myagent-ai 1.19.9 → 1.20.0

package/core/deps_checker.py

@@ -107,8 +107,7 @@ DEPENDENCIES: List[DepInfo] = [
             note="PyTorch 音频处理库 (SenseVoice 必需)"),
     DepInfo("funasr", "funasr", "1.1.0", "stt", "all",
             note="[v1.18.8] SenseVoice 中文语音识别（首选，阿里达摩院）"),
-    DepInfo("faster_whisper", "faster-whisper", "1.0.0", "stt", "all",
-            note="Whisper 本地语音识别引擎 (备选，需 C++ 编译)"),
+
     DepInfo("speech_recognition", "SpeechRecognition", "3.10.0", "stt", "all",
             note="在线语音识别 (Google API，纯 Python 无需编译，Termux 兼容)"),
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "myagent-ai",
-  "version": "1.19.9",
+  "version": "1.20.0",
   "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/requirements-optional.txt

@@ -6,5 +6,5 @@
 funasr>=1.1.0
 torch>=2.0.0
 torchaudio>=2.0.0
-faster-whisper>=1.0.0
+
 pydub>=0.25.1

package/setup.py

@@ -41,7 +41,6 @@ setup(
         "funasr>=1.1.0",
         "torch>=2.0.0",
         "torchaudio>=2.0.0",
-        "faster-whisper>=1.0.0",
         # 浏览器自动化 (ChromeDev MCP, 无需 Playwright)
         # 桌面 GUI 自动化 (内置技能)
         "pynput>=1.7.6",
@@ -53,7 +52,7 @@ setup(
         "discord": ["discord.py>=2.3.0"],
         "anthropic": ["anthropic>=0.18.0"],
         "communication": ["cryptography>=41.0.0", "websockets>=12.0"],
-        "voice": ["funasr>=1.1.0", "torch>=2.0.0", "torchaudio>=2.0.0", "faster-whisper>=1.0.0"],
+        "voice": ["funasr>=1.1.0", "torch>=2.0.0", "torchaudio>=2.0.0"],
         "all": [
             "python-telegram-bot>=21.0",
             "discord.py>=2.3.0",
@@ -63,7 +62,6 @@ setup(
             "funasr>=1.1.0",
             "torch>=2.0.0",
             "torchaudio>=2.0.0",
-            "faster-whisper>=1.0.0",
         ],
     },
     entry_points={

package/web/api_server.py

@@ -1549,11 +1549,10 @@ window.toggleFullscreen = function() {{
 
         接受音频文件（WAV/WEBM/OGG），使用本地 STT 引擎转录。
         支持的引擎（按优先级）：
-        1. [v1.18.7] SenseVoice（推荐，中文识别最佳，需：pip install funasr torch torchaudio）
-        2. faster-whisper（备选，需安装：pip install faster-whisper）
-        3. vosk（备选，需安装：pip install vosk）
-        4. LLM API Whisper 兼容端点
-        5. SpeechRecognition（Google，需外网）
+        1. SenseVoice（推荐，中文识别最佳，需：pip install funasr torch torchaudio）
+        2. vosk（备选，需安装：pip install vosk）
+        3. LLM API Whisper 兼容端点
+        4. SpeechRecognition（Google，需外网）
         """
         try:
             reader = await request.multipart()
@@ -1639,117 +1638,6 @@ window.toggleFullscreen = function() {{
             except Exception as e:
                 logger.warning(f"SenseVoice 转录失败: {e}")
 
-            # ── 尝试 faster-whisper ──
-            try:
-                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')
-                    self._whisper_model = WhisperModel("tiny", device="cpu", compute_type="int8",
-                                                       download_root=model_dir)
-                    whisper_model = self._whisper_model
-                    logger.info("faster-whisper tiny 模型已加载 (CPU int8)")
-
-                # faster-whisper 需要 16kHz WAV
-                # [v1.15.8] 使用 pydub+ffmpeg 正确转换 WebM/Opus/OGG 等格式
-                wav_buf = io.BytesIO()
-                try:
-                    from pydub import AudioSegment
-                    audio_buf = io.BytesIO(audio_data)
-                    seg = AudioSegment.from_file(audio_buf, format=audio_format or "webm")
-                    seg = seg.set_channels(1).set_frame_rate(16000).set_sample_width(2)
-                    seg.export(wav_buf, format="wav")
-                    wav_buf.seek(0)
-                except Exception as _pydub_err:
-                    # pydub 不可用时 fallback：仅处理已是 WAV 的情况
-                    import wave
-                    audio_buf = io.BytesIO(audio_data)
-                    try:
-                        with wave.open(audio_buf, 'rb') as rf:
-                            wav_buf = io.BytesIO()
-                            with wave.open(wav_buf, 'wb') as wf:
-                                wf.setnchannels(1)
-                                wf.setsampwidth(2)
-                                wf.setframerate(16000)
-                                frames = rf.readframes(rf.getnframes())
-                                wf.writeframes(frames)
-                            wav_buf.seek(0)
-                    except Exception:
-                        logger.warning(f"音频格式转换失败（pydub: {_pydub_err}）")
-                        return web.json_response({"error": "音频格式不支持，需要 WAV 或安装 pydub+ffmpeg"}, 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()
-
-                if text:
-                    return web.json_response({"text": text, "engine": "faster-whisper"})
-            except ImportError:
-                logger.debug("faster-whisper 未安装，尝试自动安装...")
-                try:
-                    from core.deps_checker import ensure_skill_deps
-                    installed = ensure_skill_deps("stt")
-                    if installed:
-                        logger.info("faster-whisper 自动安装成功，重新尝试转录")
-                        from faster_whisper import WhisperModel
-                        import os
-                        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)
-                        whisper_model = self._whisper_model
-                        # 重新执行转录（跳过上面的 try 已完成的逻辑，直接进入转录流程）
-                        import io
-                        # [v1.15.8] 使用 pydub+ffmpeg 正确转换音频格式
-                        wav_buf = io.BytesIO()
-                        try:
-                            from pydub import AudioSegment
-                            audio_buf = io.BytesIO(audio_data)
-                            seg = AudioSegment.from_file(audio_buf, format=audio_format or "webm")
-                            seg = seg.set_channels(1).set_frame_rate(16000).set_sample_width(2)
-                            seg.export(wav_buf, format="wav")
-                            wav_buf.seek(0)
-                        except Exception as _pydub_err2:
-                            import wave
-                            audio_buf = io.BytesIO(audio_data)
-                            try:
-                                with wave.open(audio_buf, 'rb') as rf:
-                                    wav_buf = io.BytesIO()
-                                    with wave.open(wav_buf, 'wb') as wf:
-                                        wf.setnchannels(1)
-                                        wf.setsampwidth(2)
-                                        wf.setframerate(16000)
-                                        frames = rf.readframes(rf.getnframes())
-                                        wf.writeframes(frames)
-                                    wav_buf.seek(0)
-                            except Exception:
-                                logger.warning(f"音频格式转换失败（pydub: {_pydub_err2}）")
-                                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()
-                        if text:
-                            return web.json_response({"text": text, "engine": "faster-whisper"})
-                except Exception as inst_err:
-                    logger.warning(f"faster-whisper 自动安装/转录失败: {inst_err}")
-            except Exception as e:
-                logger.warning(f"faster-whisper 转录失败: {e}")
-
             # ── 尝试 vosk ──
             try:
                 import vosk
@@ -1872,9 +1760,8 @@ window.toggleFullscreen = function() {{
                 "error": "未检测到可用的 STT 引擎。请尝试以下方案：\n"
                          "  1. pip install funasr torch torchaudio  (SenseVoice，中文最佳，推荐)\n"
                          "  2. 配置支持 Whisper 的 LLM API（自动使用，无需安装）\n"
-                         "  3. pip install faster-whisper  (离线本地，需 C++ 编译环境)\n"
-                         "  4. pip install vosk             (离线本地，需下载模型)\n"
-                         "  5. pip install SpeechRecognition (需外网，国内不可用)",
+                         "  3. pip install vosk             (离线本地，需下载模型)\n"
+                         "  4. pip install SpeechRecognition (需外网，国内不可用)",
                 "available": False,
             }, status=503)
 
@@ -6772,8 +6659,8 @@ window.toggleFullscreen = function() {{
         except Exception:
             pass
 
-        # [v1.18.8] 后台预加载 STT 模型，避免首次语音识别时等待数秒
-        # 优先加载 SenseVoice（中文识别最佳），失败时回退到 faster-whisper
+        # 后台预加载 STT 模型，避免首次语音识别时等待数秒
+        # 仅加载 SenseVoice（中文识别最佳）
         try:
             import threading
             def _preload_stt():
@@ -6784,30 +6671,16 @@ window.toggleFullscreen = function() {{
                     os.environ.setdefault("HF_HUB_DISABLE_PROGRESS_BARS", "1")
                     os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
 
-                    # 首选: SenseVoice (funasr)
                     try:
                         from funasr import AutoModel
                         model_dir = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), 'models', 'sensevoice')
                         self._sensevoice_model = AutoModel(model="iic/SenseVoiceSmall", model_dir=model_dir,
                                                              device="cpu", disable_pbar=True, disable_update=True)
-                        logger.info("STT SenseVoice 模型预加载完成 (首选引擎)")
-                        return  # 成功则不加载 whisper
-                    except ImportError:
-                        logger.debug("SenseVoice (funasr) 未安装，尝试 faster-whisper")
-                    except Exception as e:
-                        logger.debug(f"SenseVoice 预加载失败: {e}，尝试 faster-whisper")
-
-                    # 备选: faster-whisper
-                    try:
-                        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 (备选引擎)")
+                        logger.info("STT SenseVoice 模型预加载完成")
                     except ImportError:
-                        logger.debug("faster-whisper 未安装，跳过 STT 预加载")
+                        logger.debug("SenseVoice (funasr) 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}")
+                        logger.debug(f"SenseVoice 预加载失败（不影响使用）: {e}")
                 except Exception as e:
                     logger.debug(f"STT 预加载异常（不影响使用）: {e}")
             threading.Thread(target=_preload_stt, daemon=True).start()