[fix] gradio audio devices

README.md

@@ -187,6 +187,8 @@ cp -r minimind-3o ./scripts/minimind-3o
 cd scripts && python web_demo_omni.py
 ```
 
+注：`scripts/web_demo_omni.py` 是 Gradio 非实时演示，上传/录音音频会在后端按实际采样率重采样到 16k；实时语音通话请使用 `webui/web_demo.py`。
+
 ## Ⅱ 🛠️ 模型训练
 
 <details style="color:rgb(128,128,128)">

README_en.md

@@ -188,6 +188,8 @@ cp -r minimind-3o ./scripts/minimind-3o
 cd scripts && python web_demo_omni.py
 ```
 
+Note: `scripts/web_demo_omni.py` is the non-realtime Gradio demo. Uploaded or recorded audio is resampled on the backend from its actual sample rate to 16 kHz; use `webui/web_demo.py` for realtime voice calls.
+
 ## Ⅱ 🛠️ Training
 
 <details style="color:rgb(128,128,128)">

scripts/web_demo_omni.py

@@ -37,6 +37,40 @@ voices_data = {}
 builtin_voices, clone_voices = set(), set()
 
 
+def default_device():
+    if torch.cuda.is_available():
+        return 'cuda'
+    if hasattr(torch.backends, 'mps') and torch.backends.mps.is_available():
+        return 'mps'
+    return 'cpu'
+
+
+def resolve_aux_device(value, fallback):
+    if value and value != 'auto':
+        return value
+    return fallback
+
+
+def get_device_type(value):
+    return torch.device(value).type
+
+
+def normalize_audio(samples, sr, target_sr=16000):
+    if len(samples.shape) > 1:
+        samples = samples.mean(axis=1)
+    if np.issubdtype(samples.dtype, np.integer):
+        max_value = max(abs(np.iinfo(samples.dtype).min), np.iinfo(samples.dtype).max)
+        samples = samples.astype(np.float32) / max_value
+    else:
+        samples = samples.astype(np.float32)
+        peak = np.abs(samples).max() if samples.size else 0
+        if peak > 1.0:
+            samples = samples / peak
+    if sr != target_sr:
+        samples = librosa.resample(samples.astype(float), orig_sr=sr, target_sr=target_sr).astype(np.float32)
+    return np.ascontiguousarray(samples, dtype=np.float32)
+
+
 def frames_to_mimi(frames):
     codes = [f for f in frames if f and len(f) == 8]
     if not codes:
@@ -111,12 +145,7 @@ def chat_stream(prompt, audio_input=None, image_input=None, voice_name="default"
 
     if audio_input is not None:
         sr, samples = audio_input
-        if len(samples.shape) > 1:
+        samples = normalize_audio(samples, sr)
-            samples = samples.mean(axis=1)
-        if samples.dtype != np.float32:
-            samples = samples.astype(np.float32) / max(np.abs(samples).max(), 1)
-        if sr != 16000:
-            samples = librosa.resample(samples.astype(float), orig_sr=sr, target_sr=16000).astype(np.float32)
         inputs = model.audio_processor(samples, sampling_rate=16000, return_tensors="pt", return_attention_mask=True)
         mel = inputs.input_features.squeeze(0)
         valid_len = inputs.attention_mask.sum().item()
@@ -266,13 +295,17 @@ if __name__ == '__main__':
     parser.add_argument('--audio_encoder', default='../model/SenseVoiceSmall', type=str)
     parser.add_argument('--vision_model', default='../model/siglip2-base-p32-256-ve', type=str)
     parser.add_argument('--mimi_path', default='../model/mimi', type=str)
-    parser.add_argument('--device', default='cuda' if torch.cuda.is_available() else 'cpu', type=str)
+    parser.add_argument('--device', default=default_device(), type=str)
+    parser.add_argument('--asr_device', default='auto', type=str, help='ASR设备；auto 在 MPS 主设备下使用 cpu，其它情况跟随 --device')
+    parser.add_argument('--mimi_device', default='auto', type=str, help='Mimi 解码设备；auto 跟随 --device')
     parser.add_argument('--open_thinking', default=0, type=int, choices=[0, 1])
     parser.add_argument('--top_p', default=0.85, type=float)
     parser.add_argument('--port', default=8888, type=int)
     args = parser.parse_args()
 
     device = args.device
+    asr_device = resolve_aux_device(args.asr_device, 'cpu' if get_device_type(device) == 'mps' else device)
+    mimi_device = resolve_aux_device(args.mimi_device, device)
     model_dict = scan_hf_models(args.load_from)
     if not model_dict:
         print(f"未在 {os.path.abspath(args.load_from)} 找到 transformers 模型")
@@ -281,14 +314,16 @@ if __name__ == '__main__':
     load_hf_model(model_dict[current_model_name])
 
     try:
-        mimi_model = MimiModel.from_pretrained(args.mimi_path).eval()
+        mimi_model = MimiModel.from_pretrained(args.mimi_path).eval().to(mimi_device)
-        print('Mimi model loaded')
+        if get_device_type(mimi_device) != 'cpu':
+            mimi_model = mimi_model.half()
+        print(f'Mimi model loaded on {mimi_device}')
     except Exception:
         mimi_model = None
         print('Mimi model not found, audio output disabled')
 
     with contextlib.redirect_stdout(io.StringIO()):
-        asr_model = AutoModel(model=args.audio_encoder, trust_remote_code=True, device=device, disable_update=True)
+        asr_model = AutoModel(model=args.audio_encoder, trust_remote_code=True, device=asr_device, disable_update=True)
     load_voices()
     print(f'Voices loaded: {list(voices_data.keys()) or "none"}')
     launch_gradio(server_port=args.port)