update:替换词功能

main/xiaozhi-server/config/config_loader.py

@@ -1,7 +1,8 @@
 import os
+import asyncio
 import yaml
 from collections.abc import Mapping
-from config.manage_api_client import init_service, get_server_config, get_agent_models
+from config.manage_api_client import init_service, get_server_config, get_agent_models, get_correct_words
 
 
 def get_project_dir():
@@ -87,7 +88,16 @@ async def get_config_from_api_async(config):
 
 async def get_private_config_from_api(config, device_id, client_id):
     """从Java API获取私有配置"""
-    return await get_agent_models(device_id, client_id, config["selected_module"])
+    results = await asyncio.gather(
+        get_agent_models(device_id, client_id, config["selected_module"]),
+        get_correct_words(device_id),
+        return_exceptions=True
+    )
+    private_config = results[0] if not isinstance(results[0], Exception) else {}
+    correct_words = results[1] if not isinstance(results[1], Exception) else None
+    if correct_words:
+        private_config["correct_words"] = correct_words
+    return private_config
 
 
 def ensure_directories(config):

main/xiaozhi-server/config/manage_api_client.py

@@ -183,6 +183,18 @@ async def get_agent_models(
     )
 
 
+async def get_correct_words(mac_address: str) -> Optional[Dict]:
+    """获取智能体替换词"""
+    try:
+        return await ManageApiClient._instance._execute_async_request(
+            "POST", "/config/correct-words",
+            json={"macAddress": mac_address}
+        )
+    except Exception as e:
+        print(f"获取替换词失败: {e}")
+        return None
+
+
 async def generate_and_save_chat_summary(session_id: str) -> Optional[Dict]:
     """生成并保存聊天记录总结"""
     try:

main/xiaozhi-server/core/connection.py

@@ -726,6 +726,13 @@ class ConnectionHandler:
         if private_config.get("context_providers", None) is not None:
             self.config["context_providers"] = private_config["context_providers"]
 
+        # 注入替换词到 TTS 模块配置
+        if private_config.get("correct_words", None) is not None:
+            select_tts_module = self.config["selected_module"]["TTS"]
+            self.config["TTS"][select_tts_module]["correct_words"] = private_config[
+                "correct_words"
+            ]
+
         # 使用 run_in_executor 在线程池中执行 initialize_modules，避免阻塞主循环
         try:
             modules = await self.loop.run_in_executor(

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

@@ -194,6 +194,8 @@ class TTSProvider(TTSProviderBase):
 
             # 过滤Markdown
             filtered_text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
 
             if filtered_text:
                 # 发送continue-task消息
@@ -480,6 +482,8 @@ class TTSProvider(TTSProviderBase):
 
                     # 发送文本
                     filtered_text = MarkdownCleaner.clean_markdown(text)
+                    if self._correct_words_pattern:
+                        filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
                     # 发送continue-task消息
                     continue_task_message = {
                         "header": {

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

@@ -295,6 +295,8 @@ class TTSProvider(TTSProviderBase):
                 logger.bind(tag=TAG).warning(f"WebSocket连接不存在，终止发送文本")
                 return
             filtered_text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
             if filtered_text:
                 run_request = {
                     "header": {
@@ -564,6 +566,8 @@ class TTSProvider(TTSProviderBase):
 
                     # 发送文本合成请求
                     filtered_text = MarkdownCleaner.clean_markdown(text)
+                    if self._correct_words_pattern:
+                        filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
                     run_request = {
                         "header": {
                             "message_id": uuid.uuid4().hex,

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

@@ -45,6 +45,21 @@ class TTSProviderBase(ABC):
         self.report_on_last = False
         # sentence_id 到文本的映射，用于流式TTS获取正确的字幕文本
         self._sentence_text_map = {}
+        # 加载替换词，用于一次性正则替换
+        raw_words = config.get("correct_words", [])
+        self.correct_words = {}
+        for item in raw_words:
+            parts = item.split("|", 1)
+            if len(parts) == 2:
+                self.correct_words[parts[0]] = parts[1]
+        # 构建正则表达式，使用最长匹配优先（排序后转义拼接）
+        if self.correct_words:
+            # 按key长度降序排列，长的先匹配，避免短词部分干扰
+            sorted_keys = sorted(self.correct_words.keys(), key=len, reverse=True)
+            pattern_str = '|'.join(re.escape(k) for k in sorted_keys)
+            self._correct_words_pattern = re.compile(pattern_str)
+        else:
+            self._correct_words_pattern = None
 
         self.tts_text_buff = []
         self.punctuations = (
@@ -89,7 +104,12 @@ class TTSProviderBase(ABC):
         self.before_stop_play_files.append((file_audio, text))
 
     def to_tts_stream(self, text, opus_handler: Callable[[bytes], None] = None) -> None:
+        # 保留原始文本用于显示/上报
+        original_text = text
         text = MarkdownCleaner.clean_markdown(text)
+        # 使用正则一次性替换，避免重复遍历和部分匹配问题
+        if self._correct_words_pattern:
+            text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], 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
@@ -97,7 +117,8 @@ class TTSProviderBase(ABC):
                 try:
                     audio_bytes = asyncio.run(self.text_to_speak(text, None))
                     if audio_bytes:
-                        self.tts_audio_queue.put((SentenceType.FIRST, None, text, getattr(self, 'current_sentence_id', None)))
+                        # 使用原始文本用于显示/上报
+                        self.tts_audio_queue.put((SentenceType.FIRST, None, original_text, getattr(self, 'current_sentence_id', None)))
                         audio_bytes_to_data_stream(
                             audio_bytes,
                             file_type=self.audio_file_type,
@@ -111,16 +132,16 @@ class TTSProviderBase(ABC):
                         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}"
+                        f"语音生成失败{5 - max_repeat_time + 1}次: {original_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"
+                    f"语音生成成功: {original_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"
+                    f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                 )
             return None
         else:
@@ -131,7 +152,7 @@ class TTSProviderBase(ABC):
                         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}"
+                            f"语音生成失败{5 - max_repeat_time + 1}次: {original_text}，错误: {e}"
                         )
                         # 未执行成功，删除文件
                         if os.path.exists(tmp_file):
@@ -140,20 +161,24 @@ class TTSProviderBase(ABC):
 
                 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"
+                        f"语音生成成功: {original_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"
+                        f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                     )
-                self.tts_audio_queue.put((SentenceType.FIRST, None, text, getattr(self, 'current_sentence_id', None)))
+                self.tts_audio_queue.put((SentenceType.FIRST, None, original_text, getattr(self, 'current_sentence_id', None)))
                 self._process_audio_file_stream(tmp_file, callback=opus_handler)
             except Exception as e:
                 logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
                 return None
     
     def to_tts(self, text):
+        # 保留原始文本用于日志/显示
+        original_text = text
         text = MarkdownCleaner.clean_markdown(text)
+        if self._correct_words_pattern:
+            text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], 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
@@ -174,16 +199,16 @@ class TTSProviderBase(ABC):
                         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}"
+                        f"语音生成失败{5 - max_repeat_time + 1}次: {original_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"
+                    f"语音生成成功: {original_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"
+                    f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                 )
             return None
         else:
@@ -194,7 +219,7 @@ class TTSProviderBase(ABC):
                         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}"
+                            f"语音生成失败{5 - max_repeat_time + 1}次: {original_text}，错误: {e}"
                         )
                         # 未执行成功，删除文件
                         if os.path.exists(tmp_file):
@@ -203,11 +228,11 @@ class TTSProviderBase(ABC):
 
                 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"
+                        f"语音生成成功: {original_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"
+                        f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                     )
 
                 return tmp_file

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

@@ -370,6 +370,8 @@ class TTSProvider(TTSProviderBase):
 
             #  过滤Markdown
             filtered_text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
 
             if filtered_text:
                 # 发送文本

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

@@ -92,22 +92,25 @@ class TTSProvider(TTSProviderBase):
     def to_tts_single_stream(self, text, is_last=False):
         try:
             max_repeat_time = 5
+            original_text = text
             text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], text)
             try:
                 asyncio.run(self.text_to_speak(text, is_last))
             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}"
+                    f"语音生成失败{5 - max_repeat_time + 1}次: {original_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"
+                    f"语音生成成功: {original_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"
+                    f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                 )
         except Exception as e:
             logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
@@ -203,6 +206,8 @@ class TTSProvider(TTSProviderBase):
         """
         start_time = time.time()
         text = MarkdownCleaner.clean_markdown(text)
+        if self._correct_words_pattern:
+            text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], text)
 
         payload = {"text": text, "character": self.voice}
 

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

@@ -84,22 +84,25 @@ class TTSProvider(TTSProviderBase):
     def to_tts_single_stream(self, text, is_last=False):
         try:
             max_repeat_time = 5
+            original_text = text
             text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], text)
             try:
                 asyncio.run(self.text_to_speak(text, is_last))
             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}"
+                    f"语音生成失败{5 - max_repeat_time + 1}次: {original_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"
+                    f"语音生成成功: {original_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"
+                    f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                 )
         except Exception as e:
             logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
@@ -202,6 +205,8 @@ class TTSProvider(TTSProviderBase):
         """
         start_time = time.time()
         text = MarkdownCleaner.clean_markdown(text)
+        if self._correct_words_pattern:
+            text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], text)
 
         params = {
             "tts_text": text,

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

@@ -148,22 +148,25 @@ class TTSProvider(TTSProviderBase):
     def to_tts_single_stream(self, text, is_last=False):
         try:
             max_repeat_time = 5
+            original_text = text
             text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], text)
             try:
                 asyncio.run(self.text_to_speak(text, is_last))
             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}"
+                    f"语音生成失败{5 - max_repeat_time + 1}次: {original_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"
+                    f"语音生成成功: {original_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"
+                    f"语音生成失败: {original_text}，请检查网络或服务是否正常"
                 )
         except Exception as e:
             logger.bind(tag=TAG).error(f"Failed to generate TTS file: {e}")
@@ -298,6 +301,8 @@ class TTSProvider(TTSProviderBase):
         """
         start_time = time.time()
         text = MarkdownCleaner.clean_markdown(text)
+        if self._correct_words_pattern:
+            text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], text)
 
         payload = {
             "model": self.model,

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

@@ -245,6 +245,8 @@ class TTSProvider(TTSProviderBase):
                 return
 
             filtered_text = MarkdownCleaner.clean_markdown(text)
+            if self._correct_words_pattern:
+                filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
             if filtered_text:
                 # 发送文本合成请求
                 run_request = self._build_base_request(status=1,text=filtered_text)
@@ -441,6 +443,8 @@ class TTSProvider(TTSProviderBase):
 
                 try:
                     filtered_text = MarkdownCleaner.clean_markdown(text)
+                    if self._correct_words_pattern:
+                        filtered_text = self._correct_words_pattern.sub(lambda m: self.correct_words[m.group(0)], filtered_text)
 
                     text_request = self._build_base_request(status=2,text=filtered_text)