update:添加indextts 流式语音合成

main/manager-api/src/main/resources/db/changelog/202508081701.sql

@@ -0,0 +1,32 @@
+-- 添加Index-TTS-vLLM流式TTS供应器
+delete from `ai_model_provider` where id = 'SYSTEM_TTS_IndexStreamTTS';
+INSERT INTO `ai_model_provider` (`id`, `model_type`, `provider_code`, `name`, `fields`, `sort`, `creator`, `create_date`, `updater`, `update_date`) VALUES
+('SYSTEM_TTS_IndexStreamTTS', 'TTS', 'index_stream', 'Index-TTS-vLLM流式语音合成', '[{"key":"api_url","label":"API服务地址","type":"string"},{"key":"voice","label":"默认音色","type":"string"},{"key":"audio_format","label":"音频格式","type":"string"},{"key":"output_dir","label":"输出目录","type":"string"}]', 16, 1, NOW(), 1, NOW());
+
+-- 添加Index-TTS-vLLM流式TTS模型配置
+delete from `ai_model_config` where id = 'TTS_IndexStreamTTS';
+INSERT INTO `ai_model_config` VALUES ('TTS_IndexStreamTTS', 'TTS', 'IndexStreamTTS', 'Index-TTS-vLLM流式语音合成', 0, 1, '{\"type\": \"index_stream\", \"api_url\": \"http://127.0.0.1:11996/tts\", \"voice\": \"jay_klee\", \"audio_format\": \"pcm\", \"output_dir\": \"tmp/\"}', NULL, NULL, 19, NULL, NULL, NULL, NULL);
+
+-- 更新Index-TTS-vLLM流式TTS配置说明
+UPDATE `ai_model_config` SET 
+`doc_link` = 'https://github.com/Ksuriuri/index-tts-vllm',
+`remark` = 'Index-TTS-vLLM流式TTS配置说明：
+1. Index-TTS-vLLM是基于Index-TTS项目的vLLM推理服务，提供流式语音合成功能
+2. 支持多种音色，音质自然，适合各种语音交互场景
+3. 需要先部署Index-TTS-vLLM服务，然后配置API地址
+4. 支持实时流式合成，具有较低的延迟
+5. 支持自定义音色，可在项目assets文件夹下注册新音色
+部署步骤：
+1. 克隆项目：git clone https://github.com/Ksuriuri/index-tts-vllm.git
+2. 安装依赖：pip install -r requirements.txt
+3. 启动服务：python app.py
+4. 服务默认运行在 http://127.0.0.1:11996
+5. 如需其他音色，可到项目assets文件夹下注册
+6. 支持多种音频格式：pcm、wav、mp3等
+如需了解更多配置，请参考：https://github.com/Ksuriuri/index-tts-vllm/blob/master/README.md
+' WHERE `id` = 'TTS_IndexStreamTTS';
+
+-- 添加Index-TTS-vLLM流式TTS音色
+delete from `ai_tts_voice` where tts_model_id = 'TTS_IndexStreamTTS';
+-- 默认音色
+INSERT INTO `ai_tts_voice` VALUES ('TTS_IndexStreamTTS_0001', 'TTS_IndexStreamTTS', 'Jay Klee', 'jay_klee', '中文及中英文混合', NULL, NULL, NULL, NULL, 1, NULL, NULL, NULL, NULL);

main/manager-api/src/main/resources/db/changelog/db.changelog-master.yaml

@@ -282,3 +282,10 @@ databaseChangeLog:
         - sqlFile:
             encoding: utf8
             path: classpath:db/changelog/202507081646.sql
+  - changeSet:
+      id: 202508081701
+      author: hrz
+      changes:
+        - sqlFile:
+            encoding: utf8
+            path: classpath:db/changelog/202508081701.sql

main/xiaozhi-server/config.yaml

@@ -902,7 +902,7 @@ TTS:
   IndexStreamTTS:
     # 基于Index-TTS-vLLM项目的TTS接口服务
     # 参照教程：https://github.com/Ksuriuri/index-tts-vllm/blob/master/README.md
-    type: IndexStream
+    type: index_stream
     api_url: http://127.0.0.1:11996/tts
     audio_format: "pcm"
     # 默认音色，如需其他音色可到项目assets文件夹下注册

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

@@ -19,7 +19,11 @@ class TTSProvider(TTSProviderBase):
     def __init__(self, config, delete_audio_file):
         super().__init__(config, delete_audio_file)
         self.interface_type = InterfaceType.SINGLE_STREAM
-        self.voice = config.get("voice", "xiao_he")  # 默认音色
+        self.voice = config.get("voice", "xiao_he")
+        if config.get("private_voice"):
+            self.voice = config.get("private_voice")
+        else:
+            self.voice = config.get("voice", "xiao_he")
         self.api_url = config.get("api_url", "http://8.138.114.124:11996/tts")
         self.audio_format = "pcm"
         self.before_stop_play_files = []
@@ -80,7 +84,7 @@ class TTSProvider(TTSProviderBase):
             bool: 是否成功处理了文本
         """
         full_text = "".join(self.tts_text_buff)
-        remaining_text = full_text[self.processed_chars:]
+        remaining_text = full_text[self.processed_chars :]
         if remaining_text:
             segment_text = textUtils.get_string_no_punctuation_or_emoji(remaining_text)
             if segment_text:
@@ -118,10 +122,7 @@ class TTSProvider(TTSProviderBase):
 
     async def text_to_speak(self, text, is_last):
         """流式处理TTS音频，每句只推送一次音频列表"""
-        payload = {
-            "text": text,
-            "character": self.voice
-        }
+        payload = {"text": text, "character": self.voice}
 
         frame_bytes = int(
             self.opus_encoder.sample_rate
@@ -132,9 +133,7 @@ class TTSProvider(TTSProviderBase):
         )  # 16-bit = 2 bytes
         try:
             async with aiohttp.ClientSession() as session:
-                async with session.post(
-                        self.api_url, json=payload, timeout=10
-                ) as resp:
+                async with session.post(self.api_url, json=payload, timeout=10) as resp:
 
                     if resp.status != 200:
                         logger.bind(tag=TAG).error(
@@ -220,15 +219,10 @@ class TTSProvider(TTSProviderBase):
         start_time = time.time()
         text = MarkdownCleaner.clean_markdown(text)
 
-        payload = {
-            "text": text,
-            "character": self.character
-        }
+        payload = {"text": text, "character": self.character}
 
         try:
-            with requests.post(
-                    self.api_url, json=payload, timeout=5
-            ) as response:
+            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}"
@@ -252,7 +246,7 @@ class TTSProvider(TTSProviderBase):
 
                 # 分帧处理PCM数据
                 for i in range(0, len(pcm_data), frame_bytes):
-                    frame = pcm_data[i: i + 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))