FunASR/funasr/runtime/python/onnxruntime

Using funasr with ONNXRuntime

Introduction

• Model comes from speech_paraformer.

Steps:

1. Export the model.

   • Command: (Tips: torch >= 1.11.0 is required.)

   More details ref to (export docs)

   • e.g., Export model from modelscope

      python -m funasr.export.export_model --model-name damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type onnx --quantize False
     

   • e.g., Export model from local path, the model'name must be model.pb.

      python -m funasr.export.export_model --model-name ./damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type onnx --quantize False
     

2. Install the funasr_onnx.

   pip install funasr_onnx -i https://pypi.Python.org/simple
   

3. Run the demo.

   • Model_dir: the model path, which contains model.onnx, config.yaml, am.mvn.
   • Input: wav formt file, support formats: str, np.ndarray, List[str]
   • Output: List[str]: recognition result.

   • Example:

     from funasr_onnx import Paraformer
     
     model_dir = "/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch"
     model = Paraformer(model_dir, batch_size=1)
     
     wav_path = ['/nfs/zhifu.gzf/export/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch/example/asr_example.wav']
     
     result = model(wav_path)
     print(result)
     

Speed

Environment：Intel(R) Xeon(R) Platinum 8163 CPU @ 2.50GHz

Test wav, 5.53s, 100 times avg.

Backend	RTF
Pytorch	0.110
Onnx	0.038

Acknowledge

1. We acknowledge SWHL for contributing the onnxruntime(python api).