- Python3.7+  
- PyTorch 1.8.1+
- torchaudio
- soundfile 
- librosa 

git clone https://github.com/gaochangfeng/lighting-asr.git
cd lighting-asr
python setup.py install

from lasr.process.asrprocess import ASRProcess
train_config="hparams.yaml" 
decode_config="decode.yaml"
model_path="model.ckpt"
asrpipeline = ASRProcess(
    train_config=train_config, 
    decode_config=decode_config, 
    model_path=model_path
)
token, text = asrpipeline("test.wav")
print(token)
print(text)