myagent-ai 1.16.9 → 1.16.11

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "myagent-ai",
-  "version": "1.16.9",
+  "version": "1.16.11",
   "description": "本地桌面端执行型AI助手 - Open Interpreter 风格 | Local Desktop Execution-Oriented AI Assistant",
   "main": "main.py",
   "bin": {

package/web/api_server.py

@@ -1093,21 +1093,18 @@ class ApiServer:
 
             # ── 尝试 faster-whisper ──
             try:
-                # [v1.16.3] 抑制 HuggingFace 未认证警告和 pydub ffmpeg 警告
-                import warnings
-                warnings.filterwarnings("ignore", message=".*HF_TOKEN.*", category=UserWarning)
-                warnings.filterwarnings("ignore", message=".*huggingface_hub.*token.*", category=UserWarning)
-                warnings.filterwarnings("ignore", message=".*ffmpeg or avconv.*", category=RuntimeWarning)
-                # 抑制 HF Hub telemetry 和进度条
-                os.environ.setdefault("HF_HUB_DISABLE_TELEMETRY", "1")
-                os.environ.setdefault("HF_HUB_DISABLE_PROGRESS_BARS", "1")
-                os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
-
-                from faster_whisper import WhisperModel
-                whisper_model = getattr(self, '_whisper_model', None)
+                whisper_model = self._whisper_model
                 if whisper_model is None:
+                    # 预加载未完成或未安装，尝试懒加载
+                    import warnings as _w
+                    _w.filterwarnings("ignore", message=".*HF_TOKEN.*", category=UserWarning)
+                    _w.filterwarnings("ignore", message=".*huggingface_hub.*token.*", category=UserWarning)
+                    _w.filterwarnings("ignore", message=".*ffmpeg or avconv.*", category=RuntimeWarning)
+                    os.environ.setdefault("HF_HUB_DISABLE_TELEMETRY", "1")
+                    os.environ.setdefault("HF_HUB_DISABLE_PROGRESS_BARS", "1")
+                    os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
+                    from faster_whisper import WhisperModel
                     model_dir = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'models', 'whisper')
-                    # 使用 tiny 模型（最轻量，~39MB），CPU int8 量化
                     self._whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8",
                                                        download_root=model_dir)
                     whisper_model = self._whisper_model
@@ -1143,6 +1140,8 @@ class ApiServer:
 
                 wav_buf.seek(0)
                 segments, info = whisper_model.transcribe(wav_buf, beam_size=1,
+                                                          language="zh",
+                                                          initial_prompt="以下是普通话的句子",
                                                           vad_filter=True, vad_parameters=dict(
                                                               min_silence_duration_ms=300))
                 text = "".join(seg.text for seg in segments).strip()
@@ -1191,6 +1190,8 @@ class ApiServer:
                                 return web.json_response({"error": "音频格式不支持"}, status=400)
                         wav_buf.seek(0)
                         segments, info = whisper_model.transcribe(wav_buf, beam_size=1,
+                                                                  language="zh",
+                                                                  initial_prompt="以下是普通话的句子",
                                                                   vad_filter=True, vad_parameters=dict(
                                                                       min_silence_duration_ms=300))
                         text = "".join(seg.text for seg in segments).strip()
@@ -5710,6 +5711,30 @@ class ApiServer:
         # 加载禁用技能列表
         self._load_disabled_skills()
 
+        # [v1.16.10] 后台预加载 STT 模型，避免首次语音识别时等待数秒
+        # 在后台线程中加载，不阻塞服务启动
+        try:
+            import threading
+            def _preload_stt():
+                try:
+                    import warnings
+                    warnings.filterwarnings("ignore")
+                    os.environ.setdefault("HF_HUB_DISABLE_TELEMETRY", "1")
+                    os.environ.setdefault("HF_HUB_DISABLE_PROGRESS_BARS", "1")
+                    os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
+                    from faster_whisper import WhisperModel
+                    model_dir = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'models', 'whisper')
+                    self._whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8",
+                                                       download_root=model_dir)
+                    logger.info("STT faster-whisper 模型预加载完成")
+                except ImportError:
+                    logger.debug("faster-whisper 未安装，跳过 STT 预加载")
+                except Exception as e:
+                    logger.debug(f"STT 模型预加载失败（不影响使用）: {e}")
+            threading.Thread(target=_preload_stt, daemon=True).start()
+        except Exception:
+            pass
+
         # 恢复被中断的任务：自动重新投递群聊任务
         try:
             tp = self._get_task_persistence()