feat: voice support zh-CN

---
STT（语音识别中文）

src/hooks/useVoice.ts 两处改成：
const rawLanguage = getInitialSettings().voiceLanguage || getInitialSettings().language
优先用 voiceLanguage 字段，没有则 fallback language。zh (中文) 已在 Deepgram 支持列表中。

TTS（中文语音输出）

src/hooks/useAutoTTS.ts 新增 resolveEdgeTTSVoice() 函数，根据语言自动选择 voice：
- zh / zh-CN → zh-CN-XiaoxiaoNeural
- zh-TW → zh-TW-HsiaoYuNeural
- zh-HK → zh-HK-HiuMaanNeural
- ja → ja-JP-NanamiNeural
- 其他 → en-US-JennyNeural（默认）

配置方法（在 ~/.claude/settings.json 里加）：
{
voiceLanguage: zh-CN,
voiceTTSVoice: zh-CN-XiaoxiaoNeural
}

或者更简单，改 language 为 zh-CN 也能自动生效（两处都会读）。

README.md

@@ -91,7 +91,12 @@ cp VersperClaw ~/.local/bin
 
   # set auto-dream config
   "autoMemoryEnabled": true,
-  "autoDreamEnabled": true
+  "autoDreamEnabled": true,
+
+  # stt-tts support zh-CN
+  "voiceLanguage": "zh-CN",
+  "voiceTTSVoice": "zh-CN-XiaoxiaoNeural"
+
 }
 ```
 

src/hooks/useAutoTTS.ts

@@ -3,6 +3,27 @@ import { speakWithEdgeTTS, playAudioFile } from '../services/voice/edgeTTS.js'
 import { getInitialSettings } from '../utils/settings/settings.js'
 import type { RenderableMessage } from '../types/message.js'
 
+// Map language codes to Edge TTS Chinese voices
+const CHINESE_VOICES: Record<string, string> = {
+  zh: 'zh-CN-XiaoxiaoNeural',
+  'zh-CN': 'zh-CN-XiaoxiaoNeural',
+  'zh-TW': 'zh-TW-HsiaoYuNeural',
+  'zh-HK': 'zh-HK-HiuMaanNeural',
+}
+
+function resolveEdgeTTSVoice(language: string | undefined, explicitVoice: string | undefined): string {
+  if (explicitVoice) return explicitVoice
+  if (!language) return 'en-US-JennyNeural'
+  const base = language.split('-')[0].toLowerCase()
+  if (base === 'zh') return CHINESE_VOICES[language] || CHINESE_VOICES['zh-CN']
+  if (base === 'ja') return 'ja-JP-NanamiNeural'
+  if (base === 'ko') return 'ko-KR-SunHiNeural'
+  if (base === 'fr') return 'fr-FR-DeniseNeural'
+  if (base === 'de') return 'de-DE-KatjaNeural'
+  if (base === 'es') return 'es-ES-ElviraNeural'
+  return 'en-US-JennyNeural'
+}
+
 export function useAutoTTS(messages: RenderableMessage[]): void {
   const triggeredIdsRef = useRef<Set<string>>(new Set())
   const pendingPlayRef = useRef<Promise<void> | null>(null)
@@ -22,7 +43,8 @@ export function useAutoTTS(messages: RenderableMessage[]): void {
       triggeredIdsRef.current.add(msg.uuid)
 
       const text = content.text
-      const voice = settings.voiceTTSVoice || 'en-US-JennyNeural'
+      const language = settings.voiceLanguage || settings.language
+      const voice = resolveEdgeTTSVoice(language, settings.voiceTTSVoice)
 
       const run = async () => {
         const result = await speakWithEdgeTTS(text, {

src/hooks/useVoice.ts

@@ -421,7 +421,8 @@ export function useVoice({
         const replayBuffer = fullAudioRef.current
         await sleep(250)
         if (isStale()) return
-        const stt = normalizeLanguageForSTT(getInitialSettings().language)
+        const rawLanguage = getInitialSettings().voiceLanguage || getInitialSettings().language
+        const stt = normalizeLanguageForSTT(rawLanguage)
         const keyterms = await getVoiceKeyterms()
         if (isStale()) return
         await new Promise<void>(resolve => {
@@ -764,7 +765,7 @@ export function useVoice({
       return
     }
 
-    const rawLanguage = getInitialSettings().language
+    const rawLanguage = getInitialSettings().voiceLanguage || getInitialSettings().language
     const stt = normalizeLanguageForSTT(rawLanguage)
     logEvent('tengu_voice_recording_started', {
       focusTriggered: focusTriggeredRef.current,