| from typing import List | |
| import argparse | |
| import torch | |
| from speechbrain.pretrained import EncoderDecoderASR | |
| def asr_model_inference(asr_model: EncoderDecoderASR, audios: List[str]) -> List[str]: | |
| return [asr_model.transcribe_file(audio) for audio in audios] | |
| if __name__ == "__main__": | |
| parser = argparse.ArgumentParser() | |
| parser.add_argument("-I", dest="audio_file", required=True) | |
| args = parser.parse_args() | |
| asr_model = EncoderDecoderASR.from_hparams( | |
| source="./infernce", hparams_file="hyperparams.yaml", savedir="inference", run_opts={"device": "cpu"}) | |
| print(asr_model_inference(asr_model, [args.audio_file])) |