fix: 唤醒机制

main/xiaozhi-server/config.yaml

@@ -59,6 +59,10 @@ log:
 delete_audio: true
 # 没有语音输入多久后断开连接(秒)，默认2分钟，即120秒
 close_connection_no_voice_time: 120
+# TTS请求超时时间(秒)
+tts_timeout: 10
+# 开启唤醒词加速
+enable_wakeup_words_response_cache: true
 # 开场是否回复唤醒词
 enable_greeting: true
 # 说完话是否开启提示音

main/xiaozhi-server/core/handle/helloHandle.py

@@ -1,5 +1,13 @@
+import time
 import json
+import random
 import asyncio
+from core.utils.dialogue import Message
+from core.utils.util import audio_to_data_stream
+from core.providers.tts.dto.dto import SentenceType
+from core.utils.wakeup_word import WakeupWordsConfig
+from core.handle.sendAudioHandle import sendAudioMessage, send_stt_message
+from core.utils.util import remove_punctuation_and_length, opus_datas_to_wav_bytes
 from core.providers.tools.device_mcp import (
     MCPClient,
     send_mcp_initialize_message,
@@ -8,6 +16,18 @@ from core.providers.tools.device_mcp import (
 
 TAG = __name__
 
+WAKEUP_CONFIG = {
+    "refresh_time": 5,
+    "words": ["你好", "你好啊", "嘿，你好", "嗨"],
+}
+
+# 创建全局的唤醒词配置管理器
+wakeup_words_config = WakeupWordsConfig()
+
+# 用于防止并发调用wakeupWordsResponse的锁
+_wakeup_response_lock = asyncio.Lock()
+
+
 async def handleHelloMessage(conn, msg_json):
     """处理hello消息"""
     audio_params = msg_json.get("audio_params")
@@ -28,4 +48,100 @@ async def handleHelloMessage(conn, msg_json):
             # 发送mcp消息，获取tools列表
             asyncio.create_task(send_mcp_tools_list_request(conn))
 
-    await conn.websocket.send(json.dumps(conn.welcome_msg))
+    await conn.websocket.send(json.dumps(conn.welcome_msg))
+
+
+async def checkWakeupWords(conn, text):
+    enable_wakeup_words_response_cache = conn.config[
+        "enable_wakeup_words_response_cache"
+    ]
+
+    if not enable_wakeup_words_response_cache or not conn.tts:
+        return False
+
+    _, filtered_text = remove_punctuation_and_length(text)
+    if filtered_text not in conn.config.get("wakeup_words"):
+        return False
+
+    conn.just_woken_up = True
+    await send_stt_message(conn, text)
+
+    # 获取当前音色
+    voice = getattr(conn.tts, "voice", "default")
+    if not voice:
+        voice = "default"
+
+    # 获取唤醒词回复配置
+    response = wakeup_words_config.get_wakeup_response(voice)
+    if not response or not response.get("file_path"):
+        response = {
+            "voice": "default",
+            "file_path": "config/assets/wakeup_words.wav",
+            "time": 0,
+            "text": "哈啰啊，我是小智啦，声音好听的台湾女孩一枚，超开心认识你耶，最近在忙啥，别忘了给我来点有趣的料哦，我超爱听八卦的啦",
+        }
+
+    # 获取音频数据
+    opus_packets = []
+    def handle_audio_frame(frame_data):
+        opus_packets.append(frame_data)
+
+    audio_to_data_stream(response.get("file_path"), is_opus=True, callback=handle_audio_frame)
+
+    # 播放唤醒词回复
+    conn.client_abort = False
+
+    conn.logger.bind(tag=TAG).info(f"播放唤醒词回复: {response.get('text')}")
+    await sendAudioMessage(conn, SentenceType.FIRST, opus_packets, response.get("text"))
+    await sendAudioMessage(conn, SentenceType.LAST, [], None)
+
+    # 补充对话
+    conn.dialogue.put(Message(role="assistant", content=response.get("text")))
+
+    # 检查是否需要更新唤醒词回复
+    if time.time() - response.get("time", 0) > WAKEUP_CONFIG["refresh_time"]:
+        if not _wakeup_response_lock.locked():
+            asyncio.create_task(wakeupWordsResponse(conn))
+    return True
+
+
+async def wakeupWordsResponse(conn):
+    if not conn.tts or not conn.llm or not conn.llm.response_no_stream:
+        return
+
+    try:
+        # 尝试获取锁，如果获取不到就返回
+        if not await _wakeup_response_lock.acquire():
+            return
+
+        # 生成唤醒词回复
+        wakeup_word = random.choice(WAKEUP_CONFIG["words"])
+        question = (
+            "此刻用户正在和你说```"
+            + wakeup_word
+            + "```。\n请你根据以上用户的内容进行20-30字回复。要符合系统设置的角色情感和态度，不要像机器人一样说话。\n"
+            + "请勿对这条内容本身进行任何解释和回应，请勿返回表情符号，仅返回对用户的内容的回复。"
+        )
+
+        result = conn.llm.response_no_stream(conn.config["prompt"], question)
+        if not result or len(result) == 0:
+            return
+
+        # 生成TTS音频
+        tts_result = await asyncio.to_thread(conn.tts.to_tts, result)
+        if not tts_result:
+            return
+
+        # 获取当前音色
+        voice = getattr(conn.tts, "voice", "default")
+
+        wav_bytes = opus_datas_to_wav_bytes(tts_result, sample_rate=16000)
+        file_path = wakeup_words_config.generate_file_path(voice)
+        with open(file_path, "wb") as f:
+            f.write(wav_bytes)
+        # 更新配置
+        wakeup_words_config.update_wakeup_response(voice, file_path, result)
+    finally:
+        # 确保在任何情况下都释放锁
+        if _wakeup_response_lock.locked():
+            _wakeup_response_lock.release()

main/xiaozhi-server/core/handle/intentHandler.py

@@ -3,6 +3,7 @@ import uuid
 import asyncio
 from core.utils.dialogue import Message
 from core.providers.tts.dto.dto import ContentType
+from core.handle.helloHandle import checkWakeupWords
 from plugins_func.register import Action, ActionResponse
 from core.handle.sendAudioHandle import send_stt_message
 from core.utils.util import remove_punctuation_and_length
@@ -27,6 +28,10 @@ async def handle_user_intent(conn, text):
     if await check_direct_exit(conn, filtered_text):
         return True
 
+    # 检查是否是唤醒词
+    if await checkWakeupWords(conn, filtered_text):
+        return True
+
     if conn.intent_type == "function_call":
         # 使用支持function calling的聊天方法,不再进行意图分析
         return False

main/xiaozhi-server/core/handle/receiveAudioHandle.py

@@ -1,5 +1,6 @@
 import time
 import json
+import asyncio
 from core.handle.abortHandle import handleAbortMessage
 from core.handle.intentHandler import handle_user_intent
 from core.utils.output_counter import check_device_output_limit
@@ -10,12 +11,16 @@ TAG = __name__
 
 
 async def handleAudioMessage(conn, audio):
-    # 检查是否在唤醒处理锁定期内
-    if getattr(conn, 'wakeup_processing_lock', 0) > time.monotonic():
-        return
-    
     # 当前片段是否有人说话
     have_voice = conn.vad.is_vad(conn, audio)
+    # 如果设备刚刚被唤醒，短暂忽略VAD检测
+    if have_voice and hasattr(conn, "just_woken_up") and conn.just_woken_up:
+        have_voice = False
+        # 设置一个短暂延迟后恢复VAD检测
+        conn.asr_audio.clear()
+        if not hasattr(conn, "vad_resume_task") or conn.vad_resume_task.done():
+            conn.vad_resume_task = asyncio.create_task(resume_vad_detection(conn))
+        return
     if have_voice:
         if conn.client_is_speaking:
             await handleAbortMessage(conn)
@@ -24,6 +29,11 @@ async def handleAudioMessage(conn, audio):
     # 接收音频
     await conn.asr.receive_audio(conn, audio, have_voice)
 
+async def resume_vad_detection(conn):
+    # 等待2秒后恢复VAD检测
+    await asyncio.sleep(1)
+    conn.just_woken_up = False
+
 async def startToChat(conn, text):
     # 检查输入是否是JSON格式（包含说话人信息）
     speaker_name = None

main/xiaozhi-server/core/handle/sendAudioHandle.py

@@ -30,13 +30,14 @@ async def sendAudioMessage(conn, sentenceType, audios, text):
 
 
 # 播放音频
-async def sendAudio(conn, audios, pre_buffer=False):
+async def sendAudio(conn, audios, pre_buffer=False, frame_duration=60):
     """
     发送单个opus包，支持流控
     Args:
         conn: 连接对象
         opus_packet: 单个opus数据包
         pre_buffer: 快速发送音频
+        frame_duration: 帧时长（毫秒），匹配 Opus 编码
     """
     if audios is None or len(audios) == 0:
         return
@@ -60,7 +61,6 @@ async def sendAudio(conn, audios, pre_buffer=False):
                 "start_time": time.perf_counter(),
             }
 
-        frame_duration=60
         flow_control = conn.audio_flow_control
         current_time = time.perf_counter()
         # 计算预期发送时间
@@ -77,6 +77,35 @@ async def sendAudio(conn, audios, pre_buffer=False):
         # 更新流控状态
         flow_control["packet_count"] += 1
         flow_control["last_send_time"] = time.perf_counter()
+    else:
+        # 流控参数优化
+        start_time = time.perf_counter()
+        play_position = 0
+
+        # 执行预缓冲
+        pre_buffer_frames = min(3, len(audios))
+        for i in range(pre_buffer_frames):
+            await conn.websocket.send(audios[i])
+        remaining_audios = audios[pre_buffer_frames:]
+
+        # 播放剩余音频帧
+        for opus_packet in remaining_audios:
+            if conn.client_abort:
+                break
+
+            # 重置没有声音的状态
+            conn.last_activity_time = time.time() * 1000
+
+            # 计算预期发送时间
+            expected_time = start_time + (play_position / 1000)
+            current_time = time.perf_counter()
+            delay = expected_time - current_time
+            if delay > 0:
+                await asyncio.sleep(delay)
+
+            await conn.websocket.send(opus_packet)
+
+            play_position += frame_duration
 
 
 async def send_tts_message(conn, state, text=None):

main/xiaozhi-server/core/handle/textHandle.py

@@ -3,8 +3,9 @@ import asyncio
 from core.utils.util import filter_sensitive_info
 from core.handle.abortHandle import handleAbortMessage
 from core.handle.helloHandle import handleHelloMessage
-from core.handle.receiveAudioHandle import handleAudioMessage
+from core.handle.reportHandle import enqueue_asr_report
 from core.providers.tools.device_mcp import handle_mcp_message
+from core.handle.receiveAudioHandle import startToChat, handleAudioMessage
 from core.handle.sendAudioHandle import send_stt_message, send_tts_message
 from core.providers.tools.device_iot import handleIotDescriptors, handleIotStatus
 
@@ -55,10 +56,16 @@ async def handleTextMessage(conn, message):
                         await send_stt_message(conn, original_text)
                         await send_tts_message(conn, "stop", None)
                         conn.client_is_speaking = False
+                    elif is_wakeup_words and enable_greeting:
+                        conn.just_woken_up = True
+                        # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
+                        enqueue_asr_report(conn, "嘿，你好呀", [])
+                        await startToChat(conn, "嘿，你好呀")
                     else:
-                        # 检测到唤醒词，开始等待后续进行声纹识别
-                        conn.wakeup_mode = True
-                        conn.logger.bind(tag=TAG).info(f"检测到唤醒词~")
+                        # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
+                        enqueue_asr_report(conn, original_text, [])
+                        # 否则需要LLM对文字内容进行答复
+                        await startToChat(conn, original_text)
         elif msg_json["type"] == "iot":
             conn.logger.bind(tag=TAG).info(f"收到iot消息：{message}")
             if "descriptors" in msg_json:

main/xiaozhi-server/core/providers/asr/base.py

@@ -53,10 +53,6 @@ class ASRProviderBase(ABC):
 
     # 接收音频
     async def receive_audio(self, conn, audio, audio_have_voice):
-        # 检查是否在唤醒处理锁定期内
-        if getattr(conn, 'wakeup_processing_lock', 0) > time.monotonic():
-            return
-            
         if conn.client_listen_mode == "auto" or conn.client_listen_mode == "realtime":
             have_voice = audio_have_voice
         else:
@@ -66,14 +62,6 @@ class ASRProviderBase(ABC):
         if not have_voice and not conn.client_have_voice:
             conn.asr_audio = conn.asr_audio[-10:]
             return
-        
-        # 检查是否处于唤醒模式
-        if getattr(conn, 'wakeup_mode', False) and len(conn.asr_audio) >= 10:
-            asr_audio_task = conn.asr_audio.copy()
-            conn.reset_vad_states()
-            conn.asr_audio.clear()
-                
-            await self.handle_voice_stop(conn, asr_audio_task)
 
         if conn.client_voice_stop:
             asr_audio_task = conn.asr_audio.copy()
@@ -102,10 +90,27 @@ class ASRProviderBase(ABC):
             if conn.voiceprint_provider and combined_pcm_data:
                 wav_data = self._pcm_to_wav(combined_pcm_data)
             
-            # 检查是否处于唤醒模式
-            wakeup_mode = getattr(conn, 'wakeup_mode', False)
+            # 定义ASR任务
+            def run_asr():
+                start_time = time.monotonic()
+                try:
+                    loop = asyncio.new_event_loop()
+                    asyncio.set_event_loop(loop)
+                    try:
+                        result = loop.run_until_complete(
+                            self.speech_to_text(asr_audio_task, conn.session_id, conn.audio_format)
+                        )
+                        end_time = time.monotonic()
+                        logger.bind(tag=TAG).info(f"ASR耗时: {end_time - start_time:.3f}s")
+                        return result
+                    finally:
+                        loop.close()
+                except Exception as e:
+                    end_time = time.monotonic()
+                    logger.bind(tag=TAG).error(f"ASR失败: {e}")
+                    return ("", None)
             
-            # 声纹识别任务（公共逻辑）
+            # 定义声纹识别任务
             def run_voiceprint():
                 if not wav_data:
                     return None
@@ -124,83 +129,48 @@ class ASRProviderBase(ABC):
                     logger.bind(tag=TAG).error(f"声纹识别失败: {e}")
                     return None
             
-            if wakeup_mode and conn.voiceprint_provider and wav_data:
-                conn.wakeup_mode = False
-                # 设置处理锁，防止后续音频片段重复处理
-                conn.wakeup_processing_lock = time.monotonic() + 3  # 3秒锁定期
+            # 使用线程池执行器并行运行
+            with concurrent.futures.ThreadPoolExecutor(max_workers=2) as thread_executor:
+                asr_future = thread_executor.submit(run_asr)
                 
-                # 唤醒模式：只执行声纹识别
-                with concurrent.futures.ThreadPoolExecutor(max_workers=1) as thread_executor:
+                if conn.voiceprint_provider and wav_data:
                     voiceprint_future = thread_executor.submit(run_voiceprint)
+                    
+                    # 等待两个线程都完成
+                    asr_result = asr_future.result(timeout=15)
                     voiceprint_result = voiceprint_future.result(timeout=15)
-
-                speaker_name = voiceprint_result
+                    
+                    results = {"asr": asr_result, "voiceprint": voiceprint_result}
+                else:
+                    asr_result = asr_future.result(timeout=15)
+                    results = {"asr": asr_result, "voiceprint": None}
+            
+            
+            # 处理结果
+            raw_text, _ = results.get("asr", ("", None))
+            speaker_name = results.get("voiceprint", None)
+            
+            # 记录识别结果
+            if raw_text:
+                logger.bind(tag=TAG).info(f"识别文本: {raw_text}")
+            if speaker_name:
                 logger.bind(tag=TAG).info(f"识别说话人: {speaker_name}")
-
-                fixed_text = "嘿，你好啊"
-                enhanced_text = self._build_enhanced_text(fixed_text, speaker_name)
+            
+            # 性能监控
+            total_time = time.monotonic() - total_start_time
+            logger.bind(tag=TAG).info(f"总处理耗时: {total_time:.3f}s")
+            
+            # 检查文本长度
+            text_len, _ = remove_punctuation_and_length(raw_text)
+            self.stop_ws_connection()
+            
+            if text_len > 0:
+                # 构建包含说话人信息的JSON字符串
+                enhanced_text = self._build_enhanced_text(raw_text, speaker_name)
                 
-                # 性能监控
-                total_time = time.monotonic() - total_start_time
-                logger.bind(tag=TAG).info(f"唤醒模式总处理耗时: {total_time:.3f}s")
-
+                # 使用自定义模块进行上报
                 await startToChat(conn, enhanced_text)
                 enqueue_asr_report(conn, enhanced_text, asr_audio_task)
-            else:
-                # 正常模式：执行声纹识别和ASR
-                def run_asr():
-                    start_time = time.monotonic()
-                    try:
-                        loop = asyncio.new_event_loop()
-                        asyncio.set_event_loop(loop)
-                        try:
-                            result = loop.run_until_complete(
-                                self.speech_to_text(asr_audio_task, conn.session_id, conn.audio_format)
-                            )
-                            end_time = time.monotonic()
-                            logger.bind(tag=TAG).info(f"ASR耗时: {end_time - start_time:.3f}s")
-                            return result
-                        finally:
-                            loop.close()
-                    except Exception as e:
-                        end_time = time.monotonic()
-                        logger.bind(tag=TAG).error(f"ASR失败: {e}")
-                        return ("", None)
-                
-                # 使用线程池执行器并行运行
-                with concurrent.futures.ThreadPoolExecutor(max_workers=2) as thread_executor:
-                    asr_future = thread_executor.submit(run_asr)
-                    
-                    if conn.voiceprint_provider and wav_data:
-                        voiceprint_future = thread_executor.submit(run_voiceprint)
-                        asr_result = asr_future.result(timeout=15)
-                        voiceprint_result = voiceprint_future.result(timeout=15)
-                        results = {"asr": asr_result, "voiceprint": voiceprint_result}
-                    else:
-                        asr_result = asr_future.result(timeout=15)
-                        results = {"asr": asr_result, "voiceprint": None}
-                
-                # 处理结果
-                raw_text, _ = results.get("asr", ("", None))
-                speaker_name = results.get("voiceprint", None)
-                
-                if raw_text:
-                    logger.bind(tag=TAG).info(f"识别文本: {raw_text}")
-                if speaker_name:
-                    logger.bind(tag=TAG).info(f"识别说话人: {speaker_name}")
-                
-                # 性能监控
-                total_time = time.monotonic() - total_start_time
-                logger.bind(tag=TAG).info(f"总处理耗时: {total_time:.3f}s")
-                
-                # 检查文本长度
-                text_len, _ = remove_punctuation_and_length(raw_text)
-                self.stop_ws_connection()
-                
-                if text_len > 0:
-                    enhanced_text = self._build_enhanced_text(raw_text, speaker_name)
-                    await startToChat(conn, enhanced_text)
-                    enqueue_asr_report(conn, enhanced_text, asr_audio_task)
                 
         except Exception as e:
             logger.bind(tag=TAG).error(f"处理语音停止失败: {e}")

main/xiaozhi-server/core/providers/tts/aliyun_stream.py

@@ -478,3 +478,142 @@ class TTSProvider(TTSProviderBase):
         finally:
             self._monitor_task = None
 
+    def to_tts(self, text: str) -> list:
+        """非流式TTS处理，用于测试及保存音频文件的场景"""
+        try:
+            # 创建新的事件循环
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+
+            # 生成会话ID
+            session_id = uuid.uuid4().hex
+            # 存储音频数据
+            audio_data = []
+
+            async def _generate_audio():
+                # 刷新Token（如果需要）
+                if self._is_token_expired():
+                    self._refresh_token()
+
+                # 建立WebSocket连接
+                ws = await websockets.connect(
+                    self.ws_url,
+                    additional_headers={"X-NLS-Token": self.token},
+                    ping_interval=30,
+                    ping_timeout=10,
+                    close_timeout=10,
+                )
+                try:
+                    # 发送StartSynthesis请求
+                    start_message_id = str(uuid.uuid4().hex)
+                    start_request = {
+                        "header": {
+                            "message_id": start_message_id,
+                            "task_id": session_id,
+                            "namespace": "FlowingSpeechSynthesizer",
+                            "name": "StartSynthesis",
+                            "appkey": self.appkey,
+                        },
+                        "payload": {
+                            "voice": self.voice,
+                            "format": self.format,
+                            "sample_rate": self.sample_rate,
+                            "volume": self.volume,
+                            "speech_rate": self.speech_rate,
+                            "pitch_rate": self.pitch_rate,
+                            "enable_subtitle": True,
+                        },
+                    }
+                    await ws.send(json.dumps(start_request))
+
+                    # 等待SynthesisStarted响应
+                    synthesis_started = False
+                    while not synthesis_started:
+                        msg = await ws.recv()
+                        if isinstance(msg, str):
+                            data = json.loads(msg)
+                            header = data.get("header", {})
+                            if header.get("name") == "SynthesisStarted":
+                                synthesis_started = True
+                                logger.bind(tag=TAG).debug("TTS合成已启动")
+                            elif header.get("name") == "TaskFailed":
+                                error_info = data.get("payload", {}).get(
+                                    "error_info", {}
+                                )
+                                error_code = error_info.get("error_code")
+                                error_message = error_info.get(
+                                    "error_message", "未知错误"
+                                )
+                                raise Exception(
+                                    f"启动合成失败: {error_code} - {error_message}"
+                                )
+
+                    # 发送文本合成请求
+                    filtered_text = MarkdownCleaner.clean_markdown(text)
+                    run_message_id = str(uuid.uuid4().hex)
+                    run_request = {
+                        "header": {
+                            "message_id": run_message_id,
+                            "task_id": session_id,
+                            "namespace": "FlowingSpeechSynthesizer",
+                            "name": "RunSynthesis",
+                            "appkey": self.appkey,
+                        },
+                        "payload": {"text": filtered_text},
+                    }
+                    await ws.send(json.dumps(run_request))
+
+                    # 发送停止合成请求
+                    stop_message_id = str(uuid.uuid4().hex)
+                    stop_request = {
+                        "header": {
+                            "message_id": stop_message_id,
+                            "task_id": session_id,
+                            "namespace": "FlowingSpeechSynthesizer",
+                            "name": "StopSynthesis",
+                            "appkey": self.appkey,
+                        }
+                    }
+                    await ws.send(json.dumps(stop_request))
+
+                    # 接收音频数据
+                    synthesis_completed = False
+                    while not synthesis_completed:
+                        msg = await ws.recv()
+                        if isinstance(msg, (bytes, bytearray)):
+                            self.opus_encoder.encode_pcm_to_opus_stream(
+                                msg,
+                                end_of_stream=False,
+                                callback=lambda opus: audio_data.append(opus)
+                            )
+                        elif isinstance(msg, str):
+                            data = json.loads(msg)
+                            header = data.get("header", {})
+                            event_name = header.get("name")
+                            if event_name == "SynthesisCompleted":
+                                synthesis_completed = True
+                                logger.bind(tag=TAG).debug("TTS合成完成")
+                            elif event_name == "TaskFailed":
+                                error_info = data.get("payload", {}).get(
+                                    "error_info", {}
+                                )
+                                error_code = error_info.get("error_code")
+                                error_message = error_info.get(
+                                    "error_message", "未知错误"
+                                )
+                                raise Exception(
+                                    f"合成失败: {error_code} - {error_message}"
+                                )
+                finally:
+                    try:
+                        await ws.close()
+                    except:
+                        pass
+
+            loop.run_until_complete(_generate_audio())
+            loop.close()
+
+            return audio_data
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"生成音频数据失败: {str(e)}")
+            return []

main/xiaozhi-server/core/providers/tts/base.py

@@ -1,21 +1,22 @@
 import os
 import re
-import queue
+import time
 import uuid
+import queue
 import asyncio
 import threading
-from typing import Callable, Any
+import traceback
 from core.utils import p3
-import time
 from datetime import datetime
 from core.utils import textUtils
+from typing import Callable, Any
 from abc import ABC, abstractmethod
 from config.logger import setup_logging
-from core.utils.util import audio_bytes_to_data_stream, audio_to_data_stream
 from core.utils.tts import MarkdownCleaner
 from core.utils.output_counter import add_device_output
 from core.handle.reportHandle import enqueue_tts_report
 from core.handle.sendAudioHandle import sendAudioMessage
+from core.utils.util import audio_bytes_to_data_stream, audio_to_data_stream
 from core.providers.tts.dto.dto import (
     TTSMessageDTO,
     SentenceType,
@@ -23,8 +24,6 @@ from core.providers.tts.dto.dto import (
     InterfaceType,
 )
 
-import traceback
-
 TAG = __name__
 logger = setup_logging()
 
@@ -144,6 +143,68 @@ class TTSProviderBase(ABC):
             except Exception as e:
                 logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
                 return None
+    
+    def to_tts(self, text):
+        text = MarkdownCleaner.clean_markdown(text)
+        max_repeat_time = 5
+        if self.delete_audio_file:
+            # 需要删除文件的直接转为音频数据
+            while max_repeat_time > 0:
+                try:
+                    audio_bytes = asyncio.run(self.text_to_speak(text, None))
+                    if audio_bytes:
+                        audio_datas = []
+                        audio_bytes_to_data_stream(
+                            audio_bytes,
+                            file_type=self.audio_file_type,
+                            is_opus=True,
+                            callback=lambda data: audio_datas.append(data)
+                        )
+                        return audio_datas
+                    else:
+                        max_repeat_time -= 1
+                except Exception as e:
+                    logger.bind(tag=TAG).warning(
+                        f"语音生成失败{5 - max_repeat_time + 1}次: {text}，错误: {e}"
+                    )
+                    max_repeat_time -= 1
+            if max_repeat_time > 0:
+                logger.bind(tag=TAG).info(
+                    f"语音生成成功: {text}，重试{5 - max_repeat_time}次"
+                )
+            else:
+                logger.bind(tag=TAG).error(
+                    f"语音生成失败: {text}，请检查网络或服务是否正常"
+                )
+            return None
+        else:
+            tmp_file = self.generate_filename()
+            try:
+                while not os.path.exists(tmp_file) and max_repeat_time > 0:
+                    try:
+                        asyncio.run(self.text_to_speak(text, tmp_file))
+                    except Exception as e:
+                        logger.bind(tag=TAG).warning(
+                            f"语音生成失败{5 - max_repeat_time + 1}次: {text}，错误: {e}"
+                        )
+                        # 未执行成功，删除文件
+                        if os.path.exists(tmp_file):
+                            os.remove(tmp_file)
+                        max_repeat_time -= 1
+
+                if max_repeat_time > 0:
+                    logger.bind(tag=TAG).info(
+                        f"语音生成成功: {text}:{tmp_file}，重试{5 - max_repeat_time}次"
+                    )
+                else:
+                    logger.bind(tag=TAG).error(
+                        f"语音生成失败: {text}，请检查网络或服务是否正常"
+                    )
+
+                return tmp_file
+            except Exception as e:
+                logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
+                return None
 
     @abstractmethod
     async def text_to_speak(self, text, output_file):

main/xiaozhi-server/core/providers/tts/huoshan_double_stream.py

@@ -628,3 +628,104 @@ class TTSProvider(TTSProviderBase):
 
     def wav_to_opus_data_audio_raw_stream(self, raw_data_var, is_end=False, callback: Callable[[Any], Any]=None):
         return self.opus_encoder.encode_pcm_to_opus_stream(raw_data_var, is_end, callback=callback)
+
+    def to_tts(self, text: str) -> list:
+        """非流式生成音频数据，用于生成音频及测试场景
+        Args:
+            text: 要转换的文本
+        Returns:
+            list: 音频数据列表
+        """
+        try:
+            # 创建事件循环
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+
+            # 生成会话ID
+            session_id = uuid.uuid4().__str__().replace("-", "")
+
+            # 存储音频数据
+            audio_data = []
+
+            async def _generate_audio():
+                # 创建新的WebSocket连接
+                ws_header = {
+                    "X-Api-App-Key": self.appId,
+                    "X-Api-Access-Key": self.access_token,
+                    "X-Api-Resource-Id": self.resource_id,
+                    "X-Api-Connect-Id": uuid.uuid4(),
+                }
+                ws = await websockets.connect(
+                    self.ws_url, additional_headers=ws_header, max_size=1000000000
+                )
+
+                try:
+                    # 启动会话
+                    header = Header(
+                        message_type=FULL_CLIENT_REQUEST,
+                        message_type_specific_flags=MsgTypeFlagWithEvent,
+                        serial_method=JSON,
+                    ).as_bytes()
+                    optional = Optional(
+                        event=EVENT_StartSession, sessionId=session_id
+                    ).as_bytes()
+                    payload = self.get_payload_bytes(
+                        event=EVENT_StartSession, speaker=self.voice
+                    )
+                    await self.send_event(ws, header, optional, payload)
+
+                    # 发送文本
+                    header = Header(
+                        message_type=FULL_CLIENT_REQUEST,
+                        message_type_specific_flags=MsgTypeFlagWithEvent,
+                        serial_method=JSON,
+                    ).as_bytes()
+                    optional = Optional(
+                        event=EVENT_TaskRequest, sessionId=session_id
+                    ).as_bytes()
+                    payload = self.get_payload_bytes(
+                        event=EVENT_TaskRequest, text=text, speaker=self.voice
+                    )
+                    await self.send_event(ws, header, optional, payload)
+
+                    # 发送结束会话请求
+                    header = Header(
+                        message_type=FULL_CLIENT_REQUEST,
+                        message_type_specific_flags=MsgTypeFlagWithEvent,
+                        serial_method=JSON,
+                    ).as_bytes()
+                    optional = Optional(
+                        event=EVENT_FinishSession, sessionId=session_id
+                    ).as_bytes()
+                    payload = str.encode("{}")
+                    await self.send_event(ws, header, optional, payload)
+
+                    # 接收音频数据
+                    while True:
+                        msg = await ws.recv()
+                        res = self.parser_response(msg)
+
+                        if (
+                            res.optional.event == EVENT_TTSResponse
+                            and res.header.message_type == AUDIO_ONLY_RESPONSE
+                        ):
+                            self.wav_to_opus_data_audio_raw_stream(res.payload, callback=lambda opus_frame: audio_data.append(opus_frame))
+                        elif res.optional.event == EVENT_SessionFinished:
+                            break
+
+                finally:
+                    # 清理资源
+                    try:
+                        await ws.close()
+                    except:
+                        pass
+
+            # 运行异步任务
+            loop.run_until_complete(_generate_audio())
+            loop.close()
+
+            return audio_data
+
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"生成音频数据失败: {str(e)}")
+            return []

main/xiaozhi-server/core/providers/tts/index_stream.py

@@ -1,8 +1,10 @@
 import os
+import time
 import queue
-import asyncio
-import traceback
 import aiohttp
+import asyncio
+import requests
+import traceback
 from config.logger import setup_logging
 from core.utils.tts import MarkdownCleaner
 from core.providers.tts.base import TTSProviderBase
@@ -177,3 +179,57 @@ class TTSProvider(TTSProviderBase):
         await super().close()
         if hasattr(self, "opus_encoder"):
             self.opus_encoder.close()
+
+    def to_tts(self, text: str) -> list:
+        """非流式TTS处理，用于测试及保存音频文件的场景
+        Args:
+            text: 要转换的文本
+        Returns:
+            list: 返回opus编码后的音频数据列表
+        """
+        start_time = time.time()
+        text = MarkdownCleaner.clean_markdown(text)
+
+        payload = {"text": text, "character": self.character}
+
+        try:
+            with requests.post(self.api_url, json=payload, timeout=5) as response:
+                if response.status_code != 200:
+                    logger.bind(tag=TAG).error(
+                        f"TTS请求失败: {response.status_code}, {response.text}"
+                    )
+                    return []
+
+                logger.info(f"TTS请求成功: {text}, 耗时: {time.time() - start_time}秒")
+
+                # 使用opus编码器处理PCM数据
+                opus_datas = []
+                pcm_data = response.content
+
+                # 计算每帧的字节数
+                frame_bytes = int(
+                    self.opus_encoder.sample_rate
+                    * self.opus_encoder.channels
+                    * self.opus_encoder.frame_size_ms
+                    / 1000
+                    * 2
+                )
+
+                # 分帧处理PCM数据
+                for i in range(0, len(pcm_data), frame_bytes):
+                    frame = pcm_data[i : i + frame_bytes]
+                    if len(frame) < frame_bytes:
+                        # 最后一帧可能不足，用0填充
+                        frame = frame + b"\x00" * (frame_bytes - len(frame))
+
+                    self.opus_encoder.encode_pcm_to_opus_stream(
+                        frame,
+                        end_of_stream=(i + frame_bytes >= len(pcm_data)),
+                        callback=lambda opus: opus_datas.append(opus)
+                    )
+
+                return opus_datas
+
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"TTS请求异常: {e}")
+            return []

main/xiaozhi-server/core/providers/tts/linkerai.py

@@ -1,7 +1,9 @@
 import os
+import time
 import queue
 import aiohttp
 import asyncio
+import requests
 import traceback
 from config.logger import setup_logging
 from core.utils.tts import MarkdownCleaner
@@ -195,3 +197,71 @@ class TTSProvider(TTSProviderBase):
         except Exception as e:
             logger.bind(tag=TAG).error(f"TTS请求异常: {e}")
             self.tts_audio_queue.put((SentenceType.LAST, [], None))
+
+    def to_tts(self, text: str) -> list:
+        """非流式TTS处理，用于测试及保存音频文件的场景
+        Args:
+            text: 要转换的文本
+        Returns:
+            list: 返回opus编码后的音频数据列表
+        """
+        start_time = time.time()
+        text = MarkdownCleaner.clean_markdown(text)
+
+        params = {
+            "tts_text": text,
+            "spk_id": self.voice,
+            "frame_duration": 60,
+            "stream": False,
+            "target_sr": 16000,
+            "audio_format": self.audio_format,
+            "instruct_text": "请生成一段自然流畅的语音",
+        }
+        headers = {
+            "Authorization": f"Bearer {self.access_token}",
+            "Content-Type": "application/json",
+        }
+
+        try:
+            with requests.get(
+                self.api_url, params=params, headers=headers, timeout=5
+            ) as response:
+                if response.status_code != 200:
+                    logger.bind(tag=TAG).error(
+                        f"TTS请求失败: {response.status_code}, {response.text}"
+                    )
+                    return []
+
+                logger.info(f"TTS请求成功: {text}, 耗时: {time.time() - start_time}秒")
+
+                # 使用opus编码器处理PCM数据
+                opus_datas = []
+                pcm_data = response.content
+
+                # 计算每帧的字节数
+                frame_bytes = int(
+                    self.opus_encoder.sample_rate
+                    * self.opus_encoder.channels
+                    * self.opus_encoder.frame_size_ms
+                    / 1000
+                    * 2
+                )
+
+                # 分帧处理PCM数据
+                for i in range(0, len(pcm_data), frame_bytes):
+                    frame = pcm_data[i : i + frame_bytes]
+                    if len(frame) < frame_bytes:
+                        # 最后一帧可能不足，用0填充
+                        frame = frame + b"\x00" * (frame_bytes - len(frame))
+
+                    self.opus_encoder.encode_pcm_to_opus_stream(
+                        frame,
+                        end_of_stream=(i + frame_bytes >= len(pcm_data)),
+                        callback=lambda opus: opus_datas.append(opus)
+                    )
+
+                return opus_datas
+
+        except Exception as e:
+            logger.bind(tag=TAG).error(f"TTS请求异常: {e}")
+            return []

main/xiaozhi-server/core/providers/tts/paddle_speech.py

@@ -1,14 +1,15 @@
-import asyncio
-import json
-import base64
-import aiohttp
-import numpy as np
 import io
 import wave
+import json
+import base64
+import asyncio
 import websockets
-from core.providers.tts.base import TTSProviderBase
-from config.logger import setup_logging
+import numpy as np
 from datetime import datetime
+from config.logger import setup_logging
+from core.providers.tts.base import TTSProviderBase
+
+
 
 TAG = __name__
 logger = setup_logging()
@@ -74,43 +75,9 @@ class TTSProvider(TTSProviderBase):
     async def text_to_speak(self, text, output_file):
         if self.protocol == "websocket":
             return await self.text_streaming(text, output_file)
-        elif self.protocol == "http":
-            return await self.text(text, output_file)
         else:
             raise ValueError("Unsupported protocol. Please use 'websocket' or 'http'.")
 
-    async def text(self, text, output_file):
-        request_json = {
-            "text": text,
-            "spk_id": self.spk_id,
-            "speed": self.speed,
-            "volume": self.volume,
-            "sample_rate": self.sample_rate,
-            "save_path": self.save_path
-        }
-
-        try:
-            async with aiohttp.ClientSession() as session:
-                async with session.post(self.url, json=request_json) as resp:
-                    if resp.status == 200:
-                        resp_json = await resp.json()
-                        if resp_json.get("success"):
-                            data = resp_json["result"]
-                            audio_bytes = base64.b64decode(data["audio"])
-                            if output_file:
-                                with open(output_file, "wb") as file_to_save:
-                                    file_to_save.write(audio_bytes)
-                            else:
-                                return audio_bytes
-                        else:
-                            raise Exception(
-                                f"Error: {resp_json.get('message', 'Unknown error')} while processing text: {text}")
-                    else:
-                        raise Exception(
-                            f"HTTP Error: {resp.status} - {await resp.text()} while processing text: {text}")
-        except Exception as e:
-            raise Exception(f"Error during TTS HTTP request: {e} while processing text: {text}")
-
     async def text_streaming(self, text, output_file):
         try:
             # 使用 websockets 异步连接到 WebSocket 服务器

main/xiaozhi-server/core/utils/util.py

@@ -2,6 +2,7 @@ import re
 import os
 import json
 import copy
+import wave
 import socket
 import requests
 import subprocess
@@ -292,6 +293,33 @@ def play_audio_frames(conn, file_path):
         callback=handle_audio_frame
     )
 
+
+def opus_datas_to_wav_bytes(opus_datas, sample_rate=16000, channels=1):
+    """
+    将opus帧列表解码为wav字节流
+    """
+    decoder = opuslib_next.Decoder(sample_rate, channels)
+    pcm_datas = []
+
+    frame_duration = 60  # ms
+    frame_size = int(sample_rate * frame_duration / 1000)  # 960
+
+    for opus_frame in opus_datas:
+        # 解码为PCM（返回bytes，2字节/采样点）
+        pcm = decoder.decode(opus_frame, frame_size)
+        pcm_datas.append(pcm)
+
+    pcm_bytes = b"".join(pcm_datas)
+
+    # 写入wav字节流
+    wav_buffer = BytesIO()
+    with wave.open(wav_buffer, "wb") as wf:
+        wf.setnchannels(channels)
+        wf.setsampwidth(2)  # 16bit
+        wf.setframerate(sample_rate)
+        wf.writeframes(pcm_bytes)
+    return wav_buffer.getvalue()
+
 def check_vad_update(before_config, new_config):
     if (
         new_config.get("selected_module") is None

main/xiaozhi-server/core/utils/wakeup_word.py

@@ -0,0 +1,140 @@
+import os
+import re
+import yaml
+import time
+import hashlib
+import portalocker
+from typing import Dict
+
+
+class FileLock:
+    def __init__(self, file, timeout=5):
+        self.file = file
+        self.timeout = timeout
+        self.start_time = None
+
+    def __enter__(self):
+        self.start_time = time.time()
+        while True:
+            try:
+                portalocker.lock(self.file, portalocker.LOCK_EX | portalocker.LOCK_NB)
+                return self.file
+            except portalocker.LockException:
+                if time.time() - self.start_time > self.timeout:
+                    raise TimeoutError("获取文件锁超时")
+                time.sleep(0.1)
+
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        portalocker.unlock(self.file)
+
+
+class WakeupWordsConfig:
+    def __init__(self):
+        self.config_file = "data/.wakeup_words.yaml"
+        self.assets_dir = "config/assets/wakeup_words"
+        self._ensure_directories()
+        self._config_cache = None
+        self._last_load_time = 0
+        self._cache_ttl = 1  # 缓存有效期（秒）
+        self._lock_timeout = 5  # 文件锁超时时间（秒）
+
+    def _ensure_directories(self):
+        """确保必要的目录存在"""
+        os.makedirs(os.path.dirname(self.config_file), exist_ok=True)
+        os.makedirs(self.assets_dir, exist_ok=True)
+
+    def _load_config(self) -> Dict:
+        """加载配置文件，使用缓存机制"""
+        current_time = time.time()
+
+        # 如果缓存有效，直接返回缓存
+        if (
+            self._config_cache is not None
+            and current_time - self._last_load_time < self._cache_ttl
+        ):
+            return self._config_cache
+
+        try:
+            with open(self.config_file, "a+") as f:
+                with FileLock(f, timeout=self._lock_timeout):
+                    f.seek(0)
+                    content = f.read()
+                    config = yaml.safe_load(content) if content else {}
+                    self._config_cache = config
+                    self._last_load_time = current_time
+                    return config
+        except (TimeoutError, IOError) as e:
+            print(f"加载配置文件失败: {e}")
+            return {}
+        except Exception as e:
+            print(f"加载配置文件时发生未知错误: {e}")
+            return {}
+
+    def _save_config(self, config: Dict):
+        """保存配置到文件，使用文件锁保护"""
+        try:
+            with open(self.config_file, "w") as f:
+                with FileLock(f, timeout=self._lock_timeout):
+                    yaml.dump(config, f, allow_unicode=True)
+                    self._config_cache = config
+                    self._last_load_time = time.time()
+        except (TimeoutError, IOError) as e:
+            print(f"保存配置文件失败: {e}")
+            raise
+        except Exception as e:
+            print(f"保存配置文件时发生未知错误: {e}")
+            raise
+
+    def get_wakeup_response(self, voice: str) -> Dict:
+        voice = hashlib.md5(voice.encode()).hexdigest()
+        """获取唤醒词回复配置"""
+        config = self._load_config()
+
+        if not config or voice not in config:
+            return None
+
+        # 检查文件大小
+        file_path = config[voice]["file_path"]
+        if not os.path.exists(file_path) or os.stat(file_path).st_size < (15 * 1024):
+            return None
+
+        return config[voice]
+
+    def update_wakeup_response(self, voice: str, file_path: str, text: str):
+        """更新唤醒词回复配置"""
+        try:
+            # 过滤表情符号
+            filtered_text = re.sub(r'[\U0001F600-\U0001F64F\U0001F900-\U0001F9FF]', '', text)
+            
+            config = self._load_config()
+            voice_hash = hashlib.md5(voice.encode()).hexdigest()
+            config[voice_hash] = {
+                "voice": voice,
+                "file_path": file_path,
+                "time": time.time(),
+                "text": filtered_text,
+            }
+            self._save_config(config)
+        except Exception as e:
+            print(f"更新唤醒词回复配置失败: {e}")
+            raise
+
+    def generate_file_path(self, voice: str) -> str:
+        """生成音频文件路径，使用voice的哈希值作为文件名"""
+        try:
+            # 生成voice的哈希值
+            voice_hash = hashlib.md5(voice.encode()).hexdigest()
+            file_path = os.path.join(self.assets_dir, f"{voice_hash}.wav")
+
+            # 如果文件已存在，先删除
+            if os.path.exists(file_path):
+                try:
+                    os.remove(file_path)
+                except Exception as e:
+                    print(f"删除已存在的音频文件失败: {e}")
+                    raise
+
+            return file_path
+        except Exception as e:
+            print(f"生成音频文件路径失败: {e}")
+            raise