add device option in wespeaker cli; set onnxruntime log level

wespeaker/cli/speaker.py

@@ -72,11 +72,7 @@ def set_resample_rate(self, resample_rate: int):
     def set_vad(self, apply_vad: bool):
         self.apply_vad = apply_vad
 
-    def set_gpu(self, device_id: int):
-        if device_id >= 0:
-            device = 'cuda:{}'.format(device_id)
-        else:
-            device = 'cpu'
+    def set_device(self, device: str):
         self.device = torch.device(device)
         self.model = self.model.to(self.device)
 
@@ -304,7 +300,7 @@ def main():
         model = load_model_local(args.pretrain)
     model.set_resample_rate(args.resample_rate)
     model.set_vad(args.vad)
-    model.set_gpu(args.gpu)
+    model.set_device(args.device)
     model.set_diarization_params(min_duration=args.diar_min_duration,
                                  window_secs=args.diar_window_secs,
                                  period_secs=args.diar_period_secs,

wespeaker/cli/utils.py

@@ -52,11 +52,12 @@ def get_args():
                         type=str,
                         default="",
                         help='model directory')
-    parser.add_argument('-g',
-                        '--gpu',
-                        type=int,
-                        default=-1,
-                        help='which gpu to use (number <0 means using cpu)')
+    parser.add_argument('--device',
+                        type=str,
+                        default='cpu',
+                        help="device type (most commonly cpu or cuda,"
+                             "but also potentially mps, xpu, xla or meta)"
+                             "and optional device ordinal for the device type.")
     parser.add_argument('--audio_file', help='audio file')
     parser.add_argument('--audio_file2',
                         help='audio file2, specifically for similarity task')

wespeaker/diar/extract_emb.py

@@ -37,7 +37,7 @@ def init_session(source, device):
     opts = ort.SessionOptions()
     opts.inter_op_num_threads = 1
     opts.intra_op_num_threads = 1
-    opts.log_severity_level = 0
+    opts.log_severity_level = 1
     session = ort.InferenceSession(source,
                                    sess_options=opts,
                                    providers=providers)