: 0,
+        guidanceScale: 1,
+        seed: 8888,
+        kling_session_id: context.value.sessionId!,
+        kling_face_id: context.value.faceId!,
+        kling_face_start_time: context.value.faceStartTime!,
+        kling_face_end_time: context.value.faceEndTime!,
+        ai_provider: 'kling',
+        voiceConfigId: context.value.voice!.rawId || context.value.voice!.id.match(/[\w-]+$/)?.[0] || context.value.voice!.id,
+        pre_generated_audio: {
+          audioBase64: context.value.audioBase64!,
+          format: context.value.audioFormat!,
+        },
+        sound_end_time: context.value.audioDurationMs!,
+      }
+
+      const res = await createLipSyncTask(taskData)
+      if (res.code !== 0) {
+        throw new Error(res.msg || '任务创建失败')
+      }
+
+      setState('completed')
+      message.success('任务已提交，请在任务中心查看生成进度')
+
+    } catch (err) {
+      setError(err as Error)
+    }
+  }
+
+  /**
+   * 重试（从 failed 状态恢复）
+   */
+  function retry(): void {
+    if (!canRetry.value) {
+      message.warning('当前状态无法重试')
+      return
+    }
+
+    error.value = null
+    // 回到 idle 重新开始
+    setState('idle')
+  }
+
+  /**
+   * 重置到初始状态
+   */
+  function reset(): void {
+    state.value = 'idle'
+    context.value = { ...INITIAL_CONTEXT }
+    error.value = null
+    history.value = ['idle']
+  }
+
+  /**
+   * 获取执行状态（用于调试）
+   */
+  function getExecutionState(): ExecutionState {
+    return {
+      current: state.value,
+      history: [...history.value],
+      context: { ...context.value },
+      canNext: state.value === 'ready',
+      canRetry: canRetry.value,
+    }
+  }
+
+  // ========== 返回 API ==========
+  return {
+    // 状态
+    state,
+    context,
+    error,
+    history,
+
+    // 计算属性
+    stateLabel,
+    stateDescription,
+    progress,
+    currentStepIndex,
+    isBusy,
+    isReady,
+    isFailed,
+    isCompleted,
+    isTerminal,
+    canRetry,
+
+    // 方法
+    run,
+    createTask,
+    retry,
+    reset,
+    getExecutionState,
+  }
+}
--- a/frontend/app/web-gold/src/views/kling/hooks/useDigitalHumanGeneration.ts
+++ b/frontend/app/web-gold/src/views/kling/hooks/useDigitalHumanGeneration.ts
@@ -1,38 +1,34 @@
 /**
- * @fileoverview useDigitalHumanGeneration Hook - 数字人生成逻辑封装
- * @author Claude Code
+ * @fileoverview useDigitalHumanGeneration Hook - 数字人生成逻辑
+ *
+ * 重构后：不管理识别状态，只提供数据和操作方法
+ * 状态由 Pipeline 统一管理
 */

 import { ref, computed } from 'vue'
 import { message } from 'ant-design-vue'
 import type {
-  UseDigitalHumanGeneration,
  VideoState,
-  IdentifyState,
+  IdentifyResult,
  Video,
 } from '../types/identify-face'
 import { identifyUploadedVideo } from '@/api/kling'
 import { useUpload } from '@/composables/useUpload'

-/**
- * 数字人生成 Hook
- * 独立管理所有状态，不依赖外部状态
- */
-export function useDigitalHumanGeneration(): UseDigitalHumanGeneration {
-  // ==================== 响应式状态 ====================
-
+export function useDigitalHumanGeneration() {
+  // ========== 状态 ==========
  const videoState = ref<VideoState>({
    uploadedVideo: '',
    videoFile: null,
    previewVideoUrl: '',
    selectedVideo: null,
+    fileId: null,
    videoSource: null,
    selectorVisible: false,
  })

-  const identifyState = ref<IdentifyState>({
-    identifying: false,
-    identified: false,
+  // 识别结果数据（不含状态标志）
+  const identifyResult = ref<IdentifyResult>({
    sessionId: '',
    faceId: '',
    faceStartTime: 0,
@@ -40,24 +36,24 @@ export function useDigitalHumanGeneration(): UseDigitalHumanGeneration {
    videoFileId: null,
  })

-  // ==================== Upload Hook ====================
  const { upload } = useUpload()

-  // ==================== 计算属性 ====================
-
-  /**
-   * 人脸出现时长
-   */
-  const faceDuration = computed(() => {
-    return identifyState.value.faceEndTime - identifyState.value.faceStartTime
+  // ========== 计算属性 ==========
+  const faceDuration = computed(function() {
+    return identifyResult.value.faceEndTime - identifyResult.value.faceStartTime
  })

-  // ==================== 核心方法 ====================
+  const hasVideo = computed(function() {
+    return !!videoState.value.uploadedVideo || !!videoState.value.selectedVideo
+  })

-  /**
-   * 处理视频文件上传
-   */
-  const handleFileUpload = async (file: File): Promise<void> => {
+  const isIdentified = computed(function() {
+    return !!identifyResult.value.sessionId
+  })
+
+  // ========== 方法 ==========
+
+  async function handleFileUpload(file: File): Promise<void> {
    if (!file.name.match(/\.(mp4|mov)$/i)) {
      message.error('仅支持 MP4 和 MOV')
      return
@@ -68,168 +64,115 @@ export function useDigitalHumanGeneration(): UseDigitalHumanGeneration {
    videoState.value.selectedVideo = null
    videoState.value.previewVideoUrl = ''
    videoState.value.videoSource = 'upload'
-
-    resetIdentifyState()
-
-    await performFaceRecognition()
+    resetIdentifyResult()
  }

-  /**
-   * 处理从素材库选择视频
-   */
-  const handleVideoSelect = (video: Video): void => {
+  async function handleVideoSelect(video: Video): Promise<void> {
    videoState.value.selectedVideo = video
    videoState.value.uploadedVideo = video.fileUrl
    videoState.value.videoFile = null
    videoState.value.videoSource = 'select'
    videoState.value.selectorVisible = false
-    resetIdentifyState()
-    identifyState.value.videoFileId = video.id
+    resetIdentifyResult()
+    identifyResult.value.videoFileId = video.fileId
  }

  /**
   * 执行人脸识别
+   * 返回识别结果供 Pipeline 使用
   */
-  const performFaceRecognition = async (): Promise<void> => {
+  async function performFaceRecognition(): Promise<IdentifyResult> {
    const hasUploadFile = videoState.value.videoFile
    const hasSelectedVideo = videoState.value.selectedVideo

    if (!hasUploadFile && !hasSelectedVideo) {
-      return
+      throw new Error('请先选择视频')
    }

-    identifyState.value.identifying = true
-
-    try {
-      let res
-      if (hasSelectedVideo) {
-        res = await identifyUploadedVideo(hasSelectedVideo)
-        identifyState.value.videoFileId = hasSelectedVideo.id
-      } else {
-        // 处理文件上传（提取封面）
-        const file = hasUploadFile!
-        let coverBase64 = null
-        try {
-          const { extractVideoCover } = await import('@/utils/video-cover')
-          const cover = await extractVideoCover(file, {
-            maxWidth: 800,
-            quality: 0.8
-          })
-          coverBase64 = cover.base64
-        } catch {
-          // 封面提取失败不影响主流程
-        }
-
-        // 使用useUpload Hook上传文件
-        const fileId = await upload(file, {
-          fileCategory: 'video',
-          groupId: null, // 数字人模块不使用groupId
-          coverBase64,
-          onStart: () => {},
-          onProgress: () => {},
-          onSuccess: () => {
-            message.success('文件上传成功')
-          },
-          onError: (err: Error) => {
-            message.error(err.message || '上传失败')
-          }
-        })
-
-        // 生成播放链接
-        // TODO: 获取播放链接逻辑
-
-        res = {
-          success: true,
-          data: {
-            fileId,
-            videoUrl: '', // TODO: 需要获取实际URL
-            sessionId: '', // TODO: 需要实际识别
-            faceId: null,
-            startTime: 0,
-            endTime: 0
-          }
-        }
-        identifyState.value.videoFileId = fileId
+    if (hasSelectedVideo) {
+      const res = await identifyUploadedVideo(hasSelectedVideo) as {
+        success: boolean;
+        data: { sessionId: string; faceId: string | null; startTime: number; endTime: number }
+      }
+      identifyResult.value.videoFileId = hasSelectedVideo.fileId
+      identifyResult.value.sessionId = res.data.sessionId
+      identifyResult.value.faceId = res.data.faceId || ''
+      identifyResult.value.faceStartTime = res.data.startTime || 0
+      identifyResult.value.faceEndTime = res.data.endTime || 0
+    } else {
+      const file = hasUploadFile!
+      let coverBase64 = null
+      try {
+        const { extractVideoCover } = await import('@/utils/video-cover')
+        const cover = await extractVideoCover(file, { maxWidth: 800, quality: 0.8 })
+        coverBase64 = cover.base64
+      } catch {
+        // 封面提取失败不影响主流程
      }

-      identifyState.value.sessionId = res.data.sessionId
-      identifyState.value.faceId = res.data.faceId
-      identifyState.value.faceStartTime = res.data.startTime || 0
-      identifyState.value.faceEndTime = res.data.endTime || 0
-      identifyState.value.identified = true
+      const fileId = await upload(file, {
+        fileCategory: 'video',
+        groupId: null,
+        coverBase64,
+        onStart: function() {},
+        onProgress: function() {},
+        onSuccess: function() {},
+        onError: function(err: Error) {
+          message.error(err.message || '上传失败')
+        }
+      })

-      const durationSec = faceDuration.value / 1000
-      const suggestedMaxChars = Math.floor(durationSec * 3.5)
-      message.success(`识别完成！人脸出现时长约 ${durationSec.toFixed(1)} 秒，建议文案不超过 ${suggestedMaxChars} 字`)
-    } catch (error: any) {
-      message.error(error.message || '识别失败')
-      throw error
-    } finally {
-      identifyState.value.identifying = false
+      identifyResult.value.videoFileId = fileId
+      // 上传后需要再调用识别接口获取人脸信息
+      // 暂时清空，等待后续识别
+      identifyResult.value.sessionId = ''
+      identifyResult.value.faceId = ''
+      identifyResult.value.faceStartTime = 0
+      identifyResult.value.faceEndTime = 0
    }
+
+    return { ...identifyResult.value }
  }

-
-  /**
-   * 重置视频状态
-   */
-  const resetVideoState = (): void => {
+  function resetVideoState(): void {
    videoState.value.uploadedVideo = ''
    videoState.value.videoFile = null
    videoState.value.selectedVideo = null
+    videoState.value.fileId = null
    videoState.value.videoSource = null
    videoState.value.previewVideoUrl = ''
    videoState.value.selectorVisible = false
-
-    resetIdentifyState()
+    resetIdentifyResult()
  }

-  /**
-   * 获取视频预览 URL
-   */
-  const getVideoPreviewUrl = (video: Video): string => {
+  function getVideoPreviewUrl(video: Video): string {
    if (video.coverBase64) {
-      if (!video.coverBase64.startsWith('data:')) {
-        return `data:image/jpeg;base64,${video.coverBase64}`
-      }
-      return video.coverBase64
+      return video.coverBase64.startsWith('data:')
+        ? video.coverBase64
+        : `data:image/jpeg;base64,${video.coverBase64}`
    }
-
-    if (video.previewUrl) {
-      return video.previewUrl
-    }
-
-    if (video.coverUrl) {
-      return video.coverUrl
-    }
-
+    if (video.previewUrl) return video.previewUrl
+    if (video.coverUrl) return video.coverUrl
    return 'data:image/svg+xml;base64,PHN2ZyB3aWR0aD0iMjAwIiBoZWlnaHQ9IjExMCIgdmlld0JveD0iMCAwIDIwMCAxMTAiIGZpbGw9Im5vbmUiIHhtbG5zPSJodHRwOi8vd3d3LnczLm9yZy8yMDAwL3N2ZyI+CjxyZWN0IHdpZHRoPSIyMDAiIGhlaWdodD0iMTEwIiBmaWxsPSIjMzc0MTUxIi8+CjxwYXRoIGQ9Ik04NSA0NUwxMTUgNjVMMTA1IDg1TDc1IDc1TDg1IDQ1WiIgZmlsbD0iIzU3MjY1MSIvPgo8L3N2Zz4K'
  }

-  /**
-   * 重置识别状态
-   */
-  const resetIdentifyState = (): void => {
-    identifyState.value.identified = false
-    identifyState.value.sessionId = ''
-    identifyState.value.faceId = ''
-    identifyState.value.videoFileId = null
+  function resetIdentifyResult(): void {
+    identifyResult.value.sessionId = ''
+    identifyResult.value.faceId = ''
+    identifyResult.value.videoFileId = null
  }

  return {
-    // 响应式状态
    videoState,
-    identifyState,
-
-    // 计算属性
+    identifyResult,
+    hasVideo,
+    isIdentified,
    faceDuration,
-
-    // 方法
    handleFileUpload,
    handleVideoSelect,
    performFaceRecognition,
    resetVideoState,
-    resetIdentifyState,
+    resetIdentifyResult,
    getVideoPreviewUrl,
  }
 }
--- a/frontend/app/web-gold/src/views/kling/hooks/useIdentifyFaceController.ts
+++ b/frontend/app/web-gold/src/views/kling/hooks/useIdentifyFaceController.ts
@@ -1,416 +1,344 @@
 /**
- * @fileoverview useIdentifyFaceController Hook - 主控制器 Hook
- * @author Claude Code
+ * @fileoverview useIdentifyFaceController Hook - 主控制器（重构版）
+ *
+ * 设计理念：
+ * - 所有操作统一通过 Pipeline 状态机
+ * - 移除独立的 identifyState，使用 pipeline 状态
+ * - 点击"生成配音" → 运行到 ready 状态
+ * - 点击"生成数字人视频" → 从 ready 继续 → completed
+ *
+ * 模块依赖关系:
+ *   ┌─────────────────────────────────────────────────┐
+ *   │            useIdentifyFaceController              │
+ *   │   ┌──────────────┐  ┌──────────────┐  ┌───────────┐│
+ *   │   │   Voice      │  │  Digital     │  │ Pipeline  ││
+ *   │   │   Generation  │  │  Human       │  │           ││
+ *   │   │              │  │  Generation  │  │  状态机    ││
+ *   │   └──────────────┘  └──────────────┘  └───────────┘│
+ *   └─────────────────────────────────────────────────┘
 */

-import { ref, computed, watch } from 'vue'
+import { computed } from 'vue'
 import { message } from 'ant-design-vue'
 import type {
-  UseIdentifyFaceController,
-  LipSyncTaskData,
-  MaterialValidation,
+  VoiceMeta,
 } from '../types/identify-face'
-// @ts-ignore
-import { createLipSyncTask } from '@/api/kling'
-
-// 导入子 Hooks
 import { useVoiceGeneration } from './useVoiceGeneration'
 import { useDigitalHumanGeneration } from './useDigitalHumanGeneration'
+import { useSimplePipeline } from './pipeline/useSimplePipeline'
+
+// ==================== 常量 ====================
+const SPEECH_RATE_MARKS = { 0.5: '0.5x', 1: '1x', 1.5: '1.5x', 2: '2x' }
+const MAX_TEXT_LENGTH = 4000

 /**
- * 识别控制器 Hook - 充当协调器
- * 内部直接创建和管理两个子 Hook
+ * 主控制器 Hook
 */
-export function useIdentifyFaceController(): UseIdentifyFaceController {
-  // ==================== 创建子 Hooks 并解构 ====================
+export function useIdentifyFaceController() {
+  // 子 Hooks
+  const voice = useVoiceGeneration()
+  const digitalHuman = useDigitalHumanGeneration()

-  // 1. 语音生成 Hook - 解构响应式变量
-  const {
-    ttsText,
-    speechRate,
-    selectedVoiceMeta,
-    audioState,
-    canGenerateAudio,
-    suggestedMaxChars,
-    generateAudio,
-    resetAudioState,
-  } = useVoiceGeneration()
-
-  // 2. 数字人生成 Hook - 解构响应式变量
-  const {
-    videoState,
-    identifyState,
-    faceDuration,
-    performFaceRecognition,
-    handleFileUpload,
-    handleVideoSelect: _handleVideoSelect,
-    getVideoPreviewUrl,
-    resetVideoState,
-    resetIdentifyState,
-  } = useDigitalHumanGeneration()
-
-  // 3. Controller 统一管理跨 Hook 的状态
-  const materialValidation = ref<MaterialValidation>({
-    videoDuration: 0,
-    audioDuration: 0,
-    isValid: false,
-    showDetails: false,
+  // Pipeline 流程配置
+  const pipeline = useSimplePipeline({
+    uploadVideo: async (_file: File) => {
+      // 上传已经在 handleFileUpload 中处理
+      return digitalHuman.identifyResult.value.videoFileId || ''
+    },
+    recognizeFromLibrary: async (video: any) => {
+      await digitalHuman.handleVideoSelect(video)
+      const result = await digitalHuman.performFaceRecognition()
+      return {
+        sessionId: result.sessionId,
+        faceId: result.faceId,
+        startTime: result.faceStartTime,
+        endTime: result.faceEndTime,
+        duration: digitalHuman.faceDuration.value,
+      }
+    },
+    recognizeUploaded: async (_fileId: string | number) => {
+      const result = await digitalHuman.performFaceRecognition()
+      return {
+        sessionId: result.sessionId,
+        faceId: result.faceId,
+        startTime: result.faceStartTime,
+        endTime: result.faceEndTime,
+        duration: digitalHuman.faceDuration.value,
+      }
+    },
+    generateAudio: async (text: string, voiceMeta: any, speechRate: number) => {
+      voice.ttsText.value = text
+      voice.selectedVoiceMeta.value = voiceMeta
+      voice.speechRate.value = speechRate
+      await voice.generateAudio()
+      const audio = voice.audioState.value.generated!
+      return {
+        audioBase64: audio.audioBase64,
+        format: audio.format || 'mp3',
+        durationMs: voice.audioState.value.durationMs,
+      }
+    },
+    createTask: async () => {
+      // 任务创建在 Pipeline 中处理
+    },
  })

-  // 4. 监听音频状态变化，自动触发素材校验
-  watch(
-    () => audioState.value.generated && audioState.value.durationMs > 0,
-    (newVal, oldVal) => {
-      if (newVal && !oldVal) {
-        // 音频生成完成，获取视频时长并校验
-        const videoDurationMs = faceDuration.value || 0
-        const audioDurationMs = audioState.value.durationMs
-
-        if (videoDurationMs > 0) {
-          validateMaterialDuration(videoDurationMs, audioDurationMs)
-        }
-      }
-    },
-    { flush: 'post' }
-  )
-
-  // 5. 监听人脸识别状态变化，更新素材校验的视频时长
-  watch(
-    () => identifyState.value.identified,
-    (newVal, oldVal) => {
-      if (newVal && !oldVal) {
-        // 人脸识别成功，获取视频时长
-        const videoDurationMs = faceDuration.value
-
-        // 如果已有音频，则重新校验
-        if (audioState.value.generated && audioState.value.durationMs > 0) {
-          const audioDurationMs = audioState.value.durationMs
-          validateMaterialDuration(videoDurationMs, audioDurationMs)
-        } else {
-          // 否则只更新视频时长
-          materialValidation.value.videoDuration = videoDurationMs
-        }
-      }
-    },
-    { flush: 'post' }
-  )
  // ==================== 计算属性 ====================

-  /**
-   * 是否可以生成数字人视频（综合检查）
-   */
-  const canGenerate = computed(() => {
-    const hasText = ttsText.value.trim()
-    const hasVoice = selectedVoiceMeta.value
-    const hasVideo = videoState.value.uploadedVideo || videoState.value.selectedVideo
+  /** 是否可以生成数字人视频 */
+  const canGenerate = computed((): boolean => {
+    // Pipeline 运行中禁用
+    if (pipeline.isBusy.value) return false

-    // 音频校验：只有生成过音频后才需要校验通过
-    const audioValidated = !audioState.value.generated || audioState.value.validationPassed
-    // 素材校验：只有进行过校验后才需要校验通过
-    const materialValidated = materialValidation.value.videoDuration === 0 || materialValidation.value.isValid
+    const hasText = voice.ttsText.value.trim()
+    const hasVoice = voice.selectedVoiceMeta.value
+    const hasVideo = digitalHuman.videoState.value.uploadedVideo || digitalHuman.videoState.value.selectedVideo
+    const hasBasicConfig = hasText && hasVoice && hasVideo

-    return !!(hasText && hasVoice && hasVideo && audioValidated && materialValidated)
+    // 未识别或未到 ready 状态需要基础配置
+    if (!pipeline.isReady.value) return !!hasBasicConfig
+
+    // 已到 ready 状态可以生成
+    return true
  })

-  /**
-   * 最大的文本长度
-   */
+  /** 最大文本长度（根据人脸时长动态计算） */
  const maxTextLength = computed(() => {
-    if (!identifyState.value.identified || faceDuration.value <= 0) {
-      return 4000
-    }
-    return Math.min(4000, Math.floor(suggestedMaxChars.value * 1.2))
+    const faceDuration = digitalHuman.faceDuration.value
+    if (faceDuration <= 0) return MAX_TEXT_LENGTH
+    return Math.min(MAX_TEXT_LENGTH, Math.floor(voice.suggestedMaxChars.value * 1.2))
  })

-  /**
-   * 文本框占位符
-   */
+  /** 文本框占位符提示 */
  const textareaPlaceholder = computed(() => {
-    if (identifyState.value.identified && faceDuration.value > 0) {
-      return `请输入文案，建议不超过${suggestedMaxChars.value}字以确保与视频匹配`
+    const faceDuration = digitalHuman.faceDuration.value
+    if (faceDuration > 0) {
+      return `请输入文案，建议不超过${voice.suggestedMaxChars.value}字以确保与视频匹配`
    }
    return '请输入你想让角色说话的内容'
  })

-  /**
-   * 语速标记
-   */
-  const speechRateMarks = { 0.5: '0.5x', 1: '1x', 1.5: '1.5x', 2: '2x' }
+  /** 语速显示文本 */
+  const speechRateDisplay = computed(() => `${voice.speechRate.value.toFixed(1)}x`)
+
+  /** 人脸时长显示（秒） */
+  const faceDurationSec = computed(() => (digitalHuman.faceDuration.value / 1000).toFixed(1))
+
+  /** 音频时长显示（秒） */
+  const audioDurationSec = computed(() => (voice.audioState.value.durationMs / 1000).toFixed(1))
+
+  /** 音频播放 URL */
+  const audioUrl = computed(() => {
+    const audio = voice.audioState.value.generated
+    if (!audio) return ''
+    return audio.audioBase64 ? `data:audio/mp3;base64,${audio.audioBase64}` : audio.audioUrl || ''
+  })

  /**
-   * 语速显示
+   * 校验是否通过
+   * 规则：音频时长 <= 人脸时长
   */
-  const speechRateDisplay = computed(() => `${speechRate.value.toFixed(1)}x`)
+  const validationPassed = computed(() => {
+    const faceDuration = digitalHuman.faceDuration.value
+    const audioDuration = voice.audioState.value.durationMs
+    return audioDuration <= faceDuration
+  })

-  // ==================== 业务流程方法 ====================
+  // ==================== 业务方法 ====================

  /**
-   * 生成数字人视频
+   * 重置所有状态
   */
-  const generateDigitalHuman = async (): Promise<void> => {
+  function resetAllStates(): void {
+    voice.resetAudioState()
+    digitalHuman.resetVideoState()
+    pipeline.reset()
+  }
+
+  /**
+   * 生成配音 - 运行 Pipeline 到 ready 状态
+   */
+  async function generateAudio(): Promise<void> {
+    const hasVideo = digitalHuman.videoState.value.uploadedVideo || digitalHuman.videoState.value.selectedVideo
+    const hasText = voice.ttsText.value.trim()
+    const hasVoice = voice.selectedVoiceMeta.value
+
+    if (!hasText) {
+      message.warning('请输入文案内容')
+      return
+    }
+
+    if (!hasVoice) {
+      message.warning('请选择音色')
+      return
+    }
+
+    if (!hasVideo) {
+      message.warning('请先选择视频')
+      return
+    }
+
+    try {
+      // 运行流程到 ready 状态（包含识别、生成、校验）
+      await pipeline.run({
+        videoFile: digitalHuman.videoState.value.videoFile,
+        selectedVideo: digitalHuman.videoState.value.selectedVideo,
+        text: voice.ttsText.value,
+        voice: voice.selectedVoiceMeta.value,
+        speechRate: voice.speechRate.value,
+      })
+    } catch {
+      // 错误已在 Pipeline 中处理
+    }
+  }
+
+  /**
+   * 生成数字人视频 - 从 ready 状态继续到 completed
+   */
+  async function generateDigitalHuman(): Promise<void> {
    if (!canGenerate.value) {
      message.warning('请先完成配置')
      return
    }

-    const text = ttsText.value.trim()
+    const text = voice.ttsText.value.trim()
+    const voiceMeta = voice.selectedVoiceMeta.value
+
    if (!text) {
      message.warning('请输入文案内容')
      return
    }

-    const voice = selectedVoiceMeta.value
-    if (!voice) {
+    if (!voiceMeta) {
      message.warning('请选择音色')
      return
    }

    try {
-      // 如果未识别，先进行人脸识别
-      if (!identifyState.value.identified) {
-        const hasUploadFile = videoState.value.videoFile
-        const hasSelectedVideo = videoState.value.selectedVideo
-
-        if (!hasUploadFile && !hasSelectedVideo) {
-          message.warning('请先选择或上传视频')
-          return
-        }
-
-        try {
-          await performFaceRecognition()
-          message.success('人脸识别完成')
-        } catch (error) {
-          return
-        }
+      // 如果还没到 ready 状态，先运行到 ready
+      if (!pipeline.isReady.value) {
+        await pipeline.run({
+          videoFile: digitalHuman.videoState.value.videoFile,
+          selectedVideo: digitalHuman.videoState.value.selectedVideo,
+          text,
+          voice: voiceMeta,
+          speechRate: voice.speechRate.value,
+        })
      }

-      const videoFileId = identifyState.value.videoFileId
-
-      const taskData: LipSyncTaskData = {
-        taskName: `数字人任务_${Date.now()}`,
-        videoFileId: videoFileId!,
-        inputText: ttsText.value,
-        speechRate: speechRate.value,
-        volume: 0,
-        guidanceScale: 1,
-        seed: 8888,
-        kling_session_id: identifyState.value.sessionId,
-        kling_face_id: identifyState.value.faceId,
-        kling_face_start_time: identifyState.value.faceStartTime,
-        kling_face_end_time: identifyState.value.faceEndTime,
-        ai_provider: 'kling',
-        voiceConfigId: voice.rawId || extractIdFromString(voice.id),
+      // 如果到达 ready 状态，创建任务
+      if (pipeline.isReady.value) {
+        await pipeline.createTask()
+        // 任务提交成功后，重置所有状态
+        resetAllStates()
      }
-
-      if (!taskData.voiceConfigId) {
-        message.warning('音色配置无效')
-        return
-      }
-
-      // 如果有预生成的音频，添加到任务数据中
-      if (audioState.value.generated && audioState.value.durationMs > 0) {
-        taskData.pre_generated_audio = {
-          audioBase64: audioState.value.generated.audioBase64,
-          format: audioState.value.generated.format || 'mp3',
-        }
-
-        taskData.sound_end_time = audioState.value.durationMs
-      }
-
-      const res = await createLipSyncTask(taskData)
-
-      if (res.code === 0) {
-        message.success('任务已提交到任务中心，请前往查看')
-      } else {
-        throw new Error(res.msg || '任务创建失败')
-      }
-    } catch (error: any) {
-      message.error(error.message || '任务提交失败')
+    } catch {
+      // 错误已在 Pipeline 中处理
    }
  }

  /**
   * 更换视频
   */
-  const replaceVideo = (): void => {
-    if (videoState.value.videoSource === 'upload') {
-      videoState.value.videoFile = null
-      videoState.value.uploadedVideo = ''
-    } else {
-      videoState.value.selectedVideo = null
-      videoState.value.videoFile = null
-      videoState.value.uploadedVideo = ''
-    }
-
-    // 重置所有状态
-    resetVideoState()
-    resetAudioState()
+  function replaceVideo(): void {
+    digitalHuman.resetVideoState()
+    voice.resetAudioState()
+    pipeline.reset()
  }

-  /**
-   * 处理音色选择
-   */
-  const handleVoiceSelect = (voice: any): void => {
-    selectedVoiceMeta.value = voice
+  // ==================== 事件处理方法 ====================
+
+  function handleVoiceSelect(voiceMeta: VoiceMeta): void {
+    voice.selectedVoiceMeta.value = voiceMeta
  }

-  /**
-   * 处理文件选择
-   */
-  const handleFileSelect = (event: Event): void => {
-    const input = event.target as HTMLInputElement
-    const file = input.files?.[0]
-    if (file) {
-      handleFileUpload(file)
-    }
+  function handleFileSelect(event: Event): void {
+    const file = (event.target as HTMLInputElement).files?.[0]
+    if (file) digitalHuman.handleFileUpload(file)
  }

-  /**
-   * 处理拖拽上传
-   */
-  const handleDrop = (event: DragEvent): void => {
+  function handleDrop(event: DragEvent): void {
    event.preventDefault()
    const file = event.dataTransfer?.files[0]
-    if (file) {
-      handleFileUpload(file)
-    }
+    if (file) digitalHuman.handleFileUpload(file)
  }

-  /**
-   * 触发文件选择
-   */
-  const triggerFileSelect = (): void => {
-    const fileInput = document.querySelector('input[type="file"]') as HTMLInputElement
-    fileInput?.click()
+  function triggerFileSelect(): void {
+    document.querySelector<HTMLInputElement>('input[type="file"]')?.click()
  }

-  /**
-   * 选择上传模式
-   */
-  const handleSelectUpload = (): void => {
-    videoState.value.videoSource = 'upload'
-    videoState.value.selectedVideo = null
-    resetIdentifyState()
+  function handleSelectUpload(): void {
+    digitalHuman.videoState.value.videoSource = 'upload'
+    digitalHuman.videoState.value.selectedVideo = null
+    digitalHuman.resetIdentifyResult()
+    pipeline.reset()
  }

-  /**
-   * 从素材库选择
-   */
-  const handleSelectFromLibrary = (): void => {
-    videoState.value.videoSource = 'select'
-    videoState.value.videoFile = null
-    videoState.value.uploadedVideo = ''
-    videoState.value.selectorVisible = true
+  function handleSelectFromLibrary(): void {
+    digitalHuman.videoState.value.videoSource = 'select'
+    digitalHuman.videoState.value.videoFile = null
+    digitalHuman.videoState.value.uploadedVideo = ''
+    digitalHuman.videoState.value.selectorVisible = true
+    pipeline.reset()
  }

-  /**
-   * 处理视频选择器选择
-   */
-  const handleVideoSelect = (video: any): void => {
-    _handleVideoSelect(video)
+  async function handleVideoSelect(video: any): Promise<void> {
+    await digitalHuman.handleVideoSelect(video)
  }

-  /**
-   * 简化文案
-   */
-  const handleSimplifyScript = (): void => {
-    const textarea = document.querySelector('.tts-textarea textarea') as HTMLTextAreaElement
-    if (textarea) {
-      textarea.focus()
-      textarea.scrollIntoView({ behavior: 'smooth', block: 'center' })
-    }
+  function handleVideoLoaded(videoUrl: string): void {
+    digitalHuman.videoState.value.previewVideoUrl = videoUrl
  }

-  /**
-   * 处理视频加载
-   */
-  const handleVideoLoaded = (videoUrl: string): void => {
-    videoState.value.previewVideoUrl = videoUrl
-  }
+  // ==================== UI 工具方法 ====================

-  // ==================== UI 辅助方法 ====================
-
-  /**
-   * 格式化时长
-   */
-  const formatDuration = (seconds: number): string => {
+  function formatDuration(seconds: number): string {
    if (!seconds) return '--:--'
-    const minutes = Math.floor(seconds / 60)
-    const remainingSeconds = Math.floor(seconds % 60)
-    return `${String(minutes).padStart(2, '0')}:${String(remainingSeconds).padStart(2, '0')}`
+    const mins = Math.floor(seconds / 60)
+    const secs = Math.floor(seconds % 60)
+    return `${String(mins).padStart(2, '0')}:${String(secs).padStart(2, '0')}`
  }

-  /**
-   * 格式化文件大小
-   */
-  const formatFileSize = (bytes: number): string => {
+  function formatFileSize(bytes: number): string {
    if (!bytes) return '0 B'
    const units = ['B', 'KB', 'MB', 'GB']
    let size = bytes
-    let unitIndex = 0
-    while (size >= 1024 && unitIndex < units.length - 1) {
+    let idx = 0
+    while (size >= 1024 && idx < units.length - 1) {
      size /= 1024
-      unitIndex++
+      idx++
    }
-    return `${size.toFixed(1)} ${units[unitIndex]}`
+    return `${size.toFixed(1)} ${units[idx]}`
  }

-  /**
-   * 重置素材校验状态
-   */
-  const resetMaterialValidation = (): void => {
-    materialValidation.value.videoDuration = 0
-    materialValidation.value.audioDuration = 0
-    materialValidation.value.isValid = false
-  }
-
-  /**
-   * 验证素材时长
-   * 视频时长必须大于音频时长
-   */
-  const validateMaterialDuration = (videoDurationMs: number, audioDurationMs: number): boolean => {
-    materialValidation.value.videoDuration = videoDurationMs
-    materialValidation.value.audioDuration = audioDurationMs
-    materialValidation.value.isValid = videoDurationMs > audioDurationMs
-
-    if (!materialValidation.value.isValid) {
-      const videoSec = (videoDurationMs / 1000).toFixed(1)
-      const audioSec = (audioDurationMs / 1000).toFixed(1)
-      message.warning(`素材校验失败：视频时长（${videoSec}s）必须大于音频时长（${audioSec}s）`)
-    }
-
-    return materialValidation.value.isValid
-  }
+  // ==================== 返回接口 ====================

  return {
-    // ==================== 语音生成相关 ====================
-    ttsText,
-    speechRate,
-    selectedVoiceMeta,
-    audioState,
-    canGenerateAudio,
-    suggestedMaxChars,
+    // 语音生成模块
+    ttsText: voice.ttsText,
+    speechRate: voice.speechRate,
+    selectedVoiceMeta: voice.selectedVoiceMeta,
+    audioState: voice.audioState,
+    canGenerateAudio: voice.canGenerateAudio,
+    suggestedMaxChars: voice.suggestedMaxChars,
    generateAudio,
-    resetAudioState,
+    resetAudioState: voice.resetAudioState,

-    // ==================== 数字人生成相关 ====================
-    videoState,
-    identifyState,
-    materialValidation,
-    faceDuration,
-    performFaceRecognition,
-    handleFileUpload,
-    getVideoPreviewUrl,
-    resetVideoState,
-    resetIdentifyState,
-    resetMaterialValidation,
-    validateMaterialDuration,
+    // 数字人生成模块
+    videoState: digitalHuman.videoState,
+    identifyResult: digitalHuman.identifyResult,
+    isIdentified: digitalHuman.isIdentified,
+    faceDuration: digitalHuman.faceDuration,
+    handleFileUpload: digitalHuman.handleFileUpload,
+    getVideoPreviewUrl: digitalHuman.getVideoPreviewUrl,
+    resetVideoState: digitalHuman.resetVideoState,
+    resetIdentifyResult: digitalHuman.resetIdentifyResult,

-    // ==================== 业务流程方法 ====================
+    // 业务方法
    generateDigitalHuman,
    replaceVideo,

-    // ==================== 事件处理方法 ====================
+    // 事件处理
    handleVoiceSelect,
    handleFileSelect,
    handleDrop,
@@ -418,26 +346,35 @@ export function useIdentifyFaceController(): UseIdentifyFaceController {
    handleSelectUpload,
    handleSelectFromLibrary,
    handleVideoSelect,
-    handleSimplifyScript,
    handleVideoLoaded,

-    // ==================== UI 辅助方法 ====================
+    // UI 工具
    formatDuration,
    formatFileSize,

-    // ==================== 计算属性 ====================
+    // 计算属性
    canGenerate,
    maxTextLength,
    textareaPlaceholder,
-    speechRateMarks,
+    speechRateMarks: SPEECH_RATE_MARKS,
    speechRateDisplay,
+    faceDurationSec,
+    audioDurationSec,
+    audioUrl,
+    validationPassed,
+
+    // Pipeline 状态（单一状态源）
+    pipelineState: pipeline.state,
+    pipelineStateLabel: pipeline.stateLabel,
+    pipelineStateDescription: pipeline.stateDescription,
+    isPipelineBusy: pipeline.isBusy,
+    isPipelineReady: pipeline.isReady,
+    isPipelineFailed: pipeline.isFailed,
+    isPipelineCompleted: pipeline.isCompleted,
+    pipelineProgress: pipeline.progress,
+    pipelineCurrentStepIndex: pipeline.currentStepIndex,
+    pipelineError: pipeline.error,
+    retryPipeline: pipeline.retry,
+    resetPipeline: pipeline.reset,
  }
 }
-
-/**
- * 从字符串中提取ID
- */
-function extractIdFromString(str: string): string {
-  const match = str.match(/[\w-]+$/)
-  return match ? match[0] : str
-}
--- a/frontend/app/web-gold/src/views/kling/hooks/useVoiceGeneration.ts
+++ b/frontend/app/web-gold/src/views/kling/hooks/useVoiceGeneration.ts
@@ -1,6 +1,5 @@
 /**
- * @fileoverview useVoiceGeneration Hook - 语音生成逻辑封装
- * @author Claude Code
+ * @fileoverview useVoiceGeneration Hook - 语音生成逻辑
 */

 import { ref, computed } from 'vue'
@@ -11,59 +10,41 @@ import type {
  VoiceMeta,
  AudioData,
 } from '../types/identify-face'
-// @ts-ignore
 import { VoiceService } from '@/api/voice'
 import { DEFAULT_VOICE_PROVIDER } from '@/config/voiceConfig'

-/**
- * 语音生成 Hook
- * 独立管理所有状态，不依赖外部状态
- */
-export function useVoiceGeneration(): UseVoiceGeneration {
-  // ==================== 响应式状态 ====================
+// ========== 常量 ==========
+const DEFAULT_MAX_TEXT_LENGTH = 4000
+const DEFAULT_SPEECH_RATE = 1.0

+export function useVoiceGeneration(): UseVoiceGeneration {
+  // ========== 状态 ==========
  const ttsText = ref<string>('')
-  const speechRate = ref<number>(1.0)
+  const speechRate = ref<number>(DEFAULT_SPEECH_RATE)
  const selectedVoiceMeta = ref<VoiceMeta | null>(null)
  const audioState = ref<AudioState>({
    generated: null,
    durationMs: 0,
-    validationPassed: false,
    generating: false,
  })

-  // ==================== 计算属性 ====================
-
-  /**
-   * 是否可以生成配音
-   */
-  const canGenerateAudio = computed(() => {
-    const hasText = ttsText.value.trim()
-    const hasVoice = selectedVoiceMeta.value
-    const hasVideo = true // 语音生成不依赖视频状态
-    return !!(hasText && hasVoice && hasVideo && !audioState.value.generating)
+  // ========== 计算属性 ==========
+  const canGenerateAudio = computed(function() {
+    return !!(ttsText.value.trim() && selectedVoiceMeta.value && !audioState.value.generating)
  })

-  /**
-   * 建议的最大字符数（需要从外部传入）
-   */
-  const suggestedMaxChars = computed(() => {
-    // 默认为 4000，需要从外部设置
-    return 4000
+  const suggestedMaxChars = computed(function() {
+    return DEFAULT_MAX_TEXT_LENGTH
  })

-  // ==================== 核心方法 ====================
+  // ========== 方法 ==========

-  /**
-   * 生成配音
-   */
-  const generateAudio = async (): Promise<void> => {
+  async function generateAudio(): Promise<void> {
    const voice = selectedVoiceMeta.value
    if (!voice) {
      message.warning('请选择音色')
      return
    }
-
    if (!ttsText.value.trim()) {
      message.warning('请输入文案内容')
      return
@@ -74,8 +55,8 @@ export function useVoiceGeneration(): UseVoiceGeneration {
    try {
      const params = {
        inputText: ttsText.value,
-        voiceConfigId: voice.rawId || extractIdFromString(voice.id),
-        speechRate: speechRate.value || 1.0,
+        voiceConfigId: voice.rawId ?? extractIdFromString(voice.id),
+        speechRate: speechRate.value,
        audioFormat: 'mp3' as const,
        providerType: DEFAULT_VOICE_PROVIDER,
      }
@@ -84,144 +65,104 @@ export function useVoiceGeneration(): UseVoiceGeneration {

      if (res.code === 0) {
        const audioData = res.data as AudioData
-
        if (!audioData.audioBase64) {
-          throw new Error('未收到音频数据，无法进行时长解析')
+          throw new Error('未收到音频数据')
        }

        audioState.value.generated = audioData
-
-        try {
-          // 解析音频时长
-          audioState.value.durationMs = await parseAudioDuration(audioData.audioBase64)
-
-          // 验证音频时长
-          validateAudioDuration()
-
-          message.success('配音生成成功！')
-        } catch (error) {
-          message.error('音频解析失败，请重新生成配音')
-          audioState.value.durationMs = 0
-          audioState.value.generated = null
-          audioState.value.validationPassed = false
-        }
+        audioState.value.durationMs = await parseAudioDuration(audioData.audioBase64)
+        message.success('配音生成成功！')
      } else {
        throw new Error(res.msg || '配音生成失败')
      }
-    } catch (error: any) {
-      message.error(error.message || '配音生成失败')
+    } catch (error: unknown) {
+      const err = error as Error
+      message.error(err.message || '配音生成失败')
+      audioState.value.generated = null
+      audioState.value.durationMs = 0
    } finally {
      audioState.value.generating = false
    }
  }

  /**
-   * 解析音频时长
+   * 解析音频时长（浏览器环境）
+   * 使用 HTML5 Audio API，添加安全边距避免精度误差
   */
-  const parseAudioDuration = async (base64Data: string): Promise<number> => {
-    return new Promise((resolve, reject) => {
-      try {
-        const base64 = base64Data.includes(',') ? base64Data.split(',')[1] : base64Data
+  async function parseAudioDuration(base64Data: string): Promise<number> {
+    const base64 = base64Data.includes(',') ? base64Data.split(',')[1] : base64Data
+    const binaryString = window.atob(base64)
+    const bytes = new Uint8Array(binaryString.length)
+    for (let i = 0; i < bytes.length; i++) {
+      bytes[i] = binaryString.charCodeAt(i)
+    }

-        const binaryString = window.atob(base64)
-        const bytes = new Uint8Array(binaryString.length)
-        for (let i = 0; i < binaryString.length; i++) {
-          bytes[i] = binaryString.charCodeAt(i)
+    return new Promise<number>(function(resolve, reject) {
+      const blob = new Blob([bytes], { type: 'audio/mp3' })
+      const audio = new Audio()
+      const objectUrl = URL.createObjectURL(blob)
+
+      const timeoutId = setTimeout(function() {
+        cleanup()
+        reject(new Error('音频时长解析超时'))
+      }, 10000)
+
+      function cleanup() {
+        clearTimeout(timeoutId)
+        URL.revokeObjectURL(objectUrl)
+        audio.removeEventListener('loadedmetadata', onLoadedMetadata)
+        audio.removeEventListener('error', onError)
+        audio.removeEventListener('canplay', onLoadedMetadata)
+      }
+
+      function onLoadedMetadata() {
+        const duration = audio.duration
+        if (!isFinite(duration) || duration <= 0) {
+          cleanup()
+          reject(new Error(`音频时长无效: ${duration}`))
+          return
        }

-        const blob = new Blob([bytes], { type: 'audio/mp3' })
-        const audio = new Audio()
-        const objectUrl = URL.createObjectURL(blob)
-
-        audio.addEventListener('loadedmetadata', () => {
-          URL.revokeObjectURL(objectUrl)
-          const durationMs = Math.round(audio.duration * 1000)
-          resolve(durationMs)
-        })
-
-        audio.addEventListener('error', (error) => {
-          URL.revokeObjectURL(objectUrl)
-          reject(error)
-        })
-
-        audio.src = objectUrl
-        audio.load()
-      } catch (error) {
-        reject(error)
+        // 减去安全边距(200ms)，避免因解析误差导致 sound_end_time 超过实际音频时长
+        const durationMs = Math.floor(duration * 1000) - 200
+        const rawDurationMs = Math.floor(duration * 1000)
+        console.log('[parseAudioDuration] 解析成功:', durationMs, 'ms (原始:', rawDurationMs, 'ms)')
+        cleanup()
+        resolve(durationMs)
      }
+
+      function onError() {
+        cleanup()
+        reject(new Error('音频解析失败，请检查音频格式'))
+      }
+
+      audio.addEventListener('loadedmetadata', onLoadedMetadata)
+      audio.addEventListener('error', onError)
+      audio.addEventListener('canplay', onLoadedMetadata, { once: true })
+      audio.src = objectUrl
+      audio.load()
    })
  }

-  /**
-   * 验证音频与人脸区间的重合时长（外部调用时传入校验参数）
-   */
-  const validateAudioDuration = (
-    faceStartTime: number = 0,
-    faceEndTime: number = 0,
-    minOverlapMs: number = 2000
-  ): boolean => {
-    if (faceStartTime <= 0 || faceEndTime <= 0) {
-      audioState.value.validationPassed = false
-      return false
-    }
-
-    const faceDurationMs = faceEndTime - faceStartTime
-    const audioDuration = audioState.value.durationMs
-
-    const overlapStart = faceStartTime
-    const overlapEnd = Math.min(faceEndTime, faceStartTime + audioDuration)
-    const overlapDuration = Math.max(0, overlapEnd - overlapStart)
-
-    const isValid = overlapDuration >= minOverlapMs
-
-    audioState.value.validationPassed = isValid
-
-    if (!isValid) {
-      const overlapSec = (overlapDuration / 1000).toFixed(1)
-      message.warning(
-        `音频时长（${(audioDuration/1000).toFixed(1)}秒）与人脸区间（${(faceDurationMs/1000).toFixed(1)}秒）不匹配，重合部分仅${overlapSec}秒，至少需要${(minOverlapMs/1000)}秒`
-      )
-    } else {
-      message.success('时长校验通过！')
-    }
-
-    return isValid
-  }
-
-  /**
-   * 重置音频状态
-   */
-  const resetAudioState = (): void => {
+  function resetAudioState(): void {
    audioState.value.generated = null
    audioState.value.durationMs = 0
-    audioState.value.validationPassed = false
    audioState.value.generating = false
  }

  return {
-    // 响应式状态
    ttsText,
    speechRate,
    selectedVoiceMeta,
    audioState,
-
-    // 计算属性
    canGenerateAudio,
    suggestedMaxChars,
-
-    // 方法
    generateAudio,
-    parseAudioDuration,
-    validateAudioDuration,
    resetAudioState,
  }
 }

-/**
- * 从字符串中提取ID
- */
 function extractIdFromString(str: string): string {
-  // 尝试从各种格式中提取ID
  const match = str.match(/[\w-]+$/)
  return match ? match[0] : str
 }
--- a/frontend/app/web-gold/src/views/kling/types/identify-face.ts
+++ b/frontend/app/web-gold/src/views/kling/types/identify-face.ts
@@ -11,6 +11,7 @@ export interface VideoState {
  videoFile: File | null
  previewVideoUrl: string
  selectedVideo: Video | null
+  fileId: string | number | null
  videoSource: 'upload' | 'select' | null
  selectorVisible: boolean
 }
@@ -20,6 +21,7 @@ export interface VideoState {
 */
 export interface Video {
  id: string | number
+  fileId: string | number
  fileName: string
  fileUrl: string
  fileSize: number
@@ -42,13 +44,23 @@ export interface IdentifyState {
  videoFileId: string | number | null
 }

+/**
+ * 人脸识别结果接口（不包含状态标志）
+ */
+export interface IdentifyResult {
+  sessionId: string
+  faceId: string
+  faceStartTime: number
+  faceEndTime: number
+  videoFileId: string | number | null
+}
+
 /**
 * 音频状态接口
 */
 export interface AudioState {
  generated: AudioData | null
  durationMs: number
-  validationPassed: boolean
  generating: boolean
 }

@@ -61,16 +73,6 @@ export interface AudioData {
  format?: string
 }

-/**
- * 素材校验接口
- */
-export interface MaterialValidation {
-  videoDuration: number
-  audioDuration: number
-  isValid: boolean
-  showDetails: boolean
-}
-
 /**
 * 音色元数据接口
 */
@@ -97,8 +99,6 @@ export interface UseVoiceGeneration {

  // 方法
  generateAudio: () => Promise<void>
-  parseAudioDuration: (base64Data: string) => Promise<number>
-  validateAudioDuration: () => boolean
  resetAudioState: () => void
 }

@@ -115,7 +115,7 @@ export interface UseDigitalHumanGeneration {

  // 方法
  handleFileUpload: (file: File) => Promise<void>
-  handleVideoSelect: (video: Video) => void
+  handleVideoSelect: (video: Video) => Promise<void>
  performFaceRecognition: () => Promise<void>
  resetVideoState: () => void
  resetIdentifyState: () => void
@@ -140,29 +140,25 @@ export interface UseIdentifyFaceController {
  // ==================== 数字人生成相关 ====================
  videoState: import('vue').Ref<VideoState>
  identifyState: import('vue').Ref<IdentifyState>
-  materialValidation: import('vue').Ref<MaterialValidation>
  faceDuration: import('vue').ComputedRef<number>
  performFaceRecognition: () => Promise<void>
  handleFileUpload: (file: File) => Promise<void>
  getVideoPreviewUrl: (video: Video) => string
  resetVideoState: () => void
  resetIdentifyState: () => void
-  resetMaterialValidation: () => void
-  validateMaterialDuration: (videoDurationMs: number, audioDurationMs: number) => boolean

  // ==================== 业务流程方法 ====================
  generateDigitalHuman: () => Promise<void>
  replaceVideo: () => void

  // ==================== 事件处理方法 ====================
-  handleVoiceSelect: (voice: VoiceMeta) => void
+  handleVoiceSelect: (voiceMeta: VoiceMeta) => void
  handleFileSelect: (event: Event) => void
  handleDrop: (event: DragEvent) => void
  triggerFileSelect: () => void
  handleSelectUpload: () => void
  handleSelectFromLibrary: () => void
-  handleVideoSelect: (video: Video) => void
-  handleSimplifyScript: () => void
+  handleVideoSelect: (video: Video) => Promise<void>
  handleVideoLoaded: (videoUrl: string) => void

  // ==================== 计算属性 ====================
@@ -171,6 +167,32 @@ export interface UseIdentifyFaceController {
  textareaPlaceholder: import('vue').ComputedRef<string>
  speechRateMarks: Record<number, string>
  speechRateDisplay: import('vue').ComputedRef<string>
+  faceDurationSec: import('vue').ComputedRef<string>
+  audioDurationSec: import('vue').ComputedRef<string>
+  showGenerateHint: import('vue').ComputedRef<boolean>
+  audioUrl: import('vue').ComputedRef<string>
+  validationPassed: import('vue').ComputedRef<boolean>
+
+  // ==================== 流程状态 ====================
+  pipelineState: import('vue').Ref<string>
+  pipelineStateLabel: import('vue').ComputedRef<string>
+  pipelineStateDescription: import('vue').ComputedRef<string>
+  isPipelineBusy: import('vue').ComputedRef<boolean>
+  isPipelineReady: import('vue').ComputedRef<boolean>
+  isPipelineFailed: import('vue').ComputedRef<boolean>
+  isPipelineCompleted: import('vue').ComputedRef<boolean>
+  pipelineProgress: import('vue').ComputedRef<number>
+  pipelineError: import('vue').Ref<string | null>
+  runPipeline: (params: {
+    videoFile: File | null
+    selectedVideo: any
+    text: string
+    voice: VoiceMeta
+    speechRate: number
+  }) => Promise<void>
+  createPipelineTask: () => Promise<void>
+  retryPipeline: () => void
+  resetPipeline: () => void

  // ==================== UI 辅助方法 ====================
  formatDuration: (seconds: number) => string
--- a/frontend/app/web-gold/src/views/system/task-management/digital-human-task/index.vue
+++ b/frontend/app/web-gold/src/views/system/task-management/digital-human-task/index.vue
@@ -59,7 +59,7 @@
        >
          <template #action>
            <a-space>
-            
+
              <a-popconfirm
                title="确定要删除选中的任务吗？删除后无法恢复。"
                @confirm="handleBatchDelete"
@@ -124,19 +124,7 @@
            <!-- 操作列 -->
            <template v-else-if="column.key === 'actions'">
              <a-space>
-                <!-- 预览按钮 -->
-                <a-button
-                  v-if="isStatus(record.status, 'success')"
-                  type="link"
-                  size="small"
-                  @click="handlePreview(record)"
-                  class="action-btn-preview"
-                >
-                  <template #icon>
-                    <PlayCircleOutlined />
-                  </template>
-                  预览
-                </a-button>
+     
                <!-- 下载按钮 -->
                <a-button
                  v-if="isStatus(record.status, 'success')"
@@ -171,23 +159,6 @@
        </a-table>
      </a-spin>
    </div>
-
-    <!-- 预览对话框 -->
-    <a-modal
-      v-model:open="previewVisible"
-      title="视频预览"
-      :footer="null"
-      width="800px"
-    >
-      <video
-        v-if="previewUrl"
-        :src="previewUrl"
-        controls
-        style="width: 100%; max-height: 600px"
-      >
-        您的浏览器不支持视频播放
-      </video>
-    </a-modal>
  </div>
 </template>

@@ -242,10 +213,6 @@ useTaskPolling(getDigitalHumanTaskPage, {
  }
 })

-// 预览相关
-const previewVisible = ref(false)
-const previewUrl = ref('')
-
 // 表格选择相关
 const selectedRowKeys = ref([])

@@ -333,78 +300,17 @@ const isStatus = (status, targetStatus) => {
  return status === targetStatus || status === targetStatus.toUpperCase()
 }

-// 预览视频
-const handlePreview = (record) => {
-  if (!record.resultVideoUrl) {
-    message.warning('该任务暂无视频结果，请稍后再试')
-    return
-  }

-  previewUrl.value = record.resultVideoUrl
-  previewVisible.value = true
-}

-// 下载视频
+// 下载视频 - 新窗口打开（浏览器自动处理下载）
 const handleDownload = (record) => {
+  console.log(record)
  if (!record.resultVideoUrl) {
    message.warning('该任务暂无视频结果，请稍后再试')
    return
  }

-  const link = document.createElement('a')
-  link.href = record.resultVideoUrl
-  link.download = `数字人视频_${record.id}_${Date.now()}.mp4`
-  link.target = '_blank'
-  document.body.appendChild(link)
-  link.click()
-  document.body.removeChild(link)
-}
-
-// 批量下载视频
-const handleBatchDownload = () => {
-  // 获取选中的已完成任务
-  const selectedTasks = list.value.filter(task =>
-    selectedRowKeys.value.includes(task.id) &&
-    isStatus(task.status, 'success')
-  )
-
-  if (selectedTasks.length === 0) {
-    message.warning('请选择已完成的任务')
-    return
-  }
-
-  let successCount = 0
-  let failCount = 0
-
-  // 逐个直接下载
-  for (const task of selectedTasks) {
-    if (!task.resultVideoUrl) {
-      failCount++
-      continue
-    }
-
-    try {
-      const link = document.createElement('a')
-      link.href = task.resultVideoUrl
-      link.download = `数字人视频_${task.id}_${Date.now()}.mp4`
-      link.target = '_blank'
-      document.body.appendChild(link)
-      link.click()
-      document.body.removeChild(link)
-      successCount++
-    } catch (error) {
-      console.error(`下载任务 ${task.id} 失败:`, error)
-      failCount++
-    }
-  }
-
-  if (failCount === 0) {
-    message.success(`已触发 ${successCount} 个文件的下载`)
-  } else if (successCount === 0) {
-    message.error('所有文件下载失败，请重试')
-  } else {
-    message.warning(`成功下载 ${successCount} 个文件，${failCount} 个文件下载失败`)
-  }
+  window.open(record.resultVideoUrl, '_blank')
 }

 // 批量删除任务
--- a/frontend/package.json
+++ b/frontend/package.json
@@ -12,6 +12,7 @@
  "description": "",
  "dependencies": {
    "@types/node": "^25.0.6",
+    "aplayer": "^1.10.1",
    "axios": "^1.12.2",
    "github-markdown-css": "^5.8.1",
    "localforage": "^1.10.0",
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/file/service/TikUserFileServiceImpl.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/file/service/TikUserFileServiceImpl.java
@@ -445,12 +445,12 @@ public class TikUserFileServiceImpl implements TikUserFileService {
    }

    @Override
-    public String getVideoPlayUrl(Long infraFileId) {
+    public String getVideoPlayUrl(Long fileId) {
        Long userId = SecurityFrameworkUtils.getLoginUserId();

-        // 查询文件（根据 infraFileId 字段查询）
+        // 查询文件（根据 fileId 字段查询）
        TikUserFileDO file = userFileMapper.selectOne(new LambdaQueryWrapperX<TikUserFileDO>()
-                .eq(TikUserFileDO::getFileId, infraFileId)
+                .eq(TikUserFileDO::getFileId, fileId)
                .eq(TikUserFileDO::getUserId, userId));

        if (file == null) {
@@ -473,12 +473,12 @@ public class TikUserFileServiceImpl implements TikUserFileService {
    }

    @Override
-    public String getAudioPlayUrl(Long infraFileId) {
+    public String getAudioPlayUrl(Long fileId) {
        Long userId = SecurityFrameworkUtils.getLoginUserId();

-        // 查询文件（根据 infraFileId 字段查询）
+        // 查询文件（根据 fileId 字段查询）
        TikUserFileDO file = userFileMapper.selectOne(new LambdaQueryWrapperX<TikUserFileDO>()
-                .eq(TikUserFileDO::getFileId, infraFileId)
+                .eq(TikUserFileDO::getFileId, fileId)
                .eq(TikUserFileDO::getUserId, userId));

        if (file == null) {
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/file/vo/app/AppTikUserFileRespVO.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/file/vo/app/AppTikUserFileRespVO.java
@@ -14,9 +14,12 @@ import java.time.LocalDateTime;
@Data
 public class AppTikUserFileRespVO {

-    @Schema(description = "文件编号", requiredMode = Schema.RequiredMode.REQUIRED, example = "1")
+    @Schema(description = "文件编号（主键）", requiredMode = Schema.RequiredMode.REQUIRED, example = "1")
    private Long id;

+    @Schema(description = "素材文件编号（关联 infra_file.id，用于获取播放URL）", example = "100")
+    private Long fileId;
+
    @Schema(description = "文件名称", requiredMode = Schema.RequiredMode.REQUIRED, example = "test.mp4")
    private String fileName;

--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/kling/client/KlingClient.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/kling/client/KlingClient.java
@@ -52,34 +52,12 @@ public class KlingClient {
        validateRequest(request);

        Map<String, Object> payload = buildPayload(request);
-        try {
-            String body = objectMapper.writeValueAsString(payload);
-            String url = properties.getBaseUrl() + "/klingai/v1/videos/identify-face";
+        String url = properties.getBaseUrl() + "/klingai/v1/videos/identify-face";
+        Request httpRequest = buildPostRequest(url, payload);

-            Request httpRequest = new Request.Builder()
-                    .url(url)
-                    .addHeader("Authorization", "Bearer " + properties.getApiKey())
-                    .addHeader("Content-Type", "application/json")
-                    .post(RequestBody.create(body.getBytes(StandardCharsets.UTF_8), JSON))
-                    .build();
-
-            try {
-                KlingIdentifyFaceResponse response = executeRequest(httpRequest, "identify-face", KlingIdentifyFaceResponse.class);
-                // 验证sessionId
-                if (StrUtil.isBlank(response.getData() == null ? null : response.getData().getSessionId())) {
-                    throw exception0(LATENTSYNC_SUBMIT_FAILED.getCode(), "可灵返回 sessionId 为空");
-                }
-                return response;
-            } catch (ServiceException ex) {
-                throw ex;
-            } catch (Exception ex) {
-                log.error("[Kling][identify-face exception]", ex);
-                throw exception(LATENTSYNC_SUBMIT_FAILED);
-            }
-        } catch (Exception ex) {
-            log.error("[Kling][build request exception]", ex);
-            throw exception(LATENTSYNC_SUBMIT_FAILED);
-        }
+        KlingIdentifyFaceResponse response = executeRequest(httpRequest, "identify-face", KlingIdentifyFaceResponse.class);
+        validateSessionId(response.getData() != null ? response.getData().getSessionId() : null, "sessionId");
+        return response;
    }

    /**
@@ -89,35 +67,13 @@ public class KlingClient {
        validateEnabled();
        validateLipSyncRequest(request);

-        try {
-            String body = objectMapper.writeValueAsString(request);
-            log.info("[Kling][create-lip-sync请求体] {}", body);
-            String url = properties.getBaseUrl() + "/klingai/v1/videos/advanced-lip-sync";
+        String url = properties.getBaseUrl() + "/klingai/v1/videos/advanced-lip-sync";
+        Request httpRequest = buildPostRequest(url, request);
+        log.info("[Kling][create-lip-sync请求体] {}", request);

-            Request httpRequest = new Request.Builder()
-                    .url(url)
-                    .addHeader("Authorization", "Bearer " + properties.getApiKey())
-                    .addHeader("Content-Type", "application/json")
-                    .post(RequestBody.create(body.getBytes(StandardCharsets.UTF_8), JSON))
-                    .build();
-
-            try {
-                KlingLipSyncCreateResponse response = executeRequest(httpRequest, "create-lip-sync", KlingLipSyncCreateResponse.class);
-                // 验证taskId
-                if (StrUtil.isBlank(response.getData() == null ? null : response.getData().getTaskId())) {
-                    throw exception0(LATENTSYNC_SUBMIT_FAILED.getCode(), "可灵返回 taskId 为空");
-                }
-                return response;
-            } catch (ServiceException ex) {
-                throw ex;
-            } catch (Exception ex) {
-                log.error("[Kling][create-lip-sync exception]", ex);
-                throw exception(LATENTSYNC_SUBMIT_FAILED);
-            }
-        } catch (Exception ex) {
-            log.error("[Kling][build request exception]", ex);
-            throw exception(LATENTSYNC_SUBMIT_FAILED);
-        }
+        KlingLipSyncCreateResponse response = executeRequest(httpRequest, "create-lip-sync", KlingLipSyncCreateResponse.class);
+        validateSessionId(response.getData() != null ? response.getData().getTaskId() : null, "taskId");
+        return response;
    }

    /**
@@ -129,29 +85,10 @@ public class KlingClient {
            throw exception0(LATENTSYNC_SUBMIT_FAILED.getCode(), "任务ID不能为空");
        }

-        try {
-            String url = properties.getBaseUrl() + "/klingai/v1/videos/advanced-lip-sync/" + taskId;
+        String url = properties.getBaseUrl() + "/klingai/v1/videos/advanced-lip-sync/" + taskId;
+        Request httpRequest = buildGetRequest(url);

-            Request httpRequest = new Request.Builder()
-                    .url(url)
-                    .addHeader("Authorization", "Bearer " + properties.getApiKey())
-                    .addHeader("Content-Type", "application/json")
-                    .get()
-                    .build();
-
-            try {
-                KlingLipSyncQueryResponse response = executeRequest(httpRequest, "get-lip-sync", KlingLipSyncQueryResponse.class);
-                return response;
-            } catch (ServiceException ex) {
-                throw ex;
-            } catch (Exception ex) {
-                log.error("[Kling][get-lip-sync exception]", ex);
-                throw exception(LATENTSYNC_SUBMIT_FAILED);
-            }
-        } catch (Exception ex) {
-            log.error("[Kling][build request exception]", ex);
-            throw exception(LATENTSYNC_SUBMIT_FAILED);
-        }
+        return executeRequest(httpRequest, "get-lip-sync", KlingLipSyncQueryResponse.class);
    }

    private void validateEnabled() {
@@ -272,10 +209,39 @@ public class KlingClient {
                    throw buildException(responseBody);
                }
            }
+
+            // 解析响应
+            T result = objectMapper.readValue(responseBody, responseClass);
+
+            // ✅ 检查业务错误码（可灵 API 可能返回 HTTP 200 但 code !== 0）
+            try {
+                JsonNode root = objectMapper.readTree(responseBody);
+                if (root.has("code")) {
+                    int code = root.get("code").asInt();
+                    if (code != 0) {
+                        String message = root.has("message") ? root.get("message").asText() :
+                                       root.has("detail") ? root.get("detail").asText() : "未知错误";
+                        String requestId = root.has("request_id") ? root.get("request_id").asText() : "unknown";
+
+                        log.error("[Kling][{} business error] code={}, message={}, request_id={}", operation, code, message, requestId);
+
+                        throw exception0(LATENTSYNC_SUBMIT_FAILED.getCode(),
+                            String.format("[%s] %s (code: %d, request_id: %s)", operation, message, code, requestId));
+                    }
+                }
+            } catch (ServiceException ex) {
+                throw ex;
+            } catch (Exception ex) {
+                log.warn("[Kling][{} check business code failed, continuing]", operation, ex);
+            }
+
            log.info("[Kling][{} success][responseBody={}]", operation, responseBody);

-            return objectMapper.readValue(responseBody, responseClass);
+            return result;
        } catch (Exception ex) {
+            if (ex instanceof ServiceException) {
+                throw (ServiceException) ex;
+            }
            log.error("[Kling][{} exception]", operation, ex);
            throw exception(LATENTSYNC_SUBMIT_FAILED);
        }
@@ -319,4 +285,43 @@ public class KlingClient {
            return exception0(LATENTSYNC_SUBMIT_FAILED.getCode(), body);
        }
    }
+
+    /**
+     * 构建 POST 请求
+     */
+    private Request buildPostRequest(String url, Object payload) {
+        try {
+            String body = objectMapper.writeValueAsString(payload);
+            return new Request.Builder()
+                    .url(url)
+                    .addHeader("Authorization", "Bearer " + properties.getApiKey())
+                    .addHeader("Content-Type", "application/json")
+                    .post(RequestBody.create(body.getBytes(StandardCharsets.UTF_8), JSON))
+                    .build();
+        } catch (Exception ex) {
+            log.error("[Kling][build POST request exception]", ex);
+            throw exception(LATENTSYNC_SUBMIT_FAILED);
+        }
+    }
+
+    /**
+     * 构建 GET 请求
+     */
+    private Request buildGetRequest(String url) {
+        return new Request.Builder()
+                .url(url)
+                .addHeader("Authorization", "Bearer " + properties.getApiKey())
+                .addHeader("Content-Type", "application/json")
+                .get()
+                .build();
+    }
+
+    /**
+     * 验证响应中的 sessionId 不为空
+     */
+    private void validateSessionId(String sessionId, String fieldName) {
+        if (StrUtil.isBlank(sessionId)) {
+            throw exception0(LATENTSYNC_SUBMIT_FAILED.getCode(), "可灵返回 " + fieldName + " 为空");
+        }
+    }
 }
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/client/SiliconFlowProvider.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/client/SiliconFlowProvider.java
@@ -58,7 +58,6 @@ public class SiliconFlowProvider implements VoiceCloneProvider {
            sfRequest.setText(getOrDefault(request.getTranscriptionText(), config.getPreviewText()));
            sfRequest.setAudio(AUDIO_MIME_TYPE + base64Audio);

-            // 调用上传参考音频 API
            String url = config.getBaseUrl() + config.getVoiceUploadUrl();
            String requestBody = JSONUtil.toJsonStr(sfRequest);
            log.debug("[SiliconFlowProvider][请求体]{}", requestBody);
@@ -67,7 +66,7 @@ public class SiliconFlowProvider implements VoiceCloneProvider {
                    .header("Authorization", "Bearer " + config.getApiKey())
                    .header("Content-Type", MediaType.APPLICATION_JSON_VALUE)
                    .body(requestBody)
-                    .timeout((int) config.getConnectTimeout().toMillis())
+                    .timeout((int) config.getReadTimeout().toMillis())
                    .execute();

            String responseBody = response.body();
@@ -116,14 +115,13 @@ public class SiliconFlowProvider implements VoiceCloneProvider {

        try {
            SiliconFlowTtsRequest sfRequest = SiliconFlowTtsRequest.builder()
-                    .model(getOrDefault(request.getModel(), config.getDefaultModel()))
+                    .model(getOrDefault(request.getModel(), getOrDefault(config.getDefaultModel(), "IndexTeam/IndexTTS-2")))
                    .input(request.getText())
                    .voice(request.getVoiceId())
                    .speed(request.getSpeechRate() != null ? request.getSpeechRate() : 1.0f)
                    .responseFormat(getOrDefault(request.getAudioFormat(), config.getAudioFormat()))
                    .build();

-            // 调用文本转语音 API
            String url = config.getBaseUrl() + config.getTtsUrl();
            String requestBody = JSONUtil.toJsonStr(sfRequest);
            log.debug("[SiliconFlowProvider][请求体]{}", requestBody);
@@ -142,7 +140,6 @@ public class SiliconFlowProvider implements VoiceCloneProvider {
                throw new RuntimeException("硅基流动文本转语音失败: " + errorBody);
            }

-            // 硅基流动直接返回二进制音频数据
            byte[] audioBytes = response.bodyBytes();
            String base64Audio = Base64.getEncoder().encodeToString(audioBytes);

--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/client/dto/SiliconFlowTtsRequest.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/client/dto/SiliconFlowTtsRequest.java
@@ -36,8 +36,6 @@ public class SiliconFlowTtsRequest {
     */
    private Float speed;

-
-
    /**
     * 响应格式（mp3, opus, wav, pcm）（API 参数名：response_format）
     */
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/config/SiliconFlowProviderConfig.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/config/SiliconFlowProviderConfig.java
@@ -32,8 +32,11 @@ public class SiliconFlowProviderConfig extends VoiceProviderProperties.ProviderC

    /**
     * 默认采样率
+     * <p>mp3: 32000, 44100 (默认 44100)</p>
+     * <p>opus: 48000</p>
+     * <p>wav/pcm: 8000, 16000, 24000, 32000, 44100 (默认 44100)</p>
     */
-    private Integer sampleRate = 24000;
+    private Integer sampleRate = 44100;

    /**
     * 默认音频格式
@@ -61,9 +64,9 @@ public class SiliconFlowProviderConfig extends VoiceProviderProperties.ProviderC
    private Duration connectTimeout = Duration.ofSeconds(10);

    /**
-     * 读取超时时间（3分钟，提升语音合成成功率）
+     * 读取超时时间（5分钟，提升语音合成成功率）
     */
-    private Duration readTimeout = Duration.ofSeconds(180);
+    private Duration readTimeout = Duration.ofSeconds(300);

    /**
     * 检查是否可用（有 API Key 即可用）
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/dal/dataobject/TikDigitalHumanTaskDO.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/dal/dataobject/TikDigitalHumanTaskDO.java
@@ -3,6 +3,7 @@ package cn.iocoder.yudao.module.tik.voice.dal.dataobject;
 import cn.iocoder.yudao.framework.common.enums.CommonStatusEnum;
 import cn.iocoder.yudao.framework.tenant.core.db.TenantBaseDO;
 import com.baomidou.mybatisplus.annotation.KeySequence;
+import com.baomidou.mybatisplus.annotation.TableField;
 import com.baomidou.mybatisplus.annotation.TableId;
 import com.baomidou.mybatisplus.annotation.TableName;
 import lombok.*;
@@ -163,6 +164,7 @@ public class TikDigitalHumanTaskDO extends TenantBaseDO {
    /**
     * 可灵口型同步任务ID（从advanced-lip-sync接口获取）
     */
+    @TableField("kling_task_id")
    private String klingTaskId;

 }
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/dal/mysql/TikDigitalHumanTaskMapper.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/dal/mysql/TikDigitalHumanTaskMapper.java
@@ -4,9 +4,11 @@ import cn.iocoder.yudao.framework.common.pojo.PageParam;
 import cn.iocoder.yudao.framework.common.pojo.PageResult;
 import cn.iocoder.yudao.framework.mybatis.core.mapper.BaseMapperX;
 import cn.iocoder.yudao.framework.mybatis.core.query.LambdaQueryWrapperX;
+import cn.iocoder.yudao.framework.tenant.core.aop.TenantIgnore;
 import cn.iocoder.yudao.module.tik.voice.dal.dataobject.TikDigitalHumanTaskDO;
 import cn.iocoder.yudao.module.tik.voice.vo.AppTikDigitalHumanPageReqVO;
 import org.apache.ibatis.annotations.Mapper;
+import org.apache.ibatis.annotations.Select;

 import java.util.List;

@@ -62,4 +64,20 @@ public interface TikDigitalHumanTaskMapper extends BaseMapperX<TikDigitalHumanTa
                .last("LIMIT " + limit));
    }

+    /**
+     * 查询待轮询的可灵任务（状态为PROCESSING且有klingTaskId，最近6小时内）
+     * 使用条件构造器，配合服务类 @TenantIgnore 注解忽略租户限制
+     */
+    @TenantIgnore
+    default List<TikDigitalHumanTaskDO> selectPendingKlingTasks() {
+        return selectList(new LambdaQueryWrapperX<TikDigitalHumanTaskDO>()
+                .eq(TikDigitalHumanTaskDO::getStatus, "PROCESSING")
+                .eq(TikDigitalHumanTaskDO::getAiProvider, "kling")
+                .isNotNull(TikDigitalHumanTaskDO::getKlingTaskId)
+                .ne(TikDigitalHumanTaskDO::getKlingTaskId, "")
+                .apply("create_time >= DATE_SUB(NOW(), INTERVAL 6 HOUR)")
+                .orderByDesc(TikDigitalHumanTaskDO::getCreateTime)
+                .last("LIMIT 50"));
+    }
+
 }
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/job/DigitalHumanTaskStatusSyncJob.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/job/DigitalHumanTaskStatusSyncJob.java
@@ -23,11 +23,11 @@ public class DigitalHumanTaskStatusSyncJob {
     */
    @Scheduled(fixedDelay = 10000)
    public void syncTaskStatus() {
-        log.debug("开始同步数字人任务状态");
+        log.info("[DigitalHumanTaskStatusSyncJob][开始同步数字人任务状态]");
        try {
            latentsyncPollingService.pollLatentsyncTasks();
        } catch (Exception e) {
-            log.error("同步数字人任务状态失败", e);
+            log.error("[DigitalHumanTaskStatusSyncJob][同步数字人任务状态失败]", e);
        }
    }

--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/service/DigitalHumanTaskServiceImpl.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/service/DigitalHumanTaskServiceImpl.java
@@ -425,12 +425,6 @@ public class DigitalHumanTaskServiceImpl implements DigitalHumanTaskService {
        // 设置当前步骤描述
        respVO.setCurrentStepDesc(DigitalHumanTaskStepEnum.getDesc(task.getCurrentStep()));

-        // 对 resultVideoUrl 进行预签名处理
-        if (StrUtil.isNotBlank(task.getResultVideoUrl())) {
-            String presignedUrl = fileApi.presignGetUrl(task.getResultVideoUrl(), PRESIGN_URL_EXPIRATION_SECONDS);
-            respVO.setResultVideoUrl(presignedUrl);
-        }
-
        return respVO;
    }

--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/service/LatentsyncPollingService.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/service/LatentsyncPollingService.java
@@ -1,18 +1,8 @@
 package cn.iocoder.yudao.module.tik.voice.service;

 import cn.hutool.core.util.StrUtil;
-import cn.hutool.http.HttpRequest;
-import cn.hutool.http.HttpResponse;
-import cn.iocoder.yudao.framework.common.util.http.HttpUtils;
-import cn.iocoder.yudao.framework.common.util.number.NumberUtils;
 import cn.iocoder.yudao.framework.tenant.core.aop.TenantIgnore;
-import cn.iocoder.yudao.module.infra.dal.dataobject.file.FileDO;
-import cn.iocoder.yudao.module.infra.dal.mysql.file.FileMapper;
-import cn.iocoder.yudao.module.infra.framework.file.core.client.FileClient;
-import cn.iocoder.yudao.module.infra.service.file.FileConfigService;
-import cn.iocoder.yudao.module.tik.file.dal.dataobject.TikUserFileDO;
-import cn.iocoder.yudao.module.tik.file.dal.mysql.TikUserFileMapper;
-import cn.iocoder.yudao.module.tik.file.service.TikOssInitService;
+import cn.iocoder.yudao.framework.tenant.core.util.TenantUtils;
 import cn.iocoder.yudao.module.tik.voice.dal.dataobject.TikDigitalHumanTaskDO;
 import cn.iocoder.yudao.module.tik.voice.dal.mysql.TikDigitalHumanTaskMapper;
 import cn.iocoder.yudao.module.tik.voice.vo.AppTikLatentsyncResultRespVO;
@@ -30,7 +20,7 @@ import java.time.LocalDateTime;
 import java.util.List;

 /**
- * Latentsync任务轮询服务 - 轻量化异步处理
+ * 数字人任务轮询服务
 * 使用@TenantIgnore忽略租户检查，因为轮询服务没有用户上下文
 *
 * @author 芋道源码
@@ -44,31 +34,22 @@ public class LatentsyncPollingService {
    private final TikDigitalHumanTaskMapper taskMapper;
    private final LatentsyncService latentsyncService;
    private final StringRedisTemplate stringRedisTemplate;
-    private final TikOssInitService ossInitService;
-    private final cn.iocoder.yudao.module.infra.api.file.FileApi fileApi;
-    private final TikUserFileMapper userFileMapper;
-    private final FileMapper fileMapper;
-    private final FileConfigService fileConfigService;
    private final KlingService klingService;

-    /**
-     * Redis键前缀
-     */
+    // ========== 常量 ==========
    private static final String REDIS_POLLING_PREFIX = "latentsync:polling:";
    private static final String REDIS_POLLING_TASKS_SET = "latentsync:polling:tasks";
    private static final String REDIS_POLLING_COUNT_PREFIX = "latentsync:polling:count:";
+    private static final String REDIS_RESULT_PREFIX = "digital_human:task:result:";

-    /**
-     * 轮询配置
-     */
    private static final int MAX_POLLING_COUNT = 90; // 最多轮询90次（15分钟）
-    private static final int POLLING_INTERVAL_SECONDS = 10; // 轮询间隔10秒
-    private static final Duration CACHE_EXPIRE_TIME = Duration.ofHours(1); // 缓存1小时
+    private static final Duration CACHE_EXPIRE_TIME = Duration.ofHours(1);
+    private static final Duration RESULT_CACHE_TIME = Duration.ofHours(24);
+
+    // ========== 公开方法 ==========

    /**
-     * 定时轮询Latentsync任务状态 - 每10秒执行一次
-     * 移除了分布式锁，通过查询条件和限制避免并发问题
-     * 注意：此方法现在由 DigitalHumanTaskStatusSyncJob 定时调用，不在服务内部使用 @Scheduled 注解
+     * 定时轮询任务状态入口
     */
    public void pollLatentsyncTasks() {
        try {
@@ -78,48 +59,15 @@ public class LatentsyncPollingService {
        }
    }

-    /**
-     * 执行轮询任务的具体逻辑
-     */
-    private void executePollingTasks() {
-        try {
-            // 轮询Latentsync任务
-            List<String> taskIds = getPendingPollingTasks();
-            if (!taskIds.isEmpty()) {
-                log.debug("[pollLatentsyncTasks][开始轮询Latentsync任务][任务数量={}]", taskIds.size());
-
-                // 逐个处理Latentsync任务
-                for (String taskIdStr : taskIds) {
-                    try {
-                        Long taskId = Long.parseLong(taskIdStr);
-                        pollSingleTask(taskId);
-                    } catch (Exception e) {
-                        log.error("[pollLatentsyncTasks][轮询Latentsync任务失败][taskId={}]", taskIdStr, e);
-                    }
-                }
-            }
-
-            // 轮询可灵任务
-            pollKlingTasks();
-
-        } catch (Exception e) {
-            log.error("[pollLatentsyncTasks][轮询任务异常]", e);
-        }
-    }
-
    /**
     * 添加任务到轮询队列
     */
    public void addTaskToPollingQueue(Long taskId, String requestId) {
        try {
-            // 存储任务信息
            String taskKey = REDIS_POLLING_PREFIX + requestId;
            stringRedisTemplate.opsForValue().set(taskKey, taskId.toString(), CACHE_EXPIRE_TIME);
-
-            // 添加到待轮询集合
            stringRedisTemplate.opsForZSet().add(REDIS_POLLING_TASKS_SET, requestId, System.currentTimeMillis());

-            // 初始化轮询次数
            String countKey = REDIS_POLLING_COUNT_PREFIX + requestId;
            stringRedisTemplate.opsForValue().set(countKey, "0", CACHE_EXPIRE_TIME);

@@ -130,184 +78,210 @@ public class LatentsyncPollingService {
    }

    /**
-     * 单个任务轮询
+     * 清理过期任务（由定时任务调用）
+     */
+    public void cleanupExpiredTasks() {
+        try {
+            stringRedisTemplate.delete(REDIS_POLLING_TASKS_SET);
+        } catch (Exception e) {
+            log.error("[cleanupExpiredTasks][清理过期任务异常]", e);
+        }
+    }
+
+    // ========== 私有方法 ==========
+
+    /**
+     * 执行轮询任务的具体逻辑
+     */
+    private void executePollingTasks() {
+        // 轮询 Latentsync 任务
+        List<String> taskIds = getPendingPollingTasks();
+        for (String taskIdStr : taskIds) {
+            try {
+                pollSingleTask(Long.parseLong(taskIdStr));
+            } catch (Exception e) {
+                log.error("[executePollingTasks][轮询任务失败][taskId={}]", taskIdStr, e);
+            }
+        }
+
+        // 轮询 Kling 任务
+        pollKlingTasks();
+    }
+
+    /**
+     * 轮询单个 Latentsync 任务
     */
    private void pollSingleTask(Long taskId) {
-        // 获取任务的requestId和轮询次数（在try块外声明，供catch块使用）
        String requestId = null;
-        String countKey = null;
        int currentCount = 0;

        try {
-            // 获取任务的requestId
            String taskKey = REDIS_POLLING_PREFIX + "task_" + taskId;
            requestId = stringRedisTemplate.opsForValue().get(taskKey);

            if (StrUtil.isBlank(requestId)) {
-                // 如果没有requestId，说明任务可能已完成或已取消，从轮询队列中移除
-                removeFromPollingQueue(taskId, null);
+                removeFromPollingQueue(requestId);
                return;
            }

-            // 检查轮询次数
-            countKey = REDIS_POLLING_COUNT_PREFIX + requestId;
+            String countKey = REDIS_POLLING_COUNT_PREFIX + requestId;
            String countStr = stringRedisTemplate.opsForValue().get(countKey);
            currentCount = countStr != null ? Integer.parseInt(countStr) : 0;

            if (currentCount >= MAX_POLLING_COUNT) {
-                // 超时，标记任务失败
-                log.warn("[pollSingleTask][任务轮询超时][taskId={}, requestId={}, count={}]", taskId, requestId, currentCount);
-                markTaskFailed(taskId, "Latentsync处理超时");
-                removeFromPollingQueue(taskId, requestId);
+                markTaskFailed(taskId, "处理超时");
+                removeFromPollingQueue(requestId);
                return;
            }

-            // 调用Latentsync API检查状态
+            // 查询任务状态
            AppTikLatentsyncResultRespVO result = latentsyncService.getTaskResult(requestId);
            String status = result.getStatus();

-            log.debug("[pollSingleTask][轮询任务状态][taskId={}, requestId={}, status={}]", taskId, requestId, status);
-
-            // 检查任务状态
            if ("COMPLETED".equals(status)) {
-                // 任务完成
-                handleTaskCompleted(taskId, result.getVideo().getUrl(), requestId);
+                completeTask(taskId, result.getVideo().getUrl(), requestId);
            } else if ("FAILED".equals(status) || "ERROR".equals(status)) {
-                // 任务失败
-                handleTaskFailed(taskId, "Latentsync处理失败: " + status, requestId);
+                markTaskFailed(taskId, "处理失败: " + status);
+                removeFromPollingQueue(requestId);
            } else {
-                // 继续轮询，更新轮询次数
                stringRedisTemplate.opsForValue().set(countKey, String.valueOf(currentCount + 1), CACHE_EXPIRE_TIME);
            }
+
        } catch (Exception e) {
            log.error("[pollSingleTask][轮询任务异常][taskId={}]", taskId, e);
-
-            // 轮询异常处理：增加轮询次数，避免无限重试
            int errorCount = currentCount + 1;
            if (errorCount >= MAX_POLLING_COUNT) {
-                // 达到最大次数，标记任务失败
-                log.warn("[pollSingleTask][任务轮询异常次数过多，标记失败][taskId={}, count={}]", taskId, errorCount);
-                if (requestId != null && StrUtil.isNotBlank(requestId)) {
-                    markTaskFailed(taskId, "Latentsync API调用异常：" + e.getMessage());
-                    removeFromPollingQueue(taskId, requestId);
-                }
+                markTaskFailed(taskId, "API调用异常");
+                removeFromPollingQueue(requestId);
            } else {
-                // 更新轮询次数，继续重试
-                if (countKey != null) {
-                    stringRedisTemplate.opsForValue().set(countKey, String.valueOf(errorCount), CACHE_EXPIRE_TIME);
-                    log.debug("[pollSingleTask][轮询异常，增加次数后继续重试][taskId={}, count={}]", taskId, errorCount);
-                }
+                String countKey = REDIS_POLLING_COUNT_PREFIX + requestId;
+                stringRedisTemplate.opsForValue().set(countKey, String.valueOf(errorCount), CACHE_EXPIRE_TIME);
            }
        }
    }

    /**
-     * 处理任务完成
+     * 轮询 Kling 任务
+     */
+    private void pollKlingTasks() {
+        try {
+            List<TikDigitalHumanTaskDO> klingTasks = TenantUtils.executeIgnore(
+                    () -> taskMapper.selectPendingKlingTasks()
+            );
+
+            for (TikDigitalHumanTaskDO task : klingTasks) {
+                try {
+                    pollKlingSingleTask(task);
+                } catch (Exception e) {
+                    log.error("[pollKlingTasks][轮询任务失败][taskId={}]", task.getId(), e);
+                }
+            }
+
+        } catch (Exception e) {
+            log.error("[pollKlingTasks][轮询Kling任务异常]", e);
+        }
+    }
+
+    /**
+     * 轮询单个 Kling 任务
+     */
+    private void pollKlingSingleTask(TikDigitalHumanTaskDO task) {
+        String klingTaskId = task.getKlingTaskId();
+        if (StrUtil.isBlank(klingTaskId)) {
+            return;
+        }
+
+        try {
+            KlingLipSyncQueryResponse response = klingService.getLipSyncTask(klingTaskId);
+            String taskStatus = response.getData().getTaskStatus();
+
+            if ("succeed".equalsIgnoreCase(taskStatus)) {
+                List<KlingLipSyncVideoVO> videos = response.getData().getTaskResult().getVideos();
+                if (videos != null && !videos.isEmpty()) {
+                    completeTask(task.getId(), videos.get(0).getUrl(), null);
+                }
+            } else if ("failed".equalsIgnoreCase(taskStatus)) {
+                String errorMsg = "可灵任务执行失败: " + response.getData().getTaskStatusMsg();
+                updateTaskStatus(task.getId(), "FAILED", task.getCurrentStep(), task.getProgress(), errorMsg);
+            } else if ("submitted".equalsIgnoreCase(taskStatus) || "processing".equalsIgnoreCase(taskStatus)) {
+                updateTaskStatus(task.getId(), "PROCESSING", "sync_lip", 70, null);
+            }
+
+        } catch (Exception e) {
+            log.error("[pollKlingSingleTask][任务({})查询失败]", task.getId(), e);
+        }
+    }
+
+    /**
+     * 完成任务（通用方法）
     */
    @Transactional(rollbackFor = Exception.class)
-    public void handleTaskCompleted(Long taskId, String videoUrl, String requestId) {
-        TikDigitalHumanTaskDO task = null;
+    private void completeTask(Long taskId, String videoUrl, String requestId) {
        try {
-            // 获取任务信息
-            task = taskMapper.selectById(taskId);
-            if (task == null) {
-                log.error("[handleTaskCompleted][任务不存在][taskId={}]", taskId);
-                return;
-            }
-
-            // 保存视频到OSS（异步处理，轻量化逻辑）
-            OssSaveResult saveResult = null;
-            try {
-                // 保存视频到OSS，避免临时URL过期
-                saveResult = saveVideoToOss(task, videoUrl);
-                log.info("[handleTaskCompleted][任务({})视频已保存到OSS][url={}]", taskId, saveResult.getUrl());
-            } catch (Exception e) {
-                log.warn("[handleTaskCompleted][任务({})保存视频失败，使用原URL][error={}]", taskId, e.getMessage());
-                saveResult = new OssSaveResult(videoUrl, 0, null, null); // 降级处理
-            }
-
-            // 更新任务状态为成功
            TikDigitalHumanTaskDO updateObj = new TikDigitalHumanTaskDO();
            updateObj.setId(taskId);
            updateObj.setStatus("SUCCESS");
            updateObj.setCurrentStep("finishing");
            updateObj.setProgress(100);
-            updateObj.setResultVideoUrl(saveResult.getUrl());
+            updateObj.setResultVideoUrl(videoUrl);
            updateObj.setFinishTime(LocalDateTime.now());
-            taskMapper.updateById(updateObj);

-            // 缓存结果到Redis（快速回显）
-            try {
-                String resultKey = "digital_human:task:result:" + taskId;
-                stringRedisTemplate.opsForValue().set(resultKey, saveResult.getUrl(), Duration.ofHours(24));
-            } catch (Exception e) {
-                log.warn("[handleTaskCompleted][任务({})缓存结果失败]", taskId, e);
+            TenantUtils.executeIgnore(() -> taskMapper.updateById(updateObj));
+
+            // 缓存结果
+            String resultKey = REDIS_RESULT_PREFIX + taskId;
+            stringRedisTemplate.opsForValue().set(resultKey, videoUrl, RESULT_CACHE_TIME);
+
+            if (requestId != null) {
+                removeFromPollingQueue(requestId);
            }

-            // 保存结果视频到用户文件表（这样用户可以在素材库中查看）
-            saveResultVideoToUserFiles(task, saveResult);
-
-            // 从轮询队列中移除
-            removeFromPollingQueue(taskId, requestId);
-
-            log.info("[handleTaskCompleted][任务完成][taskId={}, requestId={}]", taskId, requestId);
+            log.info("[completeTask][任务完成][taskId={}, videoUrl={}]", taskId, videoUrl);
        } catch (Exception e) {
-            log.error("[handleTaskCompleted][处理任务完成失败][taskId={}]", taskId, e);
+            log.error("[completeTask][处理任务完成失败][taskId={}]", taskId, e);
        }
    }

    /**
-     * 处理任务失败
+     * 标记任务失败
     */
    @Transactional(rollbackFor = Exception.class)
-    public void handleTaskFailed(Long taskId, String errorMessage, String requestId) {
+    private void markTaskFailed(Long taskId, String errorMessage) {
        try {
-            // 更新任务状态为失败
            TikDigitalHumanTaskDO updateObj = new TikDigitalHumanTaskDO();
            updateObj.setId(taskId);
            updateObj.setStatus("FAILED");
            updateObj.setErrorMessage(errorMessage);
            updateObj.setFinishTime(LocalDateTime.now());
-            taskMapper.updateById(updateObj);

-            // 从轮询队列中移除
-            removeFromPollingQueue(taskId, requestId);
+            TenantUtils.executeIgnore(() -> taskMapper.updateById(updateObj));

-            log.warn("[handleTaskFailed][任务失败][taskId={}, requestId={}, error={}]", taskId, requestId, errorMessage);
+            log.warn("[markTaskFailed][任务失败][taskId={}, error={}]", taskId, errorMessage);
        } catch (Exception e) {
-            log.error("[handleTaskFailed][处理任务失败失败][taskId={}]", taskId, e);
+            log.error("[markTaskFailed][标记任务失败失败][taskId={}]", taskId, e);
        }
    }

    /**
-     * 标记任务失败（内部使用）
+     * 更新任务状态
     */
-    private void markTaskFailed(Long taskId, String errorMessage) {
-        handleTaskFailed(taskId, errorMessage, null);
-    }
+    private void updateTaskStatus(Long taskId, String status, String currentStep, Integer progress, String errorMessage) {
+        TikDigitalHumanTaskDO updateObj = new TikDigitalHumanTaskDO();
+        updateObj.setId(taskId);
+        updateObj.setStatus(status);
+        updateObj.setCurrentStep(currentStep);
+        updateObj.setProgress(progress);

-    /**
-     * 从轮询队列中移除任务
-     */
-    private void removeFromPollingQueue(Long taskId, String requestId) {
-        try {
-            if (StrUtil.isNotBlank(requestId)) {
-                // 移除具体任务
-                String taskKey = REDIS_POLLING_PREFIX + requestId;
-                stringRedisTemplate.delete(taskKey);
-
-                String countKey = REDIS_POLLING_COUNT_PREFIX + requestId;
-                stringRedisTemplate.delete(countKey);
-
-                stringRedisTemplate.opsForZSet().remove(REDIS_POLLING_TASKS_SET, requestId);
-            } else {
-                // 尝试通过taskId找到requestId并移除
-                String pattern = REDIS_POLLING_PREFIX + "task_" + taskId;
-                // 这里可以优化为直接查询，但为了简单起见，先不实现
-            }
-        } catch (Exception e) {
-            log.error("[removeFromPollingQueue][移除任务失败][taskId={}, requestId={}]", taskId, requestId, e);
+        if ("SUCCESS".equals(status)) {
+            updateObj.setFinishTime(LocalDateTime.now());
+        } else if ("PROCESSING".equals(status)) {
+            updateObj.setStartTime(LocalDateTime.now());
+        } else if ("FAILED".equals(status)) {
+            updateObj.setErrorMessage(errorMessage);
+            updateObj.setFinishTime(LocalDateTime.now());
        }
+
+        TenantUtils.executeIgnore(() -> taskMapper.updateById(updateObj));
    }

    /**
@@ -315,8 +289,6 @@ public class LatentsyncPollingService {
     */
    private List<String> getPendingPollingTasks() {
        try {
-            // 获取所有待轮询的任务
-            // 注意：这里返回的是requestId，不是taskId
            List<String> requestIds = stringRedisTemplate.opsForZSet()
                    .range(REDIS_POLLING_TASKS_SET, 0, -1)
                    .stream()
@@ -326,12 +298,8 @@ public class LatentsyncPollingService {
                return List.of();
            }

-            // 将requestId转换为taskId
            return requestIds.stream()
-                    .map(requestId -> {
-                        String taskKey = REDIS_POLLING_PREFIX + requestId;
-                        return stringRedisTemplate.opsForValue().get(taskKey);
-                    })
+                    .map(requestId -> stringRedisTemplate.opsForValue().get(REDIS_POLLING_PREFIX + requestId))
                    .filter(StrUtil::isNotBlank)
                    .toList();
        } catch (Exception e) {
@@ -341,353 +309,18 @@ public class LatentsyncPollingService {
    }

    /**
-     * 清理过期任务（每天凌晨2点执行）
-     * 注意：此方法现在由外部调度器调用，不在服务内部使用 @Scheduled 注解
+     * 从轮询队列中移除任务
     */
-    public void cleanupExpiredTasks() {
+    private void removeFromPollingQueue(String requestId) {
        try {
-            log.info("[cleanupExpiredTasks][开始清理过期轮询任务]");
-
-            // 清理过期的轮询记录
-            stringRedisTemplate.delete(REDIS_POLLING_TASKS_SET);
-
-            // 可以添加更多清理逻辑
-
-            log.info("[cleanupExpiredTasks][清理完成]");
+            if (StrUtil.isNotBlank(requestId)) {
+                stringRedisTemplate.delete(REDIS_POLLING_PREFIX + requestId);
+                stringRedisTemplate.delete(REDIS_POLLING_COUNT_PREFIX + requestId);
+                stringRedisTemplate.opsForZSet().remove(REDIS_POLLING_TASKS_SET, requestId);
+            }
        } catch (Exception e) {
-            log.error("[cleanupExpiredTasks][清理过期任务异常]", e);
+            log.error("[removeFromPollingQueue][移除任务失败][requestId={}]", requestId, e);
        }
    }

-    /**
-     * 保存视频到OSS - 直接保存到 infra_file 避免重复
-     * 返回保存结果，包含URL、文件大小和文件ID
-     */
-    private OssSaveResult saveVideoToOss(TikDigitalHumanTaskDO task, String remoteVideoUrl) throws Exception {
-        log.info("[saveVideoToOss][任务({})开始下载并保存视频到OSS][remoteUrl={}]", task.getId(), remoteVideoUrl);
-
-        try {
-            // 1. 下载远程视频文件
-            byte[] videoBytes = downloadRemoteFile(remoteVideoUrl);
-
-            // 2. 内存检查：超过50MB记录警告
-            int sizeMB = videoBytes.length / 1024 / 1024;
-            if (sizeMB > 50) {
-                log.warn("[saveVideoToOss][任务({})视频文件较大][size={}MB]", task.getId(), sizeMB);
-            }
-
-            // 3. 获取OSS目录和文件名
-            Long userId = task.getUserId();
-            String baseDirectory = ossInitService.getOssDirectoryByCategory(userId, "generate");
-            String fileName = String.format("数字人视频_%d_%d.mp4", task.getId(), System.currentTimeMillis());
-
-            // 4. 获取FileClient并上传到OSS
-            FileClient client = fileConfigService.getMasterFileClient();
-            if (client == null) {
-                throw new Exception("获取FileClient失败");
-            }
-
-            // 5. 生成上传路径（包含日期前缀和时间戳后缀）
-            String filePath = generateUploadPath(fileName, baseDirectory);
-
-            // 6. 上传到OSS
-            String presignedUrl = client.upload(videoBytes, filePath, "video/mp4");
-
-            // 7. 移除预签名参数，获取基础URL
-            String cleanUrl = HttpUtils.removeUrlQuery(presignedUrl);
-
-            // 8. 保存到 infra_file 表
-            FileDO infraFile = new FileDO()
-                    .setConfigId(client.getId())
-                    .setName(fileName)
-                    .setPath(filePath)
-                    .setUrl(cleanUrl)
-                    .setType("video/mp4")
-                    .setSize(videoBytes.length);
-            fileMapper.insert(infraFile);
-            Long infraFileId = infraFile.getId();
-
-            log.info("[saveVideoToOss][任务({})视频保存完成][infraFileId={}, size={}MB]",
-                    task.getId(), infraFileId, sizeMB);
-            return new OssSaveResult(cleanUrl, videoBytes.length, filePath, infraFileId);
-
-        } catch (Exception e) {
-            log.error("[saveVideoToOss][任务({})保存视频失败][remoteUrl={}]", task.getId(), remoteVideoUrl, e);
-            return new OssSaveResult(remoteVideoUrl, 0, null, null);
-        }
-    }
-
-    /**
-     * 生成上传路径（与 FileService 保持一致）
-     */
-    private String generateUploadPath(String name, String directory) {
-        String prefix = cn.hutool.core.date.LocalDateTimeUtil.format(
-                cn.hutool.core.date.LocalDateTimeUtil.now(),
-                cn.hutool.core.date.DatePattern.PURE_DATE_PATTERN);
-        String suffix = String.valueOf(System.currentTimeMillis());
-
-        String ext = cn.hutool.core.io.FileUtil.extName(name);
-        if (StrUtil.isNotEmpty(ext)) {
-            name = cn.hutool.core.io.FileUtil.mainName(name) + "_" + suffix + "." + ext;
-        } else {
-            name = name + "_" + suffix;
-        }
-
-        if (StrUtil.isNotEmpty(prefix)) {
-            name = prefix + "/" + name;
-        }
-        if (StrUtil.isNotEmpty(directory)) {
-            name = directory + "/" + name;
-        }
-        return name;
-    }
-
-    /**
-     * OSS保存结果
-     */
-    private static class OssSaveResult {
-        private final String url;
-        private final int fileSize;
-        private final String filePath;
-        private final Long infraFileId;
-
-        public OssSaveResult(String url, int fileSize, String filePath, Long infraFileId) {
-            this.url = url;
-            this.fileSize = fileSize;
-            this.filePath = filePath;
-            this.infraFileId = infraFileId;
-        }
-
-        public String getUrl() {
-            return url;
-        }
-
-        public int getFileSize() {
-            return fileSize;
-        }
-
-        public String getFilePath() {
-            return filePath;
-        }
-
-        public Long getInfraFileId() {
-            return infraFileId;
-        }
-    }
-
-    /**
-     * 下载远程文件 - 内存优化
-     */
-    private byte[] downloadRemoteFile(String remoteUrl) throws Exception {
-        log.info("[downloadRemoteFile][下载文件][url={}]", remoteUrl);
-
-        try (HttpResponse response = HttpRequest.get(remoteUrl)
-                .execute()) {
-
-            if (!response.isOk()) {
-                throw new Exception("下载文件失败: HTTP " + response.getStatus());
-            }
-
-            // 流式读取：分块处理避免大文件OOM
-            byte[] bytes = response.bodyBytes();
-            int sizeMB = bytes.length / 1024 / 1024;
-            log.info("[downloadRemoteFile][文件下载完成][size={} bytes, {}MB]", bytes.length, sizeMB);
-            return bytes;
-        }
-    }
-
-    /**
-     * 保存结果视频到用户文件表
-     */
-    private void saveResultVideoToUserFiles(TikDigitalHumanTaskDO task, OssSaveResult saveResult) {
-        try {
-            Long userId = task.getUserId();
-            Long infraFileId = saveResult.getInfraFileId();
-
-            // 验证必要参数
-            if (userId == null || infraFileId == null) {
-                log.warn("[saveResultVideoToUserFiles][任务({})参数不完整，无法保存][userId={}, infraFileId={}]",
-                        task.getId(), userId, infraFileId);
-                return;
-            }
-
-            // 创建用户文件记录
-            TikUserFileDO userFile = new TikUserFileDO();
-            userFile.setUserId(userId);
-            userFile.setFileId(infraFileId);
-            userFile.setFileName(String.format("数字人视频_%d_%d.mp4", task.getId(), System.currentTimeMillis()));
-            userFile.setFileType("video/mp4");
-            userFile.setFileCategory("generate");
-            userFile.setFileUrl(saveResult.getUrl());
-            userFile.setFilePath(saveResult.getFilePath());
-            userFile.setFileSize((long) saveResult.getFileSize());
-
-            userFileMapper.insert(userFile);
-
-            log.info("[saveResultVideoToUserFiles][任务({})文件记录已保存][userFileId={}, infraFileId={}]",
-                    task.getId(), userFile.getId(), infraFileId);
-        } catch (Exception e) {
-            log.error("[saveResultVideoToUserFiles][任务({})保存失败]", task.getId(), e);
-        }
-    }
-
-    /**
-     * 轮询可灵任务状态
-     */
-    private void pollKlingTasks() {
-        try {
-            // 参考混剪任务实现：添加时间和数量限制，避免并发问题
-            // 1. 时间范围限制：只检查最近6小时内的任务（避免检查历史任务）
-            // 2. 数量限制：每次最多检查50个任务（避免单次查询过多）
-            LocalDateTime startTime = LocalDateTime.now().minusHours(6);
-
-            // 查询有待轮询的可灵任务（状态为PROCESSING且有klingTaskId，限制时间和数量）
-            List<TikDigitalHumanTaskDO> klingTasks = taskMapper.selectList(
-                    new cn.iocoder.yudao.framework.mybatis.core.query.LambdaQueryWrapperX<TikDigitalHumanTaskDO>()
-                            .eq(TikDigitalHumanTaskDO::getStatus, "PROCESSING")
-                            .eq(TikDigitalHumanTaskDO::getAiProvider, "kling")
-                            .isNotNull(TikDigitalHumanTaskDO::getKlingTaskId)
-                            .ne(TikDigitalHumanTaskDO::getKlingTaskId, "")
-                            .ge(TikDigitalHumanTaskDO::getCreateTime, startTime)  // 只检查最近6小时
-                            .orderByDesc(TikDigitalHumanTaskDO::getCreateTime)
-                            .last("LIMIT 50") // 限制数量，避免并发
-            );
-
-            if (klingTasks.isEmpty()) {
-                return;
-            }
-
-            log.debug("[pollKlingTasks][开始轮询可灵任务][任务数量={}]", klingTasks.size());
-
-            // 逐个处理可灵任务
-            for (TikDigitalHumanTaskDO task : klingTasks) {
-                try {
-                    pollKlingSingleTask(task);
-                } catch (Exception e) {
-                    log.error("[pollKlingTasks][轮询可灵任务失败][taskId={}]", task.getId(), e);
-                }
-            }
-
-        } catch (Exception e) {
-            log.error("[pollKlingTasks][轮询可灵任务异常]", e);
-        }
-    }
-
-    /**
-     * 轮询单个可灵任务
-     */
-    private void pollKlingSingleTask(TikDigitalHumanTaskDO task) {
-        String klingTaskId = task.getKlingTaskId();
-        if (StrUtil.isBlank(klingTaskId)) {
-            log.warn("[pollKlingSingleTask][任务({})缺少klingTaskId]", task.getId());
-            return;
-        }
-
-        try {
-            // 查询可灵任务状态
-            KlingLipSyncQueryResponse response = klingService.getLipSyncTask(klingTaskId);
-            String taskStatus = response.getData().getTaskStatus();
-            String taskStatusMsg = response.getData().getTaskStatusMsg();
-
-            log.debug("[pollKlingSingleTask][任务({})状态更新][klingTaskId={}, status={}]",
-                    task.getId(), klingTaskId, taskStatus);
-
-            // 根据状态更新任务
-            if ("succeed".equalsIgnoreCase(taskStatus)) {
-                // 任务成功完成
-                List<KlingLipSyncVideoVO> videos = response.getData().getTaskResult().getVideos();
-                if (videos != null && !videos.isEmpty()) {
-                    String videoUrl = videos.get(0).getUrl();
-
-                    // 保存视频到OSS（异步处理，轻量化逻辑）
-                    OssSaveResult saveResult = null;
-                    try {
-                        // 保存视频到OSS，避免临时URL过期
-                        saveResult = saveVideoToOss(task, videoUrl);
-                        log.info("[pollKlingSingleTask][任务({})视频已保存到OSS][url={}]", task.getId(), saveResult.getUrl());
-                    } catch (Exception e) {
-                        log.warn("[pollKlingSingleTask][任务({})保存视频失败，使用原URL][error={}]", task.getId(), e.getMessage());
-                        saveResult = new OssSaveResult(videoUrl, 0, null, null); // 降级处理
-                    }
-
-                    // 更新任务状态为成功
-                    TikDigitalHumanTaskDO updateObj = new TikDigitalHumanTaskDO();
-                    updateObj.setId(task.getId());
-                    updateObj.setStatus("SUCCESS");
-                    updateObj.setCurrentStep("finishing");
-                    updateObj.setProgress(100);
-                    updateObj.setResultVideoUrl(saveResult.getUrl());
-                    updateObj.setFinishTime(LocalDateTime.now());
-                    taskMapper.updateById(updateObj);
-
-                    // 缓存结果到Redis（快速回显）
-                    try {
-                        String resultKey = "digital_human:task:result:" + task.getId();
-                        stringRedisTemplate.opsForValue().set(resultKey, saveResult.getUrl(), Duration.ofHours(24));
-                    } catch (Exception e) {
-                        log.warn("[pollKlingSingleTask][任务({})缓存结果失败]", task.getId(), e);
-                    }
-
-                    // 保存结果视频到用户文件表
-                    saveResultVideoToUserFiles(task, saveResult);
-
-                    log.info("[pollKlingSingleTask][任务({})完成][videoUrl={}]", task.getId(), saveResult.getUrl());
-                } else {
-                    log.warn("[pollKlingSingleTask][任务({})成功但无视频结果]", task.getId());
-                }
-
-            } else if ("failed".equalsIgnoreCase(taskStatus)) {
-                // 任务失败
-                String errorMsg = "可灵任务执行失败: " + (StrUtil.isNotBlank(taskStatusMsg) ? taskStatusMsg : "未知错误");
-                updateTaskStatus(task.getId(), "FAILED", task.getCurrentStep(), task.getProgress(), errorMsg, null, errorMsg);
-                log.error("[pollKlingSingleTask][任务({})失败][error={}]", task.getId(), errorMsg);
-
-            } else if ("submitted".equalsIgnoreCase(taskStatus) || "processing".equalsIgnoreCase(taskStatus)) {
-                // 任务还在处理中，更新进度
-                updateTaskStatus(task.getId(), "PROCESSING", "sync_lip", 70, "口型同步处理中", null);
-                log.debug("[pollKlingSingleTask][任务({})处理中]", task.getId());
-
-            } else {
-                log.warn("[pollKlingSingleTask][任务({})未知状态][status={}]", task.getId(), taskStatus);
-            }
-
-        } catch (Exception e) {
-            log.error("[pollKlingSingleTask][任务({})查询失败]", task.getId(), e);
-            // 不更新任务状态，避免误判
-        }
-    }
-
-    /**
-     * 更新任务状态
-     */
-    private void updateTaskStatus(Long taskId, String status, String currentStep, Integer progress,
-                                  String message, String resultVideoUrl) {
-        updateTaskStatus(taskId, status, currentStep, progress, message, resultVideoUrl, null);
-    }
-
-    /**
-     * 更新任务状态（带错误详情）
-     */
-    private void updateTaskStatus(Long taskId, String status, String currentStep, Integer progress,
-                                  String message, String resultVideoUrl, String errorDetail) {
-        TikDigitalHumanTaskDO updateObj = new TikDigitalHumanTaskDO();
-        updateObj.setId(taskId);
-        updateObj.setStatus(status);
-        updateObj.setCurrentStep(currentStep);
-        updateObj.setProgress(progress);
-
-        if ("SUCCESS".equals(status)) {
-            updateObj.setResultVideoUrl(resultVideoUrl);
-            updateObj.setFinishTime(LocalDateTime.now());
-        } else if ("PROCESSING".equals(status)) {
-            updateObj.setStartTime(LocalDateTime.now());
-        } else if ("FAILED".equals(status)) {
-            updateObj.setErrorMessage(message);
-            updateObj.setErrorDetail(errorDetail);
-            updateObj.setFinishTime(LocalDateTime.now());
-        }
-
-        taskMapper.updateById(updateObj);
-    }
-
 }
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/service/TikUserVoiceServiceImpl.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/service/TikUserVoiceServiceImpl.java
@@ -1,6 +1,7 @@
 package cn.iocoder.yudao.module.tik.voice.service;

 import cn.hutool.core.collection.CollUtil;
+import cn.hutool.core.util.ObjectUtil;
 import cn.hutool.core.util.StrUtil;
 import cn.hutool.http.HttpUtil;
 import cn.hutool.json.JSONArray;
@@ -144,34 +145,38 @@ public class TikUserVoiceServiceImpl implements TikUserVoiceService {
                .setLanguage(StrUtil.blankToDefault(createReqVO.getLanguage(), "zh-CN"))
                .setGender(StrUtil.blankToDefault(createReqVO.getGender(), "female"))
                .setNote(createReqVO.getNote())
-                .setTranscription(null); // 初始为空，表示未识别
+                .setTranscription(createReqVO.getText()); // 使用前端传入的文本
        voiceMapper.insert(voice);

        // 4. 调用语音克隆服务，生成 voice_id
-        try {
-            log.info("[createVoice][开始语音复刻，配音编号({})，文件ID({})，供应商({})]",
-                    voice.getId(), fileDO.getId(), createReqVO.getProviderType());
-            String fileAccessUrl = fileApi.presignGetUrl(fileDO.getUrl(), PRESIGN_URL_EXPIRATION_SECONDS);
+        if (StrUtil.isNotBlank(createReqVO.getText())) {
+            try {
+                log.info("[createVoice][开始语音复刻，配音编号({})，文件ID({})，供应商({})]",
+                        voice.getId(), fileDO.getId(), createReqVO.getProviderType());
+                String fileAccessUrl = fileApi.presignGetUrl(fileDO.getUrl(), PRESIGN_URL_EXPIRATION_SECONDS);

-            VoiceCloneProvider provider = voiceProviderFactory.getProvider(createReqVO.getProviderType());
-            String providerType = getProviderType(createReqVO.getProviderType(), provider);
-            String model = getModelByProvider(providerType);
+                VoiceCloneProvider provider = voiceProviderFactory.getProvider(createReqVO.getProviderType());
+                String providerType = getProviderType(createReqVO.getProviderType(), provider);
+                String model = getModelByProvider(providerType);

-            VoiceCloneRequest cloneRequest = new VoiceCloneRequest();
-            cloneRequest.setAudioUrl(fileAccessUrl);
-            cloneRequest.setModel(model);
-            cloneRequest.setPrefix("voice" + voice.getId());
-            cloneRequest.setTranscriptionText(voice.getTranscription());
+                VoiceCloneRequest cloneRequest = new VoiceCloneRequest();
+                cloneRequest.setAudioUrl(fileAccessUrl);
+                cloneRequest.setModel(model);
+                cloneRequest.setPrefix("voice" + voice.getId());
+                cloneRequest.setTranscriptionText(createReqVO.getText()); // 使用前端传入的文本

-            VoiceCloneResult cloneResult = provider.cloneVoice(cloneRequest);
-            String voiceId = cloneResult.getVoiceId();
+                VoiceCloneResult cloneResult = provider.cloneVoice(cloneRequest);
+                String voiceId = cloneResult.getVoiceId();

-            voice.setVoiceId(voiceId);
-            voiceMapper.updateById(voice);
+                voice.setVoiceId(voiceId);
+                voiceMapper.updateById(voice);

-            log.info("[createVoice][语音复刻成功，配音编号({})，voice_id({})]", voice.getId(), voiceId);
-        } catch (Exception e) {
-            log.error("[createVoice][语音复刻失败，配音编号({})，错误信息: {}]", voice.getId(), e.getMessage(), e);
+                log.info("[createVoice][语音复刻成功，配音编号({})，voice_id({})]", voice.getId(), voiceId);
+            } catch (Exception e) {
+                log.error("[createVoice][语音复刻失败，配音编号({})，错误信息: {}]", voice.getId(), e.getMessage(), e);
+            }
+        } else {
+            log.info("[createVoice][未提供文本，跳过语音复刻，配音编号({})]", voice.getId());
        }


@@ -495,9 +500,9 @@ public class TikUserVoiceServiceImpl implements TikUserVoiceService {
        log.info("[previewVoice][试听，voiceConfigId={}, voiceId={}, userId={}]",
                voiceConfigId, reqVO.getVoiceId(), userId);

-        String voiceId = null;
-        String fileUrl = null;
-        String referenceText = null;
+        String voiceId;
+        String fileUrl;
+        String referenceText;

        // 1. 通过语音URL合成
        if (StrUtil.isNotBlank(reqVO.getFileUrl()) && StrUtil.isNotBlank(reqVO.getTranscriptionText())) {
@@ -506,6 +511,7 @@ public class TikUserVoiceServiceImpl implements TikUserVoiceService {
                    ? reqVO.getFileUrl()
                    : fileApi.presignGetUrl(rawFileUrl, PRESIGN_URL_EXPIRATION_SECONDS);
            referenceText = reqVO.getTranscriptionText();
+            voiceId = null;
        }
        // 2. 用户配音
        else if (voiceConfigId != null) {
@@ -514,8 +520,10 @@ public class TikUserVoiceServiceImpl implements TikUserVoiceService {
                throw exception(VOICE_NOT_EXISTS, "配音不存在");
            }

-            if (StrUtil.isNotBlank(voice.getVoiceId())) {
-                voiceId = voice.getVoiceId();
+            voiceId = voice.getVoiceId();
+            if (StrUtil.isNotBlank(voiceId)) {
+                fileUrl = null;
+                referenceText = null;
            } else {
                FileDO fileDO = fileMapper.selectById(voice.getFileId());
                if (fileDO == null) {
@@ -534,14 +542,14 @@ public class TikUserVoiceServiceImpl implements TikUserVoiceService {
            if (StrUtil.isBlank(voiceId)) {
                throw exception(VOICE_NOT_EXISTS, "系统配音音色ID不能为空");
            }
+            fileUrl = null;
+            referenceText = null;
        }

-        // 统一处理：使用前端传入的 inputText，否则使用默认试听文本
        String finalText = StrUtil.blankToDefault(reqVO.getInputText(), getPreviewText());
-
        String instruction = reqVO.getInstruction();
-        Float speechRate = reqVO.getSpeechRate() != null ? reqVO.getSpeechRate() : 1.0f;
-        Float volume = reqVO.getVolume() != null ? reqVO.getVolume() : 0f;
+        Float speechRate = ObjectUtil.defaultIfNull(reqVO.getSpeechRate(), 1.0f);
+        Float volume = ObjectUtil.defaultIfNull(reqVO.getVolume(), 0f);
        String audioFormat = StrUtil.blankToDefault(reqVO.getAudioFormat(), "mp3");

        // 缓存
--- a/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/vo/AppTikUserVoiceCreateReqVO.java
+++ b/yudao-module-tik/src/main/java/cn/iocoder/yudao/module/tik/voice/vo/AppTikUserVoiceCreateReqVO.java
@@ -3,6 +3,7 @@ package cn.iocoder.yudao.module.tik.voice.vo;
 import io.swagger.v3.oas.annotations.media.Schema;
 import jakarta.validation.constraints.NotBlank;
 import jakarta.validation.constraints.NotNull;
+import jakarta.validation.constraints.Size;
 import lombok.Data;

 /**
@@ -34,6 +35,10 @@ public class AppTikUserVoiceCreateReqVO {
    @Schema(description = "备注", example = "这是一个测试配音")
    private String note;

+    @Schema(description = "音频文本（用于语音复刻，前端通过音频识别获取）")
+    @Size(max = 4000, message = "音频文本不能超过 4000 个字符")
+    private String text;
+
    @Schema(description = "供应商类型：cosyvoice-阿里云，siliconflow-硅基流动（不传则使用默认）", example = "cosyvoice")
    private String providerType;