feat: feishu voice voxcpm and edge-tts

src/commands/feishu/feishu.tsx

@@ -25,6 +25,8 @@ type Step =
   | { type: 'edit-allowed-chats' }
   | { type: 'edit-mention-policy' }
   | { type: 'edit-tts-settings' }
+  | { type: 'edit-tts-provider' }
+  | { type: 'edit-tts-ref-audio' }
   | { type: 'confirm-clear' }
   | { type: 'scanning' }
 
@@ -199,10 +201,18 @@ function FeishuDialog({
         label: `TTS 语音回复: ${config?.ttsEnabled ? '开' : '关'}`,
         action: () => { void toggleTts() },
       },
+      {
+        label: `TTS 引擎: ${config?.ttsProvider === 'voxcpm' ? 'VoxCPM' : 'Edge TTS'}`,
+        action: () => { void toggleTtsProvider() },
+      },
       {
         label: `TTS 语音: ${config?.ttsVoice || 'zh-CN-XiaoxiaoNeural (中文)'}`,
         action: () => setStep({ type: 'edit-tts-settings' }),
       },
+      {
+        label: `TTS 参考音频: ${config?.ttsReferenceAudio ? '已配置' : '未配置'}`,
+        action: () => setStep({ type: 'edit-tts-ref-audio' }),
+      },
       {
         label:
           serviceState.status === 'running' ? '停止当前飞书 Bot' : '启动当前飞书 Bot',
@@ -498,6 +508,52 @@ function FeishuDialog({
     [refreshConfig],
   )
 
+  const toggleTtsProvider = React.useCallback(
+    async () => {
+      setBusy(true)
+      try {
+        const current = getFeishuConfig()
+        const next = current.ttsProvider === 'edge' ? 'voxcpm' : 'edge'
+        saveFeishuConfig({ ...current, ttsProvider: next })
+        refreshConfig()
+        setNotice({
+          text: `TTS 引擎已切换为 ${next === 'edge' ? 'Edge TTS' : 'VoxCPM'}。`,
+          tone: 'success',
+        })
+      } catch (error) {
+        setNotice({
+          text: error instanceof Error ? error.message : '保存失败',
+          tone: 'error',
+        })
+      } finally {
+        setBusy(false)
+        setStep({ type: 'menu' })
+      }
+    },
+    [refreshConfig],
+  )
+
+  const saveTtsRefAudio = React.useCallback(
+    async (value: string) => {
+      setBusy(true)
+      try {
+        const current = getFeishuConfig()
+        saveFeishuConfig({ ...current, ttsReferenceAudio: value || undefined })
+        refreshConfig()
+        setNotice({ text: '参考音频路径已保存。', tone: 'success' })
+      } catch (error) {
+        setNotice({
+          text: error instanceof Error ? error.message : '保存失败',
+          tone: 'error',
+        })
+      } finally {
+        setBusy(false)
+        setStep({ type: 'menu' })
+      }
+    },
+    [refreshConfig],
+  )
+
   const clearConfigAndStop = React.useCallback(async () => {
     setBusy(true)
     try {
@@ -695,6 +751,20 @@ function FeishuDialog({
     )
   }
 
+  if (step.type === 'edit-tts-ref-audio') {
+    return (
+      <Dialog title="VoxCPM 参考音频" onCancel={() => setStep({ type: 'menu' })}>
+        <TextInput
+          title="参考音频路径"
+          hint="VoxCPM 语音克隆的参考音频文件路径（支持 WAV/MP3）。留空清空配置。"
+          initialValue={config?.ttsReferenceAudio}
+          onSubmit={value => { void saveTtsRefAudio(value) }}
+          onCancel={() => setStep({ type: 'menu' })}
+        />
+      </Dialog>
+    )
+  }
+
   if (step.type === 'confirm-clear') {
     return (
       <Dialog title="清空飞书配置" onCancel={() => setStep({ type: 'menu' })}>

src/services/feishu/FeishuService.ts

@@ -517,43 +517,69 @@ class FeishuService {
     if (!this.channel) return
     try {
       const config = getFeishuConfig()
-      console.log('[feishu][tts] config.ttsEnabled:', config.ttsEnabled, 'voice:', config.ttsVoice)
       if (!config.ttsEnabled) return
 
       const { readFile, unlink } = await import('node:fs/promises')
       const { spawn } = await import('node:child_process')
 
-      const voice = config.ttsVoice || 'zh-CN-XiaoxiaoNeural'
-      const mp3Path = path.join(os.tmpdir(), `feishu_tts_${Date.now()}.mp3`)
-      const oggPath = mp3Path.replace(/\.mp3$/, '.ogg')
+      const provider = config.ttsProvider || 'edge'
+      const timestamp = Date.now()
+      const rawBase = path.join(os.tmpdir(), `feishu_tts_raw_${timestamp}`)
+      const oggPath = path.join(os.tmpdir(), `feishu_tts_${timestamp}.ogg`)
 
-      console.log('[feishu][tts] generating MP3 with voice:', voice, 'text len:', text.length)
+      if (provider === 'voxcpm') {
+        const refAudio = config.ttsReferenceAudio
+        if (!refAudio) {
+          console.log('[feishu][tts] voxcpm: no ttsReferenceAudio configured')
+          return
+        }
+        console.log('[feishu][tts] voxcpm: synthesizing with ref:', refAudio, 'output base:', rawBase)
 
-      // Generate MP3 via edge-tts CLI
-      await new Promise<void>((resolve, reject) => {
-        const child = spawn('edge-tts', [
-          '--voice', voice,
-          '--text', text,
-          '--write-media', mp3Path,
-        ])
-        let stderr = ''
-        child.stderr?.on('data', c => { stderr += String(c) })
-        child.on('close', code => {
-          console.log('[feishu][tts] edge-tts exit code:', code, 'stderr:', stderr.slice(0, 100))
-          if (code === 0) resolve()
-          else reject(new Error(`edge-tts exit ${code}: ${stderr.slice(0, 200)}`))
+        await new Promise<void>((resolve, reject) => {
+          const child = spawn(
+            'voxcpm', ['clone', '--text', text, '--reference-audio', refAudio, '--denoise', '--output', `${rawBase}.wav`],
+            { shell: false },
+          )
+          let stdout = ''
+          let stderr = ''
+          child.stdout?.on('data', c => { stdout += String(c) })
+          child.stderr?.on('data', c => { stderr += String(c) })
+          child.on('close', code => {
+            console.log('[feishu][tts] voxcpm exit code:', code)
+            if (code === 0) resolve()
+            else reject(new Error(`voxcpm exit ${code}: ${stderr.slice(0, 200)}`))
+          })
+          child.on('error', reject)
         })
-        child.on('error', err => {
-          console.log('[feishu][tts] edge-tts spawn error:', err.message)
-          reject(err)
+      } else {
+        const voice = config.ttsVoice || 'zh-CN-XiaoxiaoNeural'
+        console.log('[feishu][tts] edge-tts: voice:', voice, 'text len:', text.length)
+
+        await new Promise<void>((resolve, reject) => {
+          const child = spawn('edge-tts', [
+            '--voice', voice,
+            '--text', text,
+            '--write-media', rawBase,
+          ])
+          let stderr = ''
+          child.stderr?.on('data', c => { stderr += String(c) })
+          child.on('close', code => {
+            console.log('[feishu][tts] edge-tts exit code:', code)
+            if (code === 0) resolve()
+            else reject(new Error(`edge-tts exit ${code}: ${stderr.slice(0, 200)}`))
+          })
+          child.on('error', reject)
         })
-      })
+      }
+
+      const rawPath = provider === 'voxcpm' ? `${rawBase}.wav` : rawBase
+      console.log('[feishu][tts] raw audio at:', rawPath)
 
-      // Convert MP3 → OGG/Opus (Feishu voice requires opus format)
-      console.log('[feishu][tts] converting MP3 → OGG')
+      // Convert raw → OGG/Opus (Feishu voice requires opus format)
+      console.log('[feishu][tts] converting to OGG/Opus')
       await new Promise<void>((resolve, reject) => {
         const child = spawn('ffmpeg', [
-          '-i', mp3Path,
+          '-i', rawPath,
           '-c:a', 'libopus',
           '-b:a', '128k',
           '-y',
@@ -573,7 +599,7 @@ class FeishuService {
       console.log('[feishu][tts] sent successfully')
 
       // Clean up temp files
-      await unlink(mp3Path).catch(() => {})
+      await unlink(rawPath).catch(() => {})
       await unlink(oggPath).catch(() => {})
     } catch (err) {
       // Don't let TTS failure break the main message flow

src/services/feishu/feishuConfig.ts

@@ -41,8 +41,12 @@ export type FeishuRuntimeConfig = {
   streamingCard?: boolean
   /** TTS 语音回复开关 — 发送消息时同时发送语音 */
   ttsEnabled?: boolean
+  /** TTS 引擎: 'edge' | 'voxcpm' */
+  ttsProvider?: 'edge' | 'voxcpm'
   /** TTS 语音名称（edge-tts 格式，默认 zh-CN-XiaoxiaoNeural） */
   ttsVoice?: string
+  /** VoxCPM 参考音频路径（用于 voxtts 模式） */
+  ttsReferenceAudio?: string
 }
 
 function getConfigPath(): string {