feat: 尝试增加一下聊天里面的语音转文字功能

2026-03-24 23:06:51 +00:00 · 2026-01-17 05:14:14 +08:00
parent 095c8f0db6
commit 72e2d82158
18 changed files with 999 additions and 66 deletions
--- a/electron/main.ts
+++ b/electron/main.ts
@@ -15,6 +15,7 @@ import { groupAnalyticsService } from './services/groupAnalyticsService'
 import { annualReportService } from './services/annualReportService'
 import { exportService, ExportOptions } from './services/exportService'
 import { KeyService } from './services/keyService'
+import { voiceTranscribeService } from './services/voiceTranscribeService'


 // 配置自动更新
@@ -442,6 +443,10 @@ function registerIpcHandlers() {
    return chatService.getVoiceData(sessionId, msgId)
  })

+  ipcMain.handle('chat:getVoiceTranscript', async (_, sessionId: string, msgId: string) => {
+    return chatService.getVoiceTranscript(sessionId, msgId)
+  })
+
  ipcMain.handle('chat:getMessageById', async (_, sessionId: string, localId: number) => {
    return chatService.getMessageById(sessionId, localId)
  })
@@ -516,6 +521,16 @@ function registerIpcHandlers() {
    return { success: true }
  })

+  ipcMain.handle('whisper:downloadModel', async (event, payload: { modelName: string; downloadDir?: string; source?: string }) => {
+    return voiceTranscribeService.downloadModel(payload, (progress) => {
+      event.sender.send('whisper:downloadProgress', progress)
+    })
+  })
+
+  ipcMain.handle('whisper:getModelStatus', async (_, payload: { modelName: string; downloadDir?: string }) => {
+    return voiceTranscribeService.getModelStatus(payload)
+  })
+
  // 群聊分析相关
  ipcMain.handle('groupAnalytics:getGroupChats', async () => {
    return groupAnalyticsService.getGroupChats()
--- a/electron/preload.ts
+++ b/electron/preload.ts
@@ -106,7 +106,8 @@ contextBridge.exposeInMainWorld('electronAPI', {
    close: () => ipcRenderer.invoke('chat:close'),
    getSessionDetail: (sessionId: string) => ipcRenderer.invoke('chat:getSessionDetail', sessionId),
    getImageData: (sessionId: string, msgId: string) => ipcRenderer.invoke('chat:getImageData', sessionId, msgId),
-    getVoiceData: (sessionId: string, msgId: string) => ipcRenderer.invoke('chat:getVoiceData', sessionId, msgId)
+    getVoiceData: (sessionId: string, msgId: string) => ipcRenderer.invoke('chat:getVoiceData', sessionId, msgId),
+    getVoiceTranscript: (sessionId: string, msgId: string) => ipcRenderer.invoke('chat:getVoiceTranscript', sessionId, msgId)
  },


@@ -174,5 +175,16 @@ contextBridge.exposeInMainWorld('electronAPI', {
      ipcRenderer.invoke('export:exportSessions', sessionIds, outputDir, options),
    exportSession: (sessionId: string, outputPath: string, options: any) =>
      ipcRenderer.invoke('export:exportSession', sessionId, outputPath, options)
+  },
+
+  whisper: {
+    downloadModel: (payload: { modelName: string; downloadDir?: string; source?: string }) =>
+      ipcRenderer.invoke('whisper:downloadModel', payload),
+    getModelStatus: (payload: { modelName: string; downloadDir?: string }) =>
+      ipcRenderer.invoke('whisper:getModelStatus', payload),
+    onDownloadProgress: (callback: (payload: { modelName: string; downloadedBytes: number; totalBytes?: number; percent?: number }) => void) => {
+      ipcRenderer.on('whisper:downloadProgress', (_, payload) => callback(payload))
+      return () => ipcRenderer.removeAllListeners('whisper:downloadProgress')
+    }
  }
 })
--- a/electron/services/chatService.ts
+++ b/electron/services/chatService.ts
@@ -16,6 +16,7 @@ import { ConfigService } from './config'
 import { wcdbService } from './wcdbService'
 import { MessageCacheService } from './messageCacheService'
 import { ContactCacheService, ContactCacheEntry } from './contactCacheService'
+import { voiceTranscribeService } from './voiceTranscribeService'

 type HardlinkState = {
  db: Database.Database
@@ -83,6 +84,10 @@ class ChatService {
  private hardlinkCache = new Map<string, HardlinkState>()
  private readonly contactCacheService: ContactCacheService
  private readonly messageCacheService: MessageCacheService
+  private voiceWavCache = new Map<string, Buffer>()
+  private voiceTranscriptCache = new Map<string, string>()
+  private voiceTranscriptPending = new Map<string, Promise<{ success: boolean; transcript?: string; error?: string }>>()
+  private readonly voiceCacheMaxEntries = 50

  constructor() {
    this.configService = new ConfigService()
@@ -1738,6 +1743,9 @@ class ChatService {

    if (includeMessages) {
      this.messageCacheService.clear()
+      this.voiceWavCache.clear()
+      this.voiceTranscriptCache.clear()
+      this.voiceTranscriptPending.clear()
    }

    for (const state of this.hardlinkCache.values()) {
@@ -2263,6 +2271,8 @@ class ChatService {
        const pcmData = readFileSync(pcmFile)
        const wavHeader = this.createWavHeader(pcmData.length, 24000, 1) // 微信语音通常 24kHz
        const wavData = Buffer.concat([wavHeader, pcmData])
+        const cacheKey = this.getVoiceCacheKey(sessionId, msgId)
+        this.cacheVoiceWav(cacheKey, wavData)

        return { success: true, data: wavData.toString('base64') }
      } finally {
@@ -2276,6 +2286,45 @@ class ChatService {
    }
  }

+  async getVoiceTranscript(sessionId: string, msgId: string): Promise<{ success: boolean; transcript?: string; error?: string }> {
+    const cacheKey = this.getVoiceCacheKey(sessionId, msgId)
+    const cached = this.voiceTranscriptCache.get(cacheKey)
+    if (cached) {
+      return { success: true, transcript: cached }
+    }
+
+    const pending = this.voiceTranscriptPending.get(cacheKey)
+    if (pending) {
+      return pending
+    }
+
+    const task = (async () => {
+      try {
+        let wavData = this.voiceWavCache.get(cacheKey)
+        if (!wavData) {
+          const voiceResult = await this.getVoiceData(sessionId, msgId)
+          if (!voiceResult.success || !voiceResult.data) {
+            return { success: false, error: voiceResult.error || '语音解码失败' }
+          }
+          wavData = Buffer.from(voiceResult.data, 'base64')
+        }
+
+        const result = await voiceTranscribeService.transcribeWavBuffer(wavData)
+        if (result.success && result.transcript) {
+          this.cacheVoiceTranscript(cacheKey, result.transcript)
+        }
+        return result
+      } catch (error) {
+        return { success: false, error: String(error) }
+      } finally {
+        this.voiceTranscriptPending.delete(cacheKey)
+      }
+    })()
+
+    this.voiceTranscriptPending.set(cacheKey, task)
+    return task
+  }
+
  private createWavHeader(pcmLength: number, sampleRate: number = 24000, channels: number = 1): Buffer {
    const header = Buffer.alloc(44)
    header.write('RIFF', 0)
@@ -2294,6 +2343,26 @@ class ChatService {
    return header
  }

+  private getVoiceCacheKey(sessionId: string, msgId: string): string {
+    return `${sessionId}:${msgId}`
+  }
+
+  private cacheVoiceWav(cacheKey: string, wavData: Buffer): void {
+    this.voiceWavCache.set(cacheKey, wavData)
+    if (this.voiceWavCache.size > this.voiceCacheMaxEntries) {
+      const oldestKey = this.voiceWavCache.keys().next().value
+      if (oldestKey) this.voiceWavCache.delete(oldestKey)
+    }
+  }
+
+  private cacheVoiceTranscript(cacheKey: string, transcript: string): void {
+    this.voiceTranscriptCache.set(cacheKey, transcript)
+    if (this.voiceTranscriptCache.size > this.voiceCacheMaxEntries) {
+      const oldestKey = this.voiceTranscriptCache.keys().next().value
+      if (oldestKey) this.voiceTranscriptCache.delete(oldestKey)
+    }
+  }
+
  async getMessageById(sessionId: string, localId: number): Promise<{ success: boolean; message?: Message; error?: string }> {
    try {
      console.info('[ChatService] getMessageById (SQL)', { sessionId, localId })
--- a/electron/services/config.ts
+++ b/electron/services/config.ts
@@ -20,6 +20,9 @@ interface ConfigSchema {
  language: string
  logEnabled: boolean
  llmModelPath: string
+  whisperModelName: string
+  whisperModelDir: string
+  whisperDownloadSource: string
 }

 export class ConfigService {
@@ -42,7 +45,10 @@ export class ConfigService {
        themeId: 'cloud-dancer',
        language: 'zh-CN',
        logEnabled: false,
-        llmModelPath: ''
+        llmModelPath: '',
+        whisperModelName: 'base',
+        whisperModelDir: '',
+        whisperDownloadSource: 'tsinghua'
      }
    })
  }
--- a/electron/services/voiceTranscribeService.ts
+++ b/electron/services/voiceTranscribeService.ts
@@ -0,0 +1,281 @@
+import { app } from 'electron'
+import { createWriteStream, existsSync, mkdirSync, statSync, unlinkSync, writeFileSync } from 'fs'
+import { join, dirname } from 'path'
+import { promisify } from 'util'
+import { execFile, spawnSync } from 'child_process'
+import * as https from 'https'
+import * as http from 'http'
+import { ConfigService } from './config'
+
+const execFileAsync = promisify(execFile)
+
+type WhisperModelInfo = {
+  name: string
+  fileName: string
+  sizeLabel: string
+  sizeBytes?: number
+}
+
+type DownloadProgress = {
+  modelName: string
+  downloadedBytes: number
+  totalBytes?: number
+  percent?: number
+}
+
+const WHISPER_MODELS: Record<string, WhisperModelInfo> = {
+  tiny: { name: 'tiny', fileName: 'ggml-tiny.bin', sizeLabel: '75 MB', sizeBytes: 75_000_000 },
+  base: { name: 'base', fileName: 'ggml-base.bin', sizeLabel: '142 MB', sizeBytes: 142_000_000 },
+  small: { name: 'small', fileName: 'ggml-small.bin', sizeLabel: '466 MB', sizeBytes: 466_000_000 },
+  medium: { name: 'medium', fileName: 'ggml-medium.bin', sizeLabel: '1.5 GB', sizeBytes: 1_500_000_000 },
+  'large-v3': { name: 'large-v3', fileName: 'ggml-large-v3.bin', sizeLabel: '2.9 GB', sizeBytes: 2_900_000_000 }
+}
+
+const WHISPER_SOURCES: Record<string, string> = {
+  official: 'https://huggingface.co/ggerganov/whisper.cpp/resolve/main',
+  tsinghua: 'https://hf-mirror.com/ggerganov/whisper.cpp/resolve/main'
+}
+
+function getStaticFfmpegPath(): string | null {
+  try {
+    // eslint-disable-next-line @typescript-eslint/no-var-requires
+    const ffmpegStatic = require('ffmpeg-static')
+    if (typeof ffmpegStatic === 'string' && existsSync(ffmpegStatic)) {
+      return ffmpegStatic
+    }
+    const devPath = join(process.cwd(), 'node_modules', 'ffmpeg-static', 'ffmpeg.exe')
+    if (existsSync(devPath)) {
+      return devPath
+    }
+    if (app.isPackaged) {
+      const resourcesPath = process.resourcesPath
+      const packedPath = join(resourcesPath, 'app.asar.unpacked', 'node_modules', 'ffmpeg-static', 'ffmpeg.exe')
+      if (existsSync(packedPath)) {
+        return packedPath
+      }
+    }
+    return null
+  } catch {
+    return null
+  }
+}
+
+export class VoiceTranscribeService {
+  private configService = new ConfigService()
+  private downloadTasks = new Map<string, Promise<{ success: boolean; path?: string; error?: string }>>()
+
+  private resolveModelInfo(modelName: string): WhisperModelInfo | null {
+    return WHISPER_MODELS[modelName] || null
+  }
+
+  private resolveModelDir(overrideDir?: string): string {
+    const configured = overrideDir || this.configService.get('whisperModelDir')
+    if (configured) return configured
+    return join(app.getPath('userData'), 'models', 'whisper')
+  }
+
+  private resolveModelPath(modelName: string, overrideDir?: string): string | null {
+    const info = this.resolveModelInfo(modelName)
+    if (!info) return null
+    return join(this.resolveModelDir(overrideDir), info.fileName)
+  }
+
+  private resolveSourceUrl(overrideSource?: string): string {
+    const configured = overrideSource || this.configService.get('whisperDownloadSource')
+    if (configured && WHISPER_SOURCES[configured]) return WHISPER_SOURCES[configured]
+    return WHISPER_SOURCES.official
+  }
+
+  async getModelStatus(payload: { modelName: string; downloadDir?: string }): Promise<{
+    success: boolean
+    exists?: boolean
+    path?: string
+    sizeBytes?: number
+    error?: string
+  }> {
+    const modelPath = this.resolveModelPath(payload.modelName, payload.downloadDir)
+    if (!modelPath) {
+      return { success: false, error: '未知模型名称' }
+    }
+    if (!existsSync(modelPath)) {
+      return { success: true, exists: false, path: modelPath }
+    }
+    const sizeBytes = statSync(modelPath).size
+    return { success: true, exists: true, path: modelPath, sizeBytes }
+  }
+
+  async downloadModel(
+    payload: { modelName: string; downloadDir?: string; source?: string },
+    onProgress?: (progress: DownloadProgress) => void
+  ): Promise<{ success: boolean; path?: string; error?: string }> {
+    const info = this.resolveModelInfo(payload.modelName)
+    if (!info) {
+      return { success: false, error: '未知模型名称' }
+    }
+
+    const modelPath = this.resolveModelPath(payload.modelName, payload.downloadDir)
+    if (!modelPath) {
+      return { success: false, error: '模型路径生成失败' }
+    }
+
+    if (existsSync(modelPath)) {
+      return { success: true, path: modelPath }
+    }
+
+    const cacheKey = `${payload.modelName}:${modelPath}`
+    const pending = this.downloadTasks.get(cacheKey)
+    if (pending) return pending
+
+    const task = (async () => {
+      try {
+        const targetDir = this.resolveModelDir(payload.downloadDir)
+        if (!existsSync(targetDir)) {
+          mkdirSync(targetDir, { recursive: true })
+        }
+
+        const baseUrl = this.resolveSourceUrl(payload.source)
+        const url = `${baseUrl}/${info.fileName}`
+        await this.downloadToFile(url, modelPath, payload.modelName, onProgress)
+        return { success: true, path: modelPath }
+      } catch (error) {
+        try { if (existsSync(modelPath)) unlinkSync(modelPath) } catch { }
+        return { success: false, error: String(error) }
+      } finally {
+        this.downloadTasks.delete(cacheKey)
+      }
+    })()
+
+    this.downloadTasks.set(cacheKey, task)
+    return task
+  }
+
+  async transcribeWavBuffer(wavData: Buffer): Promise<{ success: boolean; transcript?: string; error?: string }> {
+    const modelName = this.configService.get('whisperModelName') || 'base'
+    const modelPath = this.resolveModelPath(modelName)
+    console.info('[VoiceTranscribe] check model', { modelName, modelPath, exists: modelPath ? existsSync(modelPath) : false })
+    if (!modelPath || !existsSync(modelPath)) {
+      return { success: false, error: '未下载语音模型，请在设置中下载' }
+    }
+
+    // 使用内置的预编译 whisper-cli.exe
+    const resourcesPath = app.isPackaged
+      ? join(process.resourcesPath, 'resources')
+      : join(app.getAppPath(), 'resources')
+    const whisperExe = join(resourcesPath, 'whisper-cli.exe')
+    
+    if (!existsSync(whisperExe)) {
+      return { success: false, error: '找不到语音转写程序，请重新安装应用' }
+    }
+
+    const ffmpegPath = getStaticFfmpegPath() || 'ffmpeg'
+    console.info('[VoiceTranscribe] ffmpeg path', ffmpegPath)
+
+    const tempDir = app.getPath('temp')
+    const fileToken = `${Date.now()}_${Math.random().toString(16).slice(2)}`
+    const inputPath = join(tempDir, `weflow_voice_${fileToken}.wav`)
+    const outputPath = join(tempDir, `weflow_voice_${fileToken}_16k.wav`)
+
+    try {
+      writeFileSync(inputPath, wavData)
+      console.info('[VoiceTranscribe] converting to 16kHz', { inputPath, outputPath })
+      await execFileAsync(ffmpegPath, ['-y', '-i', inputPath, '-ar', '16000', '-ac', '1', outputPath])
+      
+      console.info('[VoiceTranscribe] transcribing with whisper', { whisperExe, modelPath })
+      const { stdout } = await execFileAsync(whisperExe, [
+        '-m', modelPath,
+        '-f', outputPath,
+        '-l', 'zh',
+        '-otxt'
+      ], {
+        maxBuffer: 10 * 1024 * 1024,
+        cwd: tempDir
+      })
+
+      // 解析输出文本
+      const txtFile = outputPath.replace(/\.[^.]+$/, '.txt')
+      let transcript = ''
+      if (existsSync(txtFile)) {
+        const { readFileSync } = await import('fs')
+        transcript = readFileSync(txtFile, 'utf-8').trim()
+        unlinkSync(txtFile)
+      } else {
+        // 从 stdout 提取
+        const lines = stdout.split('\n').filter(line => {
+          const trimmed = line.trim()
+          return trimmed && !trimmed.startsWith('[') && !trimmed.startsWith('whisper_')
+        })
+        transcript = lines.join(' ').trim()
+      }
+
+      console.info('[VoiceTranscribe] success', { transcript })
+      return { success: true, transcript }
+    } catch (error) {
+      console.error('[VoiceTranscribe] failed', error)
+      return { success: false, error: String(error) }
+    } finally {
+      try { if (existsSync(inputPath)) unlinkSync(inputPath) } catch { }
+      try { if (existsSync(outputPath)) unlinkSync(outputPath) } catch { }
+    }
+  }
+
+  private downloadToFile(
+    url: string,
+    targetPath: string,
+    modelName: string,
+    onProgress?: (progress: DownloadProgress) => void,
+    remainingRedirects = 3
+  ): Promise<void> {
+    return new Promise((resolve, reject) => {
+      const protocol = url.startsWith('https') ? https : http
+      const request = protocol.get(url, (response) => {
+        if ([301, 302, 303, 307, 308].includes(response.statusCode || 0) && response.headers.location) {
+          if (remainingRedirects <= 0) {
+            reject(new Error('下载重定向次数过多'))
+            return
+          }
+          this.downloadToFile(response.headers.location, targetPath, modelName, onProgress, remainingRedirects - 1)
+            .then(resolve)
+            .catch(reject)
+          return
+        }
+
+        if (response.statusCode !== 200) {
+          reject(new Error(`下载失败: ${response.statusCode}`))
+          return
+        }
+
+        const totalBytes = Number(response.headers['content-length'] || 0) || undefined
+        let downloadedBytes = 0
+
+        const writer = createWriteStream(targetPath)
+
+        response.on('data', (chunk) => {
+          downloadedBytes += chunk.length
+          const percent = totalBytes ? (downloadedBytes / totalBytes) * 100 : undefined
+          onProgress?.({ modelName, downloadedBytes, totalBytes, percent })
+        })
+
+        response.on('error', (error) => {
+          try { writer.close() } catch { }
+          reject(error)
+        })
+
+        writer.on('error', (error) => {
+          try { writer.close() } catch { }
+          reject(error)
+        })
+
+        writer.on('finish', () => {
+          writer.close()
+          resolve()
+        })
+
+        response.pipe(writer)
+      })
+
+      request.on('error', reject)
+    })
+  }
+}
+
+export const voiceTranscribeService = new VoiceTranscribeService()
--- a/electron/types/whisper-node.d.ts
+++ b/electron/types/whisper-node.d.ts
@@ -0,0 +1,22 @@
+declare module 'whisper-node' {
+  export type WhisperSegment = {
+    start: string
+    end: string
+    speech: string
+  }
+
+  export type WhisperOptions = {
+    modelName?: string
+    modelPath?: string
+    whisperOptions?: {
+      language?: string
+      gen_file_txt?: boolean
+      gen_file_subtitle?: boolean
+      gen_file_vtt?: boolean
+      word_timestamps?: boolean
+      timestamp_size?: number
+    }
+  }
+
+  export default function whisper(filePath: string, options?: WhisperOptions): Promise<WhisperSegment[]>
+}
--- a/package-lock.json
+++ b/package-lock.json
@@ -25,6 +25,7 @@
        "react-dom": "^19.2.3",
        "react-router-dom": "^7.1.1",
        "wechat-emojis": "^1.0.2",
+        "whisper-node": "^1.1.1",
        "zustand": "^5.0.2"
      },
      "devDependencies": {
@@ -3746,7 +3747,6 @@
      "version": "1.0.2",
      "resolved": "https://registry.npmmirror.com/balanced-match/-/balanced-match-1.0.2.tgz",
      "integrity": "sha512-3oSeUO0TMV67hN1AmbXsK4yaqU7tjiHlbxRDZOpH0KW9+CeX4bRAaX0Anxt0tx2MrpRpWwQaPwIlISEJhYU5Pw==",
-      "dev": true,
      "license": "MIT"
    },
    "node_modules/base64-arraybuffer": {
@@ -3852,7 +3852,6 @@
      "version": "1.1.12",
      "resolved": "https://registry.npmmirror.com/brace-expansion/-/brace-expansion-1.1.12.tgz",
      "integrity": "sha512-9T9UjW3r0UW5c1Q7GTwllptXwhvYmEzFhzMfZ9H7FQWt+uZePjZPjBP/W1ZEyZ1twGWom5/56TF4lPcqjnDHcg==",
-      "dev": true,
      "license": "MIT",
      "dependencies": {
        "balanced-match": "^1.0.0",
@@ -4451,7 +4450,6 @@
      "version": "0.0.1",
      "resolved": "https://registry.npmmirror.com/concat-map/-/concat-map-0.0.1.tgz",
      "integrity": "sha512-/Srv4dswyQNBfohGpz9o6Yb3Gz3SrUDqBH5rTuhGR7ahtlbYKnVxw2bCFMRljaA7EXHaXZ8wsHdodFvbkhKmqg==",
-      "dev": true,
      "license": "MIT"
    },
    "node_modules/concat-stream": {
@@ -5792,7 +5790,6 @@
      "version": "1.0.0",
      "resolved": "https://registry.npmmirror.com/fs.realpath/-/fs.realpath-1.0.0.tgz",
      "integrity": "sha512-OO0pH2lK6a0hZnAdau5ItzHPI6pUlvI7jMVnxUQRtw4owF2wk8lOSabtGDCTP4Ggrg2MbGnWO9X8K1t4+fGMDw==",
-      "dev": true,
      "license": "ISC"
    },
    "node_modules/fsevents": {
@@ -5814,7 +5811,6 @@
      "version": "1.1.2",
      "resolved": "https://registry.npmmirror.com/function-bind/-/function-bind-1.1.2.tgz",
      "integrity": "sha512-7XHNxH7qX9xG5mIwxkhumTox/MIRNcOgDrxWsMt2pAr23WHp6MrRlN7FBSFpCpr+oVO0F744iUgR82nJMfG2SA==",
-      "dev": true,
      "license": "MIT",
      "funding": {
        "url": "https://github.com/sponsors/ljharb"
@@ -5933,7 +5929,6 @@
      "resolved": "https://registry.npmmirror.com/glob/-/glob-7.2.3.tgz",
      "integrity": "sha512-nFR0zLpU2YCaRxwoCJvL6UvCH2JFyFVIvwTLsIf21AuHlMskA1hhTdk+LlYJtOlYt9v6dvszD2BGRqBL+iQK9Q==",
      "deprecated": "Glob versions prior to v9 are no longer supported",
-      "dev": true,
      "license": "ISC",
      "dependencies": {
        "fs.realpath": "^1.0.0",
@@ -5954,7 +5949,6 @@
      "version": "3.1.2",
      "resolved": "https://registry.npmmirror.com/minimatch/-/minimatch-3.1.2.tgz",
      "integrity": "sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==",
-      "dev": true,
      "license": "ISC",
      "dependencies": {
        "brace-expansion": "^1.1.7"
@@ -6109,7 +6103,6 @@
      "version": "2.0.2",
      "resolved": "https://registry.npmmirror.com/hasown/-/hasown-2.0.2.tgz",
      "integrity": "sha512-0hJU9SCPvmMzIBdZFqNPXWa6dqh7WdH0cII9y+CyS8rG3nL48Bclra9HmKhVVUHyPWNH5Y7xDwAB7bfgSjkUMQ==",
-      "dev": true,
      "license": "MIT",
      "dependencies": {
        "function-bind": "^1.1.2"
@@ -6334,7 +6327,6 @@
      "resolved": "https://registry.npmmirror.com/inflight/-/inflight-1.0.6.tgz",
      "integrity": "sha512-k92I/b08q4wvFscXCLvqfsHCrjrF7yiXsQuIVvVE7N82W3+aqpzuUdBbfhWcy/FZR3/4IgflMgKLOsvPDrGCJA==",
      "deprecated": "This module is not supported, and leaks memory. Do not use it. Check out lru-cache if you want a good and tested way to coalesce async requests by a key value, which is much more comprehensive and powerful.",
-      "dev": true,
      "license": "ISC",
      "dependencies": {
        "once": "^1.3.0",
@@ -6353,6 +6345,15 @@
      "integrity": "sha512-JV/yugV2uzW5iMRSiZAyDtQd+nxtUnjeLt0acNdw98kKLrvuRVyB80tsREOE7yvGVgalhZ6RNXCmEHkUKBKxew==",
      "license": "ISC"
    },
+    "node_modules/interpret": {
+      "version": "1.4.0",
+      "resolved": "https://registry.npmmirror.com/interpret/-/interpret-1.4.0.tgz",
+      "integrity": "sha512-agE4QfB2Lkp9uICn7BAqoscw4SZP9kTE2hxiFI3jBPmXJfdqiahTbUuKGsMoN2GtqL9AxhYioAcVvgsb1HvRbA==",
+      "license": "MIT",
+      "engines": {
+        "node": ">= 0.10"
+      }
+    },
    "node_modules/ip-address": {
      "version": "10.1.0",
      "resolved": "https://registry.npmmirror.com/ip-address/-/ip-address-10.1.0.tgz",
@@ -6376,6 +6377,21 @@
        "is-ci": "bin.js"
      }
    },
+    "node_modules/is-core-module": {
+      "version": "2.16.1",
+      "resolved": "https://registry.npmmirror.com/is-core-module/-/is-core-module-2.16.1.tgz",
+      "integrity": "sha512-UfoeMA6fIJ8wTYFEUjelnaGI67v6+N7qXJEvQuIGa99l4xsCruSYOVSQ0uPANn4dAzm8lkYPaKLrrijLq7x23w==",
+      "license": "MIT",
+      "dependencies": {
+        "hasown": "^2.0.2"
+      },
+      "engines": {
+        "node": ">= 0.4"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/ljharb"
+      }
+    },
    "node_modules/is-extglob": {
      "version": "2.1.1",
      "resolved": "https://registry.npmmirror.com/is-extglob/-/is-extglob-2.1.1.tgz",
@@ -7506,7 +7522,6 @@
      "version": "1.0.1",
      "resolved": "https://registry.npmmirror.com/path-is-absolute/-/path-is-absolute-1.0.1.tgz",
      "integrity": "sha512-AVbw3UJ2e9bq64vSaS9Am0fje1Pa8pbGqTTsmXfaIiMpnr5DlDhfJOuLj9Sf95ZPVDAUerDfEk88MPmPe7UCQg==",
-      "dev": true,
      "license": "MIT",
      "engines": {
        "node": ">=0.10.0"
@@ -7522,6 +7537,12 @@
        "node": ">=8"
      }
    },
+    "node_modules/path-parse": {
+      "version": "1.0.7",
+      "resolved": "https://registry.npmmirror.com/path-parse/-/path-parse-1.0.7.tgz",
+      "integrity": "sha512-LDJzPVEEEPR+y48z93A0Ed0yXb8pAByGWo/k5YYdYgpY2/2EsOsksJrq7lOHxryrVOn1ejG6oAp8ahvOIQD8sw==",
+      "license": "MIT"
+    },
    "node_modules/path-scurry": {
      "version": "1.11.1",
      "resolved": "https://registry.npmmirror.com/path-scurry/-/path-scurry-1.11.1.tgz",
@@ -7911,6 +7932,26 @@
        "url": "https://paulmillr.com/funding/"
      }
    },
+    "node_modules/readline-sync": {
+      "version": "1.4.10",
+      "resolved": "https://registry.npmmirror.com/readline-sync/-/readline-sync-1.4.10.tgz",
+      "integrity": "sha512-gNva8/6UAe8QYepIQH/jQ2qn91Qj0B9sYjMBBs3QOB8F2CXcKgLxQaJRP76sWVRQt+QU+8fAkCbCvjjMFu7Ycw==",
+      "license": "MIT",
+      "engines": {
+        "node": ">= 0.8.0"
+      }
+    },
+    "node_modules/rechoir": {
+      "version": "0.6.2",
+      "resolved": "https://registry.npmmirror.com/rechoir/-/rechoir-0.6.2.tgz",
+      "integrity": "sha512-HFM8rkZ+i3zrV+4LQjwQ0W+ez98pApMGM3HUrN04j3CqzPOzl9nmP15Y8YXNm8QHGv/eacOVEjqhmWpkRV0NAw==",
+      "dependencies": {
+        "resolve": "^1.1.6"
+      },
+      "engines": {
+        "node": ">= 0.10"
+      }
+    },
    "node_modules/require-directory": {
      "version": "2.1.1",
      "resolved": "https://registry.npmmirror.com/require-directory/-/require-directory-2.1.1.tgz",
@@ -7948,6 +7989,26 @@
        "url": "https://github.com/sponsors/jet2jet"
      }
    },
+    "node_modules/resolve": {
+      "version": "1.22.11",
+      "resolved": "https://registry.npmmirror.com/resolve/-/resolve-1.22.11.tgz",
+      "integrity": "sha512-RfqAvLnMl313r7c9oclB1HhUEAezcpLjz95wFH4LVuhk9JF/r22qmVP9AMmOU4vMX7Q8pN8jwNg/CSpdFnMjTQ==",
+      "license": "MIT",
+      "dependencies": {
+        "is-core-module": "^2.16.1",
+        "path-parse": "^1.0.7",
+        "supports-preserve-symlinks-flag": "^1.0.0"
+      },
+      "bin": {
+        "resolve": "bin/resolve"
+      },
+      "engines": {
+        "node": ">= 0.4"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/ljharb"
+      }
+    },
    "node_modules/resolve-alpn": {
      "version": "1.2.1",
      "resolved": "https://registry.npmmirror.com/resolve-alpn/-/resolve-alpn-1.2.1.tgz",
@@ -8281,6 +8342,23 @@
        "node": ">=8"
      }
    },
+    "node_modules/shelljs": {
+      "version": "0.8.5",
+      "resolved": "https://registry.npmmirror.com/shelljs/-/shelljs-0.8.5.tgz",
+      "integrity": "sha512-TiwcRcrkhHvbrZbnRcFYMLl30Dfov3HKqzp5tO5b4pt6G/SezKcYhmDg15zXVBswHmctSAQKznqNW2LO5tTDow==",
+      "license": "BSD-3-Clause",
+      "dependencies": {
+        "glob": "^7.0.0",
+        "interpret": "^1.0.0",
+        "rechoir": "^0.6.2"
+      },
+      "bin": {
+        "shjs": "bin/shjs"
+      },
+      "engines": {
+        "node": ">=4"
+      }
+    },
    "node_modules/signal-exit": {
      "version": "3.0.7",
      "resolved": "https://registry.npmmirror.com/signal-exit/-/signal-exit-3.0.7.tgz",
@@ -8588,6 +8666,18 @@
        "node": ">=8"
      }
    },
+    "node_modules/supports-preserve-symlinks-flag": {
+      "version": "1.0.0",
+      "resolved": "https://registry.npmmirror.com/supports-preserve-symlinks-flag/-/supports-preserve-symlinks-flag-1.0.0.tgz",
+      "integrity": "sha512-ot0WnXS9fgdkgIcePe6RHNk1WA8+muPa6cSjeR3V8K27q9BB1rTE3R1p7Hv0z1ZyAc8s6Vvv8DIyWf681MAt0w==",
+      "license": "MIT",
+      "engines": {
+        "node": ">= 0.4"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/ljharb"
+      }
+    },
    "node_modules/tar": {
      "version": "6.2.1",
      "resolved": "https://registry.npmmirror.com/tar/-/tar-6.2.1.tgz",
@@ -9219,6 +9309,19 @@
        "node": ">= 8"
      }
    },
+    "node_modules/whisper-node": {
+      "version": "1.1.1",
+      "resolved": "https://registry.npmmirror.com/whisper-node/-/whisper-node-1.1.1.tgz",
+      "integrity": "sha512-s1czx7pL0g63QOz0X9oAu7vOf4GzmFfQIy6J7msOAH5Yyiy+4a3w6+Uv0hiHvHkfBWk/+hG8nY3VEFdIapF83g==",
+      "license": "MIT",
+      "dependencies": {
+        "readline-sync": "^1.4.10",
+        "shelljs": "^0.8.5"
+      },
+      "bin": {
+        "download": "dist/download.js"
+      }
+    },
    "node_modules/wide-align": {
      "version": "1.1.5",
      "resolved": "https://registry.npmmirror.com/wide-align/-/wide-align-1.1.5.tgz",
--- a/package.json
+++ b/package.json
@@ -30,6 +30,7 @@
    "react-dom": "^19.2.3",
    "react-router-dom": "^7.1.1",
    "wechat-emojis": "^1.0.2",
+    "whisper-node": "^1.1.1",
    "zustand": "^5.0.2"
  },
  "devDependencies": {
--- a/resources/whisper-cli.exe
+++ b/resources/whisper-cli.exe
--- a/resources/whisper-main.exe
+++ b/resources/whisper-main.exe
--- a/src/pages/ChatPage.scss
+++ b/src/pages/ChatPage.scss
@@ -1303,6 +1303,12 @@
  cursor: pointer;
 }

+.voice-stack {
+  display: flex;
+  flex-direction: column;
+  gap: 6px;
+}
+
 .message-bubble.sent .voice-message {
  background: rgba(255, 255, 255, 0.18);
 }
@@ -1391,6 +1397,34 @@
  color: #d9480f;
 }

+.voice-transcript {
+  max-width: 260px;
+  padding: 8px 12px;
+  border-radius: 14px;
+  font-size: 13px;
+  line-height: 1.5;
+  background: var(--bg-secondary);
+  color: var(--text-primary);
+  border: 1px solid var(--border-color);
+  word-break: break-word;
+  white-space: pre-wrap;
+}
+
+.voice-transcript.sent {
+  background: rgba(255, 255, 255, 0.9);
+  color: var(--text-primary);
+  border-color: transparent;
+}
+
+.voice-transcript.received {
+  background: var(--card-bg);
+}
+
+.voice-transcript.error {
+  color: #d9480f;
+  cursor: pointer;
+}
+
@keyframes voicePulse {
  0% {
    height: 6px;
--- a/src/pages/ChatPage.tsx
+++ b/src/pages/ChatPage.tsx
@@ -1280,6 +1280,7 @@ function ChatPage(_props: ChatPageProps) {
 const emojiDataUrlCache = new Map<string, string>()
 const imageDataUrlCache = new Map<string, string>()
 const voiceDataUrlCache = new Map<string, string>()
+const voiceTranscriptCache = new Map<string, string>()
 const senderAvatarCache = new Map<string, { avatarUrl?: string; displayName?: string }>()
 const senderAvatarLoading = new Map<string, Promise<{ avatarUrl?: string; displayName?: string } | null>>()

@@ -1312,6 +1313,9 @@ function MessageBubble({ message, session, showTime, myAvatarUrl, isGroupChat }:
  const [voiceLoading, setVoiceLoading] = useState(false)
  const [isVoicePlaying, setIsVoicePlaying] = useState(false)
  const voiceAudioRef = useRef<HTMLAudioElement | null>(null)
+  const [voiceTranscriptLoading, setVoiceTranscriptLoading] = useState(false)
+  const [voiceTranscriptError, setVoiceTranscriptError] = useState(false)
+  const voiceTranscriptRequestedRef = useRef(false)
  const [showImagePreview, setShowImagePreview] = useState(false)

  // 从缓存获取表情包 data URL
@@ -1327,6 +1331,10 @@ function MessageBubble({ message, session, showTime, myAvatarUrl, isGroupChat }:
  const [voiceDataUrl, setVoiceDataUrl] = useState<string | undefined>(
    () => voiceDataUrlCache.get(voiceCacheKey)
  )
+  const voiceTranscriptCacheKey = `voice-transcript:${message.localId}`
+  const [voiceTranscript, setVoiceTranscript] = useState<string | undefined>(
+    () => voiceTranscriptCache.get(voiceTranscriptCacheKey)
+  )

  const formatTime = (timestamp: number): string => {
    if (!Number.isFinite(timestamp) || timestamp <= 0) return '未知时间'
@@ -1604,6 +1612,37 @@ function MessageBubble({ message, session, showTime, myAvatarUrl, isGroupChat }:
    }
  }, [isVoice])

+  const requestVoiceTranscript = useCallback(async () => {
+    if (voiceTranscriptLoading || voiceTranscriptRequestedRef.current) return
+    voiceTranscriptRequestedRef.current = true
+    setVoiceTranscriptLoading(true)
+    setVoiceTranscriptError(false)
+    try {
+      const result = await window.electronAPI.chat.getVoiceTranscript(session.username, String(message.localId))
+      if (result.success) {
+        const transcriptText = (result.transcript || '').trim()
+        voiceTranscriptCache.set(voiceTranscriptCacheKey, transcriptText)
+        setVoiceTranscript(transcriptText)
+      } else {
+        setVoiceTranscriptError(true)
+        voiceTranscriptRequestedRef.current = false
+      }
+    } catch {
+      setVoiceTranscriptError(true)
+      voiceTranscriptRequestedRef.current = false
+    } finally {
+      setVoiceTranscriptLoading(false)
+    }
+  }, [message.localId, session.username, voiceTranscriptCacheKey, voiceTranscriptLoading])
+
+  useEffect(() => {
+    if (!isVoice) return
+    if (!voiceDataUrl) return
+    if (voiceTranscriptError) return
+    if (voiceTranscriptLoading || voiceTranscript !== undefined || voiceTranscriptRequestedRef.current) return
+    void requestVoiceTranscript()
+  }, [isVoice, voiceDataUrl, voiceTranscript, voiceTranscriptError, voiceTranscriptLoading, requestVoiceTranscript])
+
  if (isSystem) {
    return (
      <div className="message-bubble system">
@@ -1762,8 +1801,21 @@ function MessageBubble({ message, session, showTime, myAvatarUrl, isGroupChat }:
      }

      const showDecryptHint = !voiceDataUrl && !voiceLoading && !isVoicePlaying
+      const showTranscript = Boolean(voiceDataUrl) && (voiceTranscriptLoading || voiceTranscriptError || voiceTranscript !== undefined)
+      const transcriptText = (voiceTranscript || '').trim()
+      const transcriptDisplay = voiceTranscriptLoading
+        ? '转写中...'
+        : voiceTranscriptError
+          ? '转写失败，点击重试'
+          : (transcriptText || '未识别到文字')
+      const handleTranscriptRetry = () => {
+        if (!voiceTranscriptError) return
+        voiceTranscriptRequestedRef.current = false
+        void requestVoiceTranscript()
+      }

      return (
+        <div className="voice-stack">
          <div className={`voice-message ${isVoicePlaying ? 'playing' : ''}`} onClick={handleToggle}>
            <button
              className="voice-play-btn"
@@ -1791,6 +1843,16 @@ function MessageBubble({ message, session, showTime, myAvatarUrl, isGroupChat }:
              {voiceError && <span className="voice-error">播放失败</span>}
            </div>
          </div>
+          {showTranscript && (
+            <div
+              className={`voice-transcript ${isSent ? 'sent' : 'received'}${voiceTranscriptError ? ' error' : ''}`}
+              onClick={handleTranscriptRetry}
+              title={voiceTranscriptError ? '点击重试语音转写' : undefined}
+            >
+              {transcriptDisplay}
+            </div>
+          )}
+        </div>
      )
    }

--- a/src/pages/SettingsPage.scss
+++ b/src/pages/SettingsPage.scss
@@ -204,6 +204,23 @@
    }
  }

+  select {
+    width: 100%;
+    padding: 10px 16px;
+    border: 1px solid var(--border-color);
+    border-radius: 9999px;
+    font-size: 14px;
+    background: var(--bg-primary);
+    color: var(--text-primary);
+    margin-bottom: 10px;
+    cursor: pointer;
+
+    &:focus {
+      outline: none;
+      border-color: var(--primary);
+    }
+  }
+  
  .input-with-toggle {
    position: relative;
    display: flex;
@@ -235,6 +252,93 @@
  }
 }

+.whisper-section {
+  .whisper-grid {
+    display: grid;
+    grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
+    gap: 12px;
+  }
+
+  .whisper-field {
+    display: flex;
+    flex-direction: column;
+  }
+
+  .field-label {
+    font-size: 12px;
+    color: var(--text-tertiary);
+    margin-bottom: 6px;
+  }
+
+  .whisper-status-line {
+    display: flex;
+    align-items: center;
+    gap: 8px;
+    font-size: 12px;
+    color: var(--text-secondary);
+    margin: 4px 0 10px;
+
+    .status {
+      padding: 2px 8px;
+      border-radius: 999px;
+      font-size: 11px;
+      font-weight: 500;
+      background: var(--bg-tertiary);
+      color: var(--text-secondary);
+    }
+
+    .status.ok {
+      background: rgba(16, 185, 129, 0.12);
+      color: #059669;
+    }
+
+    .status.warn {
+      background: rgba(245, 158, 11, 0.12);
+      color: #d97706;
+    }
+
+    .path {
+      flex: 1;
+      min-width: 0;
+      font-size: 11px;
+      color: var(--text-tertiary);
+      overflow: hidden;
+      text-overflow: ellipsis;
+      white-space: nowrap;
+    }
+  }
+
+  .whisper-progress {
+    display: flex;
+    align-items: center;
+    gap: 12px;
+    width: 100%;
+    max-width: 320px;
+
+    .progress-bar {
+      flex: 1;
+      height: 6px;
+      background: var(--bg-tertiary);
+      border-radius: 999px;
+      overflow: hidden;
+
+      .progress-fill {
+        height: 100%;
+        background: var(--primary);
+        border-radius: 999px;
+        transition: width 0.2s ease;
+      }
+    }
+
+    span {
+      font-size: 12px;
+      color: var(--text-secondary);
+      min-width: 36px;
+      text-align: right;
+    }
+  }
+}
+
 .log-toggle-line {
  display: flex;
  align-items: center;
--- a/src/pages/SettingsPage.tsx
+++ b/src/pages/SettingsPage.tsx
@@ -1,4 +1,4 @@
-import { useState, useEffect, useRef } from 'react'
+import { useState, useEffect, useRef } from 'react'
 import { useAppStore } from '../stores/appStore'
 import { useThemeStore, themes } from '../stores/themeStore'
 import { useAnalyticsStore } from '../stores/analyticsStore'
@@ -7,15 +7,29 @@ import * as configService from '../services/config'
 import {
  Eye, EyeOff, FolderSearch, FolderOpen, Search, Copy,
  RotateCcw, Trash2, Save, Plug, Check, Sun, Moon,
-  Palette, Database, Download, HardDrive, Info, RefreshCw, ChevronDown
+  Palette, Database, Download, HardDrive, Info, RefreshCw, ChevronDown, Mic
 } from 'lucide-react'
 import './SettingsPage.scss'

-type SettingsTab = 'appearance' | 'database' | 'cache' | 'about'
+type SettingsTab = 'appearance' | 'database' | 'whisper' | 'cache' | 'about'
+
+const whisperModels = [
+  { value: 'tiny', label: 'tiny (75 MB)' },
+  { value: 'base', label: 'base (142 MB)' },
+  { value: 'small', label: 'small (466 MB)' },
+  { value: 'medium', label: 'medium (1.5 GB)' },
+  { value: 'large-v3', label: 'large-v3 (2.9 GB)' }
+]
+
+const whisperSources = [
+  { value: 'official', label: 'HuggingFace 官方' },
+  { value: 'tsinghua', label: '清华镜像 (hf-mirror)' }
+]

 const tabs: { id: SettingsTab; label: string; icon: React.ElementType }[] = [
  { id: 'appearance', label: '外观', icon: Palette },
  { id: 'database', label: '数据库连接', icon: Database },
+  { id: 'whisper', label: '语音识别模型', icon: Mic },
  { id: 'cache', label: '缓存', icon: HardDrive },
  { id: 'about', label: '关于', icon: Info }
 ]
@@ -41,6 +55,12 @@ function SettingsPage() {
  const wxidDropdownRef = useRef<HTMLDivElement>(null)
  const [cachePath, setCachePath] = useState('')
  const [logEnabled, setLogEnabled] = useState(false)
+  const [whisperModelName, setWhisperModelName] = useState('base')
+  const [whisperModelDir, setWhisperModelDir] = useState('')
+  const [whisperDownloadSource, setWhisperDownloadSource] = useState('tsinghua')
+  const [isWhisperDownloading, setIsWhisperDownloading] = useState(false)
+  const [whisperDownloadProgress, setWhisperDownloadProgress] = useState(0)
+  const [whisperModelStatus, setWhisperModelStatus] = useState<{ exists: boolean; path?: string } | null>(null)

  const [isLoading, setIsLoadingState] = useState(false)
  const [isTesting, setIsTesting] = useState(false)
@@ -102,6 +122,9 @@ function SettingsPage() {
      const savedLogEnabled = await configService.getLogEnabled()
      const savedImageXorKey = await configService.getImageXorKey()
      const savedImageAesKey = await configService.getImageAesKey()
+      const savedWhisperModelName = await configService.getWhisperModelName()
+      const savedWhisperModelDir = await configService.getWhisperModelDir()
+      const savedWhisperSource = await configService.getWhisperDownloadSource()

      if (savedKey) setDecryptKey(savedKey)
      if (savedPath) setDbPath(savedPath)
@@ -112,6 +135,9 @@ function SettingsPage() {
      }
      if (savedImageAesKey) setImageAesKey(savedImageAesKey)
      setLogEnabled(savedLogEnabled)
+      if (savedWhisperModelName) setWhisperModelName(savedWhisperModelName)
+      if (savedWhisperModelDir) setWhisperModelDir(savedWhisperModelDir)
+      if (savedWhisperSource) setWhisperDownloadSource(savedWhisperSource)
    } catch (e) {
      console.error('加载配置失败:', e)
    }
@@ -119,6 +145,20 @@ function SettingsPage() {



+  const refreshWhisperStatus = async (modelNameValue = whisperModelName, modelDirValue = whisperModelDir) => {
+    try {
+      const result = await window.electronAPI.whisper?.getModelStatus({
+        modelName: modelNameValue,
+        downloadDir: modelDirValue || undefined
+      })
+      if (result?.success) {
+        setWhisperModelStatus({ exists: Boolean(result.exists), path: result.path })
+      }
+    } catch {
+      setWhisperModelStatus(null)
+    }
+  }
+
  const loadAppVersion = async () => {
    try {
      const version = await window.electronAPI.app.getVersion()
@@ -136,6 +176,20 @@ function SettingsPage() {
    return () => removeListener?.()
  }, [])

+  useEffect(() => {
+    const removeListener = window.electronAPI.whisper?.onDownloadProgress?.((payload) => {
+      if (payload.modelName !== whisperModelName) return
+      if (typeof payload.percent === 'number') {
+        setWhisperDownloadProgress(payload.percent)
+      }
+    })
+    return () => removeListener?.()
+  }, [whisperModelName])
+
+  useEffect(() => {
+    void refreshWhisperStatus(whisperModelName, whisperModelDir)
+  }, [whisperModelName, whisperModelDir])
+
  const handleCheckUpdate = async () => {
    setIsCheckingUpdate(true)
    setUpdateInfo(null)
@@ -143,9 +197,9 @@ function SettingsPage() {
      const result = await window.electronAPI.app.checkForUpdates()
      if (result.hasUpdate) {
        setUpdateInfo(result)
-        showMessage(`发现新版本 ${result.version}`, true)
+        showMessage(`发现新版：${result.version}`, true)
      } else {
-        showMessage('当前已是最新版本', true)
+        showMessage('当前已是最新版', true)
      }
    } catch (e) {
      showMessage(`检查更新失败: ${e}`, false)
@@ -257,6 +311,60 @@ function SettingsPage() {



+  const handleSelectWhisperModelDir = async () => {
+    try {
+      const result = await dialog.openFile({ title: '选择 Whisper 模型下载目录', properties: ['openDirectory'] })
+      if (!result.canceled && result.filePaths.length > 0) {
+        const dir = result.filePaths[0]
+        setWhisperModelDir(dir)
+        await configService.setWhisperModelDir(dir)
+        showMessage('已选择 Whisper 模型目录', true)
+      }
+    } catch (e) {
+      showMessage('选择目录失败', false)
+    }
+  }
+
+  const handleWhisperModelChange = async (value: string) => {
+    setWhisperModelName(value)
+    setWhisperDownloadProgress(0)
+    await configService.setWhisperModelName(value)
+  }
+
+  const handleWhisperSourceChange = async (value: string) => {
+    setWhisperDownloadSource(value)
+    await configService.setWhisperDownloadSource(value)
+  }
+
+  const handleDownloadWhisperModel = async () => {
+    if (isWhisperDownloading) return
+    setIsWhisperDownloading(true)
+    setWhisperDownloadProgress(0)
+    try {
+      const result = await window.electronAPI.whisper.downloadModel({
+        modelName: whisperModelName,
+        downloadDir: whisperModelDir || undefined,
+        source: whisperDownloadSource
+      })
+      if (result.success) {
+        setWhisperDownloadProgress(100)
+        showMessage('Whisper 模型下载完成', true)
+        await refreshWhisperStatus(whisperModelName, whisperModelDir)
+      } else {
+        showMessage(result.error || 'Whisper 模型下载失败', false)
+      }
+    } catch (e) {
+      showMessage(`Whisper 模型下载失败: ${e}`, false)
+    } finally {
+      setIsWhisperDownloading(false)
+    }
+  }
+
+  const handleResetWhisperModelDir = async () => {
+    setWhisperModelDir('')
+    await configService.setWhisperModelDir('')
+  }
+
  const handleAutoGetDbKey = async () => {
    if (isFetchingDbKey) return
    setIsFetchingDbKey(true)
@@ -367,6 +475,9 @@ function SettingsPage() {
      } else {
        await configService.setImageAesKey('')
      }
+      await configService.setWhisperModelName(whisperModelName)
+      await configService.setWhisperModelDir(whisperModelDir)
+      await configService.setWhisperDownloadSource(whisperDownloadSource)
      await configService.setOnboardingDone(true)

      showMessage('配置保存成功，正在测试连接...', true)
@@ -387,7 +498,7 @@ function SettingsPage() {
  }

  const handleClearConfig = async () => {
-    const confirmed = window.confirm('确定要清除当前配置吗？清除后需要重新完成首次配置。')
+    const confirmed = window.confirm('确定要清除当前配置吗？清除后需要重新完成首次配置？')
    if (!confirmed) return
    setIsLoadingState(true)
    setLoading(true, '正在清除配置...')
@@ -402,6 +513,12 @@ function SettingsPage() {
      setWxid('')
      setCachePath('')
      setLogEnabled(false)
+      setWhisperModelName('base')
+      setWhisperModelDir('')
+      setWhisperDownloadSource('tsinghua')
+      setWhisperModelStatus(null)
+      setWhisperDownloadProgress(0)
+      setIsWhisperDownloading(false)
      setDbConnected(false)
      await window.electronAPI.window.openOnboardingWindow()
    } catch (e) {
@@ -608,16 +725,6 @@ function SettingsPage() {
        {isFetchingImageKey && <div className="form-hint status-text">正在扫描内存，请稍候...</div>}
      </div>

-      <div className="form-group">
-        <label>缓存目录 <span className="optional">(可选)</span></label>
-        <span className="form-hint">留空使用默认目录</span>
-        <input type="text" placeholder="留空使用默认目录" value={cachePath} onChange={(e) => setCachePath(e.target.value)} />
-        <div className="btn-row">
-          <button className="btn btn-secondary" onClick={handleSelectCachePath}><FolderOpen size={16} /> 浏览选择</button>
-          <button className="btn btn-secondary" onClick={() => setCachePath('')}><RotateCcw size={16} /> 恢复默认</button>
-        </div>
-      </div>
-
      <div className="form-group">
        <label>调试日志</label>
        <span className="form-hint">开启后写入 WCDB 调试日志，便于排查连接问题</span>
@@ -650,12 +757,82 @@ function SettingsPage() {
      </div>
    </div>
  )
-
-
-
+  const renderWhisperTab = () => (
+    <div className="tab-content">
+      <p className="section-desc">语音解密后自动转写为文字</p>
+      <div className="form-group whisper-section">
+        <label>语音识别模型 (Whisper)</label>
+        <span className="form-hint">语音解密后自动转文字，模型越大越准确但下载更慢</span>
+        <div className="whisper-grid">
+          <div className="whisper-field">
+            <span className="field-label">模型</span>
+            <select
+              value={whisperModelName}
+              onChange={(e) => handleWhisperModelChange(e.target.value)}
+            >
+              {whisperModels.map((model) => (
+                <option key={model.value} value={model.value}>{model.label}</option>
+              ))}
+            </select>
+          </div>
+          <div className="whisper-field">
+            <span className="field-label">下载源</span>
+            <select
+              value={whisperDownloadSource}
+              onChange={(e) => handleWhisperSourceChange(e.target.value)}
+            >
+              {whisperSources.map((source) => (
+                <option key={source.value} value={source.value}>{source.label}</option>
+              ))}
+            </select>
+          </div>
+        </div>
+        <span className="form-hint">模型下载目录</span>
+        <input
+          type="text"
+          placeholder="留空使用默认目录"
+          value={whisperModelDir}
+          onChange={(e) => setWhisperModelDir(e.target.value)}
+          onBlur={() => configService.setWhisperModelDir(whisperModelDir)}
+        />
+        <div className="btn-row">
+          <button className="btn btn-secondary" onClick={handleSelectWhisperModelDir}><FolderOpen size={16} /> 选择目录</button>
+          <button className="btn btn-secondary" onClick={handleResetWhisperModelDir}><RotateCcw size={16} /> 默认目录</button>
+        </div>
+        <div className="whisper-status-line">
+          <span className={`status ${whisperModelStatus?.exists ? 'ok' : 'warn'}`}>
+            {whisperModelStatus?.exists ? '已下载' : '未下载'}
+          </span>
+          {whisperModelStatus?.path && <span className="path">{whisperModelStatus.path}</span>}
+        </div>
+        {isWhisperDownloading ? (
+          <div className="whisper-progress">
+            <div className="progress-bar">
+              <div className="progress-fill" style={{ width: `${whisperDownloadProgress}%` }} />
+            </div>
+            <span>{whisperDownloadProgress.toFixed(0)}%</span>
+          </div>
+        ) : (
+          <button className="btn btn-primary" onClick={handleDownloadWhisperModel}>
+            <Download size={16} /> 下载模型
+          </button>
+        )}
+      </div>
+    </div>
+  )
  const renderCacheTab = () => (
    <div className="tab-content">
      <p className="section-desc">管理应用缓存数据</p>
+      <div className="form-group">
+        <label>缓存目录 <span className="optional">(可选)</span></label>
+        <span className="form-hint">留空使用默认目录</span>
+        <input type="text" placeholder="留空使用默认目录" value={cachePath} onChange={(e) => setCachePath(e.target.value)} />
+        <div className="btn-row">
+          <button className="btn btn-secondary" onClick={handleSelectCachePath}><FolderOpen size={16} /> 浏览选择</button>
+          <button className="btn btn-secondary" onClick={() => setCachePath('')}><RotateCcw size={16} /> 恢复默认</button>
+        </div>
+      </div>
+
      <div className="btn-row">
        <button className="btn btn-secondary" onClick={handleClearAnalyticsCache} disabled={isClearingCache}>
          <Trash2 size={16} /> 清除分析缓存
@@ -664,8 +841,7 @@ function SettingsPage() {
          <Trash2 size={16} /> 清除图片缓存
        </button>
        <button className="btn btn-danger" onClick={handleClearAllCache} disabled={isClearingCache}>
-          <Trash2 size={16} /> 清除所有缓存
-        </button>
+          <Trash2 size={16} /> 清除所有缓存</button>
      </div>
      <div className="divider" />
      <p className="section-desc">清除当前配置并重新开始首次引导</p>
@@ -690,7 +866,7 @@ function SettingsPage() {
        <div className="about-update">
          {updateInfo?.hasUpdate ? (
            <>
-              <p className="update-hint">新版本 v{updateInfo.version} 可用</p>
+              <p className="update-hint">新版 v{updateInfo.version} 可用</p>
              {isDownloading ? (
                <div className="download-progress">
                  <div className="progress-bar">
@@ -747,7 +923,7 @@ function SettingsPage() {
                  onClick={() => handleSelectWxid(opt.wxid)}
                >
                  <span className="wxid-id">{opt.wxid}</span>
-                  <span className="wxid-date">最后修改: {new Date(opt.modifiedTime).toLocaleString()}</span>
+                  <span className="wxid-date">最后修改 {new Date(opt.modifiedTime).toLocaleString()}</span>
                </div>
              ))}
            </div>
@@ -782,6 +958,7 @@ function SettingsPage() {
      <div className="settings-body">
        {activeTab === 'appearance' && renderAppearanceTab()}
        {activeTab === 'database' && renderDatabaseTab()}
+        {activeTab === 'whisper' && renderWhisperTab()}
        {activeTab === 'cache' && renderCacheTab()}
        {activeTab === 'about' && renderAboutTab()}
      </div>
@@ -790,3 +967,5 @@ function SettingsPage() {
 }

 export default SettingsPage
+
+
--- a/src/services/config.ts
+++ b/src/services/config.ts
@@ -17,7 +17,10 @@ export const CONFIG_KEYS = {
  ONBOARDING_DONE: 'onboardingDone',
  LLM_MODEL_PATH: 'llmModelPath',
  IMAGE_XOR_KEY: 'imageXorKey',
-  IMAGE_AES_KEY: 'imageAesKey'
+  IMAGE_AES_KEY: 'imageAesKey',
+  WHISPER_MODEL_NAME: 'whisperModelName',
+  WHISPER_MODEL_DIR: 'whisperModelDir',
+  WHISPER_DOWNLOAD_SOURCE: 'whisperDownloadSource'
 } as const

 // 获取解密密钥
@@ -144,6 +147,39 @@ export async function setLlmModelPath(path: string): Promise<void> {
  await config.set(CONFIG_KEYS.LLM_MODEL_PATH, path)
 }

+// 获取 Whisper 模型名称
+export async function getWhisperModelName(): Promise<string | null> {
+  const value = await config.get(CONFIG_KEYS.WHISPER_MODEL_NAME)
+  return (value as string) || null
+}
+
+// 设置 Whisper 模型名称
+export async function setWhisperModelName(name: string): Promise<void> {
+  await config.set(CONFIG_KEYS.WHISPER_MODEL_NAME, name)
+}
+
+// 获取 Whisper 模型目录
+export async function getWhisperModelDir(): Promise<string | null> {
+  const value = await config.get(CONFIG_KEYS.WHISPER_MODEL_DIR)
+  return (value as string) || null
+}
+
+// 设置 Whisper 模型目录
+export async function setWhisperModelDir(dir: string): Promise<void> {
+  await config.set(CONFIG_KEYS.WHISPER_MODEL_DIR, dir)
+}
+
+// 获取 Whisper 下载源
+export async function getWhisperDownloadSource(): Promise<string | null> {
+  const value = await config.get(CONFIG_KEYS.WHISPER_DOWNLOAD_SOURCE)
+  return (value as string) || null
+}
+
+// 设置 Whisper 下载源
+export async function setWhisperDownloadSource(source: string): Promise<void> {
+  await config.set(CONFIG_KEYS.WHISPER_DOWNLOAD_SOURCE, source)
+}
+
 // 清除所有配置
 export async function clearConfig(): Promise<void> {
  await config.clear()
--- a/src/types/electron.d.ts
+++ b/src/types/electron.d.ts
@@ -95,6 +95,7 @@ export interface ElectronAPI {
    }>
    getImageData: (sessionId: string, msgId: string) => Promise<{ success: boolean; data?: string; error?: string }>
    getVoiceData: (sessionId: string, msgId: string) => Promise<{ success: boolean; data?: string; error?: string }>
+    getVoiceTranscript: (sessionId: string, msgId: string) => Promise<{ success: boolean; transcript?: string; error?: string }>
  }

  image: {
@@ -295,6 +296,11 @@ export interface ElectronAPI {
      error?: string
    }>
  }
+  whisper: {
+    downloadModel: (payload: { modelName: string; downloadDir?: string; source?: string }) => Promise<{ success: boolean; path?: string; error?: string }>
+    getModelStatus: (payload: { modelName: string; downloadDir?: string }) => Promise<{ success: boolean; exists?: boolean; path?: string; sizeBytes?: number; error?: string }>
+    onDownloadProgress: (callback: (payload: { modelName: string; downloadedBytes: number; totalBytes?: number; percent?: number }) => void) => () => void
+  }
 }

 export interface ExportOptions {
--- a/tsconfig.node.json
+++ b/tsconfig.node.json
@@ -13,6 +13,7 @@
  },
  "include": [
    "vite.config.ts",
-    "electron/**/*.ts"
+    "electron/**/*.ts",
+    "electron/**/*.d.ts"
  ]
 }
--- a/vite.config.ts
+++ b/vite.config.ts
@@ -30,7 +30,9 @@ export default defineConfig({
              external: [
                'better-sqlite3',
                'koffi',
-                'fsevents'
+                'fsevents',
+                'whisper-node',
+                'shelljs'
              ]
            }
          }