AssertionError assert len(wav_np.shape) == 1

[lihuining]

Traceback (most recent call last):
File "/root/paddlejob/workspace/env_run/lihuining01/ASR/FireRedASR/infer_aed.py", line 14, in
results = model.transcribe(
File "/root/anaconda3/envs/fireredasr/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/root/paddlejob/workspace/env_run/lihuining01/ASR/FireRedASR/fireredasr/models/fireredasr.py", line 44, in transcribe
feats, lengths, durs = self.feat_extractor(batch_wav_path)
File "/root/paddlejob/workspace/env_run/lihuining01/ASR/FireRedASR/fireredasr/data/asr_feat.py", line 22, in call
fbank = self.fbank((sample_rate, wav_np))
File "/root/paddlejob/workspace/env_run/lihuining01/ASR/FireRedASR/fireredasr/data/asr_feat.py", line 96, in call
assert len(wav_np.shape) == 1
AssertionError

I encountered this error when running infer_aed.py, and I also print the audio, Audio shape: (3529728, 2)
Audio dtype: int16, how to avoid this error?

[kaituoxu]

ffmpeg -i input_audio -ar 16000 -ac 1 -acodec pcm_s16le -f wav output.wav

use this to preprocess you audio file