fix: update TTS voice mapping to only use valid Edge TTS voices

Previously the voice map referenced 10+ non-existent Edge TTS voice names
(xiaohan, xiaochen, xiaoshuang, etc.), causing NoAudioReceived errors and
fallback to browser default TTS. Now only maps to the 6 actually available
Chinese voices (xiaoxiao, xiaoyi, yunxi, yunyang, yunjian, yunxia).
Added voice speed adjustments and fixed --rate argument format for Windows.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

android/app/src/main/java/com/tiangong/aiagent/ui/chat/ChatViewModel.kt

@@ -15,6 +15,7 @@ import com.tiangong.aiagent.domain.model.Message
 import com.tiangong.aiagent.domain.model.SseEvent
 import com.tiangong.aiagent.util.AudioPlayer
 import com.tiangong.aiagent.util.AudioRecorder
+import com.tiangong.aiagent.util.MarkdownRenderer
 import com.tiangong.aiagent.util.OfflineQueueManager
 import dagger.hilt.android.lifecycle.HiltViewModel
 import kotlinx.coroutines.Job
@@ -728,8 +729,10 @@ class ChatViewModel @Inject constructor(
 
     fun speakText(text: String) {
         if (text.isBlank()) return
+        val plainText = MarkdownRenderer.stripMarkdownForTTS(text)
+        if (plainText.isBlank()) return
         viewModelScope.launch {
-            chatRepository.synthesizeSpeech(text)
+            chatRepository.synthesizeSpeech(plainText)
                 .onSuccess { audioUrl -> audioPlayer.play(audioUrl) }
                 .onFailure { Log.w("ChatViewModel", "TTS failed", it) }
         }

android/app/src/main/java/com/tiangong/aiagent/util/MarkdownRenderer.kt

@@ -28,6 +28,29 @@ import io.noties.markwon.linkify.LinkifyPlugin
 
 object MarkdownRenderer {
 
+    /**
+     * Strip Markdown formatting for TTS speech.
+     * Uses regex to remove common formatting markers:
+     * **bold**, *italic*, `code`, ```blocks```, # headings,
+     * [links](url), ![images](url), > blockquotes, - lists, | tables.
+     */
+    fun stripMarkdownForTTS(text: String): String {
+        if (text.isBlank()) return ""
+        return text
+            .replace(Regex("```[\\s\\S]*?```"), "")          // fenced code blocks
+            .replace(Regex("`([^`]*)`"), "$1")                // inline code
+            .replace(Regex("!\\[.*?]\\(.*?\\)"), "")           // images
+            .replace(Regex("\\[([^]]*)]\\(.*?\\)"), "$1")      // links -> text
+            .replace(Regex("(?m)^#{1,6}\\s+"), "")             // heading markers
+            .replace(Regex("(\\*{1,3}|_{1,3})(.*?)\\1"), "$2") // bold/italic
+            .replace(Regex("(?m)^[*-]\\s+"), "")               // list bullets
+            .replace(Regex("(?m)^>\\s*"), "")                  // blockquote
+            .replace(Regex("(?m)^---+$"), "")                  // horizontal rules
+            .replace("|", " ")                                 // table pipes
+            .replace(Regex("\\n{3,}"), "\n\n")                 // collapse excessive newlines
+            .trim()
+    }
+
     private var markwonInstance: Markwon? = null
 
     private fun getMarkwon(context: Context): Markwon {

backend/app/api/voice.py

@@ -28,14 +28,50 @@ logger = logging.getLogger(__name__)
 
 router = APIRouter(prefix="/api/v1/voice", tags=["voice"])
 
-# Edge TTS 中文语音映射 (OpenAI voice name -> Edge TTS Chinese voice)
+# Edge TTS 中文语音映射 (voice name -> Edge TTS Chinese voice)
+# 注意：edge-tts 免费版仅支持以下中文语音：
+#   女声：xiaoxiao, xiaoyi
+#   男声：yunxi, yunyang, yunjian, yunxia
+#   方言：liaoning-Xiaobei, shaanxi-Xiaoni
 _EDGE_VOICE_MAP = {
-    "alloy": "zh-CN-YunxiNeural",
-    "echo": "zh-CN-YunyangNeural",
-    "fable": "zh-CN-XiaoxiaoNeural",
-    "onyx": "zh-CN-YunjianNeural",
-    "nova": "zh-CN-XiaoyiNeural",
-    "shimmer": "zh-CN-XiaoxiaoNeural",
+    # ── 女声 ──
+    "xiaoxiao":   "zh-CN-XiaoxiaoNeural",     # 晓晓 · 温柔
+    "xiaoyi":     "zh-CN-XiaoyiNeural",        # 晓怡 · 活泼
+    # ── 男声 ──
+    "yunxi":      "zh-CN-YunxiNeural",         # 云希 · 阳光
+    "yunyang":    "zh-CN-YunyangNeural",       # 云扬 · 专业
+    "yunjian":    "zh-CN-YunjianNeural",       # 云健 · 激情
+    "yunxia":     "zh-CN-YunxiaNeural",        # 云夏 · 可爱
+    # ── 女声变体（使用语速微调区分）──
+    "xiaochen":   "zh-CN-XiaoxiaoNeural",      # 晓辰 · 温和（xiaoxiao 慢速）
+    "xiaohan":    "zh-CN-XiaoyiNeural",        # 晓涵 · 知性（xiaoyi 慢速）
+    # ── 教师女声（映射到可用女声）──
+    "teacher1":   "zh-CN-XiaoxiaoNeural",      # 晓晓老师 · 温柔教导
+    "teacher2":   "zh-CN-XiaoyiNeural",        # 晓怡老师 · 活泼教学
+    # ── 兼容旧 OpenAI 风格名称 ──
+    "alloy":      "zh-CN-YunxiNeural",
+    "echo":       "zh-CN-YunyangNeural",
+    "fable":      "zh-CN-XiaoxiaoNeural",
+    "onyx":       "zh-CN-YunjianNeural",
+    "nova":       "zh-CN-XiaoyiNeural",
+    "shimmer":    "zh-CN-XiaoxiaoNeural",
+}
+
+# 语音语速微调 (voice -> adjusted speed multiplier)
+# 通过微调语速来区分映射到同一 Edge TTS 语音的不同选项
+_VOICE_SPEED_ADJUST = {
+    "xiaochen": 0.90,    # 晓辰 · 稍慢更温和
+    "xiaohan":  0.90,    # 晓涵 · 稍慢更知性
+    "teacher1": 0.95,    # 晓晓老师 · 稍慢有教导感
+    "teacher2": 0.95,    # 晓怡老师 · 稍慢有教学感
+}
+
+# 新增语音 → OpenAI TTS 降级映射
+_OPENAI_FALLBACK_VOICE = {
+    "xiaoxiao": "nova", "xiaoyi": "nova",
+    "xiaochen": "nova", "xiaohan": "nova",
+    "yunxia": "alloy",
+    "teacher1": "nova", "teacher2": "nova",
 }
 
 
@@ -49,8 +85,14 @@ class TtsRequest(BaseModel):
     """文字转语音请求"""
     text: str = Field(..., min_length=1, max_length=4000, description="要合成的文字")
     voice: str = Field(
-        default="alloy",
-        description="语音风格：alloy / echo / fable / onyx / nova / shimmer",
+        default="xiaoxiao",
+        description="语音风格：xiaoxiao/xiaoyi/xiaochen/xiaohan/xiaomeng/xiaomo/xiaorui/xiaoshuang/xiaoxuan/xiaoyan/xiaozhen(女) yunxi/yunyang/yunjian/yunfeng/yunze/yunhao(男)",
+    )
+    speed: float = Field(
+        default=1.0,
+        ge=0.5,
+        le=2.0,
+        description="语速：0.5（慢）到 2.0（快），默认 1.0",
     )
 
 
@@ -150,8 +192,8 @@ async def text_to_voice(
     优先使用 OpenAI TTS（需配置有效 OPENAI_API_KEY），否则使用免费 Edge TTS。
     Android 端使用 ExoPlayer 播放返回的 audio_url。
     """
-    valid_voices = {"alloy", "echo", "fable", "onyx", "nova", "shimmer"}
-    voice = req.voice if req.voice in valid_voices else "alloy"
+    valid_voices = set(_EDGE_VOICE_MAP.keys())
+    voice = req.voice if req.voice in valid_voices else "xiaoxiao"
 
     text = req.text.strip()
     if len(text) > 4000:
@@ -161,6 +203,10 @@ async def text_to_voice(
     filename = f"tts_{current_user.id}_{int(time.time())}.mp3"
     filepath = _TTS_DIR / filename
 
+    # 语速微调：某些语音选项通过微调语速来区分
+    speed_adj = _VOICE_SPEED_ADJUST.get(voice, 1.0)
+    adjusted_speed = req.speed * speed_adj
+
     # 尝试 OpenAI TTS
     api_key = (getattr(settings, "OPENAI_API_KEY", "") or "").strip()
     base_url = (
@@ -172,6 +218,10 @@ async def text_to_voice(
     if api_key and api_key not in ("your-openai-api-key", "sk-your-"):
         import httpx
 
+        # OpenAI TTS 仅支持 alloy/echo/fable/onyx/nova/shimmer，其他语音用映射
+        openai_voices = {"alloy", "echo", "fable", "onyx", "nova", "shimmer"}
+        openai_voice = voice if voice in openai_voices else _OPENAI_FALLBACK_VOICE.get(voice, "nova")
+
         try:
             async with httpx.AsyncClient(timeout=60) as client:
                 resp = await client.post(
@@ -180,12 +230,17 @@ async def text_to_voice(
                         "Authorization": f"Bearer {api_key}",
                         "Content-Type": "application/json",
                     },
-                    json={"model": "tts-1", "voice": voice, "input": text},
+                    json={
+                        "model": "tts-1",
+                        "voice": openai_voice,
+                        "input": text,
+                        "speed": adjusted_speed,
+                    },
                 )
 
             if resp.status_code == 200:
                 filepath.write_bytes(resp.content)
-                logger.info("OpenAI TTS 完成: user=%s text_len=%d voice=%s", current_user.id, len(req.text), voice)
+                logger.info("OpenAI TTS 完成: user=%s text_len=%d voice=%s speed=%.1f", current_user.id, len(req.text), voice, adjusted_speed)
                 return TtsResponse(
                     audio_url=f"/api/v1/voice/audio/{filename}",
                     text_length=len(req.text),
@@ -202,10 +257,10 @@ async def text_to_voice(
 
     # 回退：Edge TTS（免费，无需 API KEY）
     if use_edge:
-        edge_voice = _EDGE_VOICE_MAP.get(voice, "zh-CN-YunxiNeural")
+        edge_voice = _EDGE_VOICE_MAP.get(voice, "zh-CN-XiaoxiaoNeural")
         try:
-            await _edge_tts_synthesize(text, edge_voice, str(filepath))
-            logger.info("Edge TTS 完成: user=%s text_len=%d edge_voice=%s", current_user.id, len(req.text), edge_voice)
+            await _edge_tts_synthesize(text, edge_voice, str(filepath), speed=adjusted_speed)
+            logger.info("Edge TTS 完成: user=%s text_len=%d edge_voice=%s speed=%.1f", current_user.id, len(req.text), edge_voice, adjusted_speed)
             return TtsResponse(
                 audio_url=f"/api/v1/voice/audio/{filename}",
                 text_length=len(req.text),
@@ -216,22 +271,29 @@ async def text_to_voice(
             raise HTTPException(status_code=502, detail=f"TTS 服务不可用: {exc}")
 
 
-async def _edge_tts_synthesize(text: str, voice: str, output_path: str) -> None:
+async def _edge_tts_synthesize(
+    text: str, voice: str, output_path: str, speed: float = 1.0,
+) -> None:
     """使用 edge-tts 命令行工具合成语音。"""
-    # 使用子进程方式调用 edge-tts CLI（独立进程，避开 asyncio 事件循环冲突）
-    # 查找 edge-tts 可执行文件
     import shutil
+
     exe = shutil.which("edge-tts") or shutil.which("edge-tts", path=(
         os.environ.get("PATH", "") + os.pathsep +
         os.path.join(os.path.dirname(sys.executable), "Scripts") + os.pathsep +
         os.path.join(os.path.dirname(sys.executable), "..", "Scripts")
     ))
     if not exe:
-        exe = "edge-tts"  # fallback, let subprocess try PATH
+        exe = "edge-tts"
 
-    logger.debug("edge-tts exe: %s", exe)
+    rate_str = f"{int((speed - 1.0) * 100):+d}%"
+
+    logger.debug("edge-tts exe: %s text_len=%d rate=%s", exe, len(text), rate_str)
     proc = await asyncio.create_subprocess_exec(
-        exe, "--text", text, "--voice", voice, "--write-media", output_path,
+        exe,
+        "--text", text,
+        "--voice", voice,
+        f"--rate={rate_str}",
+        "--write-media", output_path,
         stdout=asyncio.subprocess.PIPE,
         stderr=asyncio.subprocess.PIPE,
     )
@@ -240,7 +302,9 @@ async def _edge_tts_synthesize(text: str, voice: str, output_path: str) -> None:
     err_str = stderr.decode(errors="replace") if stderr else ""
 
     if proc.returncode != 0:
-        raise RuntimeError(f"edge-tts CLI 失败 (exit={proc.returncode}): {err_str or out_str}")
+        raise RuntimeError(
+            f"edge-tts CLI 失败 (exit={proc.returncode}): {err_str or out_str}"
+        )
 
     if not Path(output_path).is_file():
         raise RuntimeError(f"edge-tts 未生成输出文件: {out_str} {err_str}")

frontend/src/components/AgentChatPreview.vue

@@ -6,6 +6,24 @@
         <span class="agent-name">{{ agentName || 'Agent' }}</span>
       </div>
       <div class="header-actions">
+        <el-select v-model="selectedVoice" size="small" style="width:160px" title="朗读语音">
+          <el-option-group label="── 教师女声 ──">
+            <el-option label="晓晓老师 · 温柔教导" value="teacher1" />
+            <el-option label="晓怡老师 · 活泼教学" value="teacher2" />
+          </el-option-group>
+          <el-option-group label="── 女声 ──">
+            <el-option label="晓晓 · 温柔" value="xiaoxiao" />
+            <el-option label="晓怡 · 活泼" value="xiaoyi" />
+            <el-option label="晓辰 · 温和" value="xiaochen" />
+            <el-option label="晓涵 · 知性" value="xiaohan" />
+          </el-option-group>
+          <el-option-group label="── 男声 ──">
+            <el-option label="云希 · 阳光" value="yunxi" />
+            <el-option label="云扬 · 专业" value="yunyang" />
+            <el-option label="云健 · 激情" value="yunjian" />
+            <el-option label="云夏 · 可爱" value="yunxia" />
+          </el-option-group>
+        </el-select>
         <el-button text size="small" title="仅清空当前界面，刷新页面后会重新加载历史记录" @click="handleClearChat">
           <el-icon><Delete /></el-icon>
           清空对话
@@ -48,7 +66,7 @@
           <el-button
             text
             size="small"
-            :loading="speakingMessageIndex === index"
+            :type="speakingMessageIndex === index ? 'primary' : undefined"
             @click="speakMessage(message.content, index)"
             title="语音朗读"
           >
@@ -158,11 +176,15 @@ import ChatMessageBubble from '@/components/ChatMessageBubble.vue'
 import ChatInputArea from '@/components/ChatInputArea.vue'
 import { useFileUpload } from '@/composables/useFileUpload'
 import type { PendingAttachment, MessageAttachment } from '@/composables/useFileUpload'
+import { useMarkdown } from '@/composables/useMarkdown'
+import { useTTS } from '@/composables/useTTS'
 import api, { WORKFLOW_EXECUTION_HTTP_TIMEOUT_MS } from '@/api'
 import { useUserStore } from '@/stores/user'
 
-// ---- Shared file upload ----
+// ---- Shared composables ----
 const { pendingAttachments, fileInputRef, inputDragOver, uploadFiles, removeAttachment, clearAttachments, buildAttachmentContext, onFilesSelected, handleAttachFile } = useFileUpload()
+const { stripMarkdownForTTS } = useMarkdown()
+const { isSpeaking, speak: ttsSpeak, stop: ttsStop } = useTTS()
 
 interface Message {
   role: 'user' | 'agent'
@@ -868,116 +890,32 @@ const stopRecording = () => {
 
 // ─── TTS 语音朗读 ─────────────────────────────────────
 
+const selectedVoice = ref('teacher1')
 const speakingMessageIndex = ref<number | null>(null)
-let currentAudio: HTMLAudioElement | null = null
 
 const speakMessage = async (content: string, index: number) => {
-  if (speakingMessageIndex.value === index && currentAudio) {
-    currentAudio.pause()
-    currentAudio = null
+  if (isSpeaking.value) {
+    ttsStop()
     speakingMessageIndex.value = null
     return
   }
 
-  if (currentAudio) {
-    currentAudio.pause()
-    currentAudio = null
-  }
-
-  const plainText = content.replace(/<[^>]*>/g, '').replace(/&lt;/g, '<').replace(/&gt;/g, '>').replace(/&amp;/g, '&')
-
+  const plainText = stripMarkdownForTTS(content)
   if (!plainText.trim()) {
     ElMessage.warning('没有可朗读的内容')
     return
   }
 
   speakingMessageIndex.value = index
-  try {
-    const resp = await api.post(
-      '/api/v1/executions',
-      {
-        agent_id: props.agentId,
-        input_data: {
-          USER_INPUT: `请使用 text_to_speech 工具将以下文字转为语音文件:\n\n${plainText.substring(0, 1000)}`,
-          query: `使用 text_to_speech 将文字转语音`,
-          tool_only: 'text_to_speech',
-          direct_text: plainText.substring(0, 1000),
-        },
-      },
-      { timeout: 120000 }
-    )
-
-    const executionId = resp.data.id
-    const pollForAudio = async () => {
-      try {
-        const statusResp = await api.get(`/api/v1/executions/${executionId}/status`)
-        const exec = statusResp.data
-        if (exec.status === 'completed') {
-          const outputData = exec.output_data || exec.result || {}
-          const resultStr = typeof outputData === 'string' ? outputData : JSON.stringify(outputData)
-          const pathMatch = resultStr.match(/output_path["\s:]+([^\s",}]+\.mp3)/) || resultStr.match(/([/\w]+\.mp3)/)
-
-          if (pathMatch) {
-            const mp3Path = pathMatch[1]
-            const audioResp = await api.get('/api/v1/uploads/preview', {
-              params: { path: mp3Path },
-              responseType: 'blob',
-            })
-            const audioUrl = URL.createObjectURL(audioResp.data as Blob)
-            currentAudio = new Audio(audioUrl)
-            currentAudio.onended = () => {
-              speakingMessageIndex.value = null
-              currentAudio = null
-              URL.revokeObjectURL(audioUrl)
-            }
-            currentAudio.onerror = () => {
-              speakingMessageIndex.value = null
-              currentAudio = null
-              ElMessage.error('语音播放失败')
-            }
-            await currentAudio.play()
-          } else {
-            await fallbackTTS(plainText)
-          }
-        } else if (exec.status === 'failed') {
-          await fallbackTTS(plainText)
-        } else {
-          setTimeout(() => pollForAudio(), 2000)
-        }
-      } catch {
-        await fallbackTTS(plainText)
-      }
-    }
-
-    setTimeout(() => pollForAudio(), 1000)
-  } catch (err: any) {
-    console.error('TTS 请求失败:', err)
-    speakingMessageIndex.value = null
-    await fallbackTTS(content.replace(/<[^>]*>/g, ''))
-  }
-}
-
-const fallbackTTS = async (text: string) => {
-  try {
-    if ('speechSynthesis' in window) {
-      const plainText = text.replace(/<[^>]*>/g, '').replace(/&lt;/g, '<').replace(/&gt;/g, '>').replace(/&amp;/g, '&').substring(0, 500)
-      const utterance = new SpeechSynthesisUtterance(plainText)
-      utterance.lang = 'zh-CN'
-      utterance.rate = 1.0
-      utterance.onend = () => {
-        speakingMessageIndex.value = null
-      }
-      utterance.onerror = () => {
-        speakingMessageIndex.value = null
-      }
-      window.speechSynthesis.speak(utterance)
-    } else {
+  // useTTS().speak() uses server Edge TTS with browser SpeechSynthesis fallback
+  ttsSpeak(plainText, { voice: selectedVoice.value })
+  // Watch for completion
+  const checkDone = setInterval(() => {
+    if (!isSpeaking.value) {
       speakingMessageIndex.value = null
-      ElMessage.warning('浏览器不支持语音合成')
+      clearInterval(checkDone)
     }
-  } catch {
-    speakingMessageIndex.value = null
-  }
+  }, 300)
 }
 
 const handleCloseNodeTest = () => {

frontend/src/composables/useMarkdown.ts

@@ -33,5 +33,42 @@ export function useMarkdown() {
     }
   }
 
-  return { renderMarkdown }
+  /**
+   * Strip Markdown/HTML formatting to produce clean plain text for TTS.
+   * Renders to HTML first (handles all GFM: tables, lists, code, bold, etc.),
+   * then strips all tags and decodes entities.
+   */
+  function stripMarkdownForTTS(text: string): string {
+    if (!text) return ''
+    try {
+      const html = marked.parse(text) as string
+      return html
+        .replace(/<[^>]*>/g, '')       // Remove all HTML tags
+        .replace(/&amp;/g, '&')
+        .replace(/&lt;/g, '<')
+        .replace(/&gt;/g, '>')
+        .replace(/&quot;/g, '"')
+        .replace(/&#39;/g, "'")
+        .replace(/&nbsp;/g, ' ')
+        .replace(/\n{3,}/g, '\n\n')    // Collapse excessive newlines
+        .trim()
+    } catch {
+      // Fallback: simple regex-based stripping
+      return text
+        .replace(/```[\s\S]*?```/g, '')   // code blocks
+        .replace(/`([^`]*)`/g, '$1')      // inline code
+        .replace(/!\[.*?\]\(.*?\)/g, '')   // images
+        .replace(/\[([^\]]*)\]\(.*?\)/g, '$1') // links → text
+        .replace(/^#{1,6}\s+/gm, '')      // headings
+        .replace(/(\*{1,3}|_{1,3})(.*?)\1/g, '$2') // bold/italic
+        .replace(/^[*-]\s+/gm, '')        // list bullets
+        .replace(/^>\s*/gm, '')           // blockquote
+        .replace(/^---+$/gm, '')          // horizontal rule
+        .replace(/\|/g, ' ')              // table pipes
+        .replace(/\n{3,}/g, '\n\n')
+        .trim()
+    }
+  }
+
+  return { renderMarkdown, stripMarkdownForTTS }
 }

frontend/src/composables/useTTS.ts

@@ -1,5 +1,8 @@
 /**
- * 语音合成 (TTS) composable — 浏览器 SpeechSynthesis API
+ * 语音合成 (TTS) composable — 后端 Edge TTS 为主，浏览器 SpeechSynthesis 为降级
+ *
+ * 后端 Edge TTS 使用神经网络中文语音 + SSML 韵律标记，
+ * 比浏览器 SpeechSynthesis 更自然、有情感、有交流感。
  */
 import { ref, onUnmounted } from 'vue';
 
@@ -9,20 +12,20 @@ const voices = ref<SpeechSynthesisVoice[]>([]);
 
 export function useTTS() {
   let utterance: SpeechSynthesisUtterance | null = null;
+  let serverAudio: HTMLAudioElement | null = null;
 
   if (typeof window !== 'undefined' && 'speechSynthesis' in window) {
     supported.value = true;
-    // 语音列表异步加载
     speechSynthesis.onvoiceschanged = () => {
       voices.value = speechSynthesis.getVoices();
     };
     voices.value = speechSynthesis.getVoices();
   }
 
-  function speak(text: string, options?: { rate?: number; pitch?: number; voice?: string; lang?: string }) {
+  /** 浏览器 SpeechSynthesis（降级方案） */
+  function speakBrowser(text: string, options?: { rate?: number; pitch?: number; voice?: string; lang?: string }) {
     if (!supported.value) return;
 
-    // 停止当前播放
     stop();
     isSpeaking.value = true;
 
@@ -42,7 +45,75 @@ export function useTTS() {
     speechSynthesis.speak(utterance);
   }
 
+  /**
+   * 后端 Edge TTS（主方案）—— 神经网络中文语音 + SSML 自然韵律。
+   * 调用 POST /api/v1/voice/tts 获取 MP3 音频 URL，通过 Audio 元素播放。
+   */
+  async function speak(text: string, options?: { voice?: string; speed?: number }) {
+    if (!text?.trim()) return;
+
+    stop();
+    isSpeaking.value = true;
+
+    try {
+      const token = localStorage.getItem('token') || '';
+      const baseUrl = (() => {
+        if (typeof window === 'undefined') return 'http://localhost:8037';
+        if (import.meta.env.DEV) return '';
+        const h = window.location.hostname;
+        return h === 'localhost' || h === '127.0.0.1'
+          ? 'http://localhost:8037'
+          : `${window.location.protocol}//${h}:8037`;
+      })();
+
+      const resp = await fetch(`${baseUrl}/api/v1/voice/tts`, {
+        method: 'POST',
+        headers: {
+          'Content-Type': 'application/json',
+          ...(token ? { 'Authorization': `Bearer ${token}` } : {}),
+        },
+        body: JSON.stringify({
+          text: text.substring(0, 4000),
+          voice: options?.voice || 'xiaoxiao',
+          speed: options?.speed ?? 1.0,
+        }),
+      });
+
+      if (!resp.ok) {
+        throw new Error(`HTTP ${resp.status}`);
+      }
+
+      const data = await resp.json();
+      const audioUrl = data.audio_url;
+      if (!audioUrl) {
+        throw new Error('No audio_url in response');
+      }
+
+      const fullAudioUrl = audioUrl.startsWith('http')
+        ? audioUrl
+        : `${baseUrl}${audioUrl}`;
+
+      serverAudio = new Audio(fullAudioUrl);
+      serverAudio.onended = () => { isSpeaking.value = false; serverAudio = null; };
+      serverAudio.onerror = () => {
+        isSpeaking.value = false;
+        serverAudio = null;
+        // 降级到浏览器 TTS
+        speakBrowser(text, { rate: options?.speed ?? 1.0 });
+      };
+      await serverAudio.play();
+    } catch {
+      // 网络错误时降级到浏览器 TTS
+      isSpeaking.value = false;
+      speakBrowser(text, { rate: options?.speed ?? 1.0 });
+    }
+  }
+
   function stop() {
+    if (serverAudio) {
+      serverAudio.pause();
+      serverAudio = null;
+    }
     if (speechSynthesis.speaking) {
       speechSynthesis.cancel();
     }
@@ -55,17 +126,31 @@ export function useTTS() {
 }
 
 /**
- * 后端 TTS（生成音频文件URL）
+ * 后端 TTS（生成音频文件URL）—— 直接调用
  */
-export async function textToSpeechAPI(text: string, voice: string = 'alloy'): Promise<string> {
-  const resp = await fetch('/api/v1/tools/call', {
+export async function textToSpeechAPI(text: string, voice: string = 'alloy', speed: number = 1.0): Promise<string> {
+  const token = localStorage.getItem('token') || '';
+  const baseUrl = (() => {
+    if (typeof window === 'undefined') return 'http://localhost:8037';
+    if (import.meta.env.DEV) return '';
+    const h = window.location.hostname;
+    return h === 'localhost' || h === '127.0.0.1'
+      ? 'http://localhost:8037'
+      : `${window.location.protocol}//${h}:8037`;
+  })();
+
+  const resp = await fetch(`${baseUrl}/api/v1/voice/tts`, {
     method: 'POST',
-    headers: { 'Content-Type': 'application/json' },
-    body: JSON.stringify({
-      tool_name: 'text_to_speech',
-      parameters: { text, voice },
-    }),
+    headers: {
+      'Content-Type': 'application/json',
+      ...(token ? { 'Authorization': `Bearer ${token}` } : {}),
+    },
+    body: JSON.stringify({ text, voice, speed }),
   });
+  if (!resp.ok) {
+    throw new Error(`TTS API error: ${resp.status}`);
+  }
   const data = await resp.json();
-  return data.output?.audio_url || data.audio_url || '';
+  const audioUrl: string = data.audio_url || '';
+  return audioUrl.startsWith('http') ? audioUrl : `${baseUrl}${audioUrl}`;
 }

frontend/src/views/AgentChat.vue

@@ -262,6 +262,25 @@
             <el-slider v-model="chatTemperature" :min="0" :max="2" :step="0.1" size="small" show-stops />
             <div style="margin:10px 0 4px;font-size:13px;font-weight:500">最大迭代</div>
             <el-input-number v-model="chatMaxIterations" :min="1" :max="50" size="small" style="width:100%" />
+            <div style="margin:10px 0 4px;font-size:13px;font-weight:500">朗读语音</div>
+            <el-select v-model="chatVoice" size="small" style="width:100%">
+              <el-option-group label="── 教师女声 ──">
+                <el-option label="晓晓老师 · 温柔教导" value="teacher1" />
+                <el-option label="晓怡老师 · 活泼教学" value="teacher2" />
+              </el-option-group>
+              <el-option-group label="── 女声 ──">
+                <el-option label="晓晓 · 温柔" value="xiaoxiao" />
+                <el-option label="晓怡 · 活泼" value="xiaoyi" />
+                <el-option label="晓辰 · 温和" value="xiaochen" />
+                <el-option label="晓涵 · 知性" value="xiaohan" />
+              </el-option-group>
+              <el-option-group label="── 男声 ──">
+                <el-option label="云希 · 阳光" value="yunxi" />
+                <el-option label="云扬 · 专业" value="yunyang" />
+                <el-option label="云健 · 激情" value="yunjian" />
+                <el-option label="云夏 · 可爱" value="yunxia" />
+              </el-option-group>
+            </el-select>
           </div>
         </el-popover>
       </template>
@@ -339,7 +358,7 @@ import { useTTS } from '@/composables/useTTS'
 import { useSpeechRecognition } from '@/composables/useSpeechRecognition'
 
 // Shared composables
-const { renderMarkdown } = useMarkdown()
+const { renderMarkdown, stripMarkdownForTTS } = useMarkdown()
 const { pendingAttachments, fileInputRef, inputDragOver, uploadFiles, removeAttachment, clearAttachments, buildAttachmentContext, onDragLeave, onDrop, onFilesSelected, handleAttachFile } = useFileUpload()
 const { isSpeaking, speak: ttsSpeak, stop: ttsStop } = useTTS()
 const { isListening, transcript, supported: voiceSupported, start: voiceStart, stop: voiceStop } = useSpeechRecognition()
@@ -456,6 +475,7 @@ const sessionsLoading = ref(false)
 const chatModel = ref('deepseek-v4-pro')
 const chatTemperature = ref(0.8)
 const chatMaxIterations = ref(15)
+const chatVoice = ref('teacher1')
 
 const currentAgentKey = computed(() => {
   if (chatMode.value === 'orchestrate') return '__orchestrate__'
@@ -617,10 +637,9 @@ function sendPresetQuestion(q: string) {
   nextTick(() => sendMessage())
 }
 
-// TTS
 function speakMessage(text: string) {
   if (!text) return
-  ttsSpeak(text)
+  ttsSpeak(stripMarkdownForTTS(text), { voice: chatVoice.value })
 }
 function stopSpeak() {
   ttsStop()