paraformer streaming bugfix

游雁

2024-03-18 cbe2ea7e07cbf364827bd89cefc42b3f643ea3be

 runtime/python/onnxruntime/funasr_onnx/vad_bin.py

@@ -31,7 +31,8 @@
                quantize: bool = False,
                intra_op_num_threads: int = 4,
                max_end_sil: int = None,
                cache_dir: str = None
                cache_dir: str = None,
                **kwargs
                ):
      
      if not Path(model_dir).exists():
@@ -54,21 +55,17 @@
      if not os.path.exists(model_file):
         print(".onnx is not exist, begin to export onnx")
         try:
            from funasr.export.export_model import ModelExport
            from funasr import AutoModel
         except:
            raise "You are exporting onnx, please install funasr and try it again. To install funasr, you could:\n" \
                  "\npip3 install -U funasr\n" \
                  "For the users in China, you could install with the command:\n" \
                  "\npip3 install -U funasr -i https://mirror.sjtu.edu.cn/pypi/web/simple"
         export_model = ModelExport(
            cache_dir=cache_dir,
            onnx=True,
            device="cpu",
            quant=quantize,
         )
         export_model.export(model_dir)
      config_file = os.path.join(model_dir, 'vad.yaml')
      cmvn_file = os.path.join(model_dir, 'vad.mvn')
			
         model = AutoModel(model=model_dir)
         model_dir = model.export(type="onnx", quantize=quantize, **kwargs)
      config_file = os.path.join(model_dir, 'config.yaml')
      cmvn_file = os.path.join(model_dir, 'am.mvn')
      config = read_yaml(config_file)
      
      self.frontend = WavFrontend(
@@ -77,8 +74,8 @@
      )
      self.ort_infer = OrtInferSession(model_file, device_id, intra_op_num_threads=intra_op_num_threads)
      self.batch_size = batch_size
      self.vad_scorer = E2EVadModel(config["vad_post_conf"])
      self.max_end_sil = max_end_sil if max_end_sil is not None else config["vad_post_conf"]["max_end_silence_time"]
      self.vad_scorer = E2EVadModel(config["model_conf"])
      self.max_end_sil = max_end_sil if max_end_sil is not None else config["model_conf"]["max_end_silence_time"]
      self.encoder_conf = config["encoder_conf"]
   
   def prepare_cache(self, in_cache: list = []):
@@ -200,7 +197,8 @@
                quantize: bool = False,
                intra_op_num_threads: int = 4,
                max_end_sil: int = None,
                cache_dir: str = None
                cache_dir: str = None,
                **kwargs
                ):
      if not Path(model_dir).exists():
         try:
@@ -222,21 +220,18 @@
      if not os.path.exists(model_file):
         print(".onnx is not exist, begin to export onnx")
         try:
            from funasr.export.export_model import ModelExport
            from funasr import AutoModel
         except:
            raise "You are exporting onnx, please install funasr and try it again. To install funasr, you could:\n" \
                  "\npip3 install -U funasr\n" \
                  "For the users in China, you could install with the command:\n" \
                  "\npip3 install -U funasr -i https://mirror.sjtu.edu.cn/pypi/web/simple"
         export_model = ModelExport(
            cache_dir=cache_dir,
            onnx=True,
            device="cpu",
            quant=quantize,
         )
         export_model.export(model_dir)
      config_file = os.path.join(model_dir, 'vad.yaml')
      cmvn_file = os.path.join(model_dir, 'vad.mvn')
			
         model = AutoModel(model=model_dir)
         model_dir = model.export(type="onnx", quantize=quantize, **kwargs)
			
      config_file = os.path.join(model_dir, 'config.yaml')
      cmvn_file = os.path.join(model_dir, 'am.mvn')
      config = read_yaml(config_file)
      
      self.frontend = WavFrontendOnline(
@@ -245,8 +240,8 @@
      )
      self.ort_infer = OrtInferSession(model_file, device_id, intra_op_num_threads=intra_op_num_threads)
      self.batch_size = batch_size
      self.vad_scorer = E2EVadModel(config["vad_post_conf"])
      self.max_end_sil = max_end_sil if max_end_sil is not None else config["vad_post_conf"]["max_end_silence_time"]
      self.vad_scorer = E2EVadModel(config["model_conf"])
      self.max_end_sil = max_end_sil if max_end_sil is not None else config["model_conf"]["max_end_silence_time"]
      self.encoder_conf = config["encoder_conf"]
   
   def prepare_cache(self, in_cache: list = []):