search

RookieJunChen / FullSubNet-plus

The official PyTorch implementation of "FullSubNet+: Channel Attention FullSubNet with Complex Spectrograms for Speech Enhancement".
Apache License 2.0
235 stars 55 forks source link

soundfile.LibsndfileError: Error opening 'xx/xx/xx.wav': File contains data in an unknown format. #15

Open QHCV opened 1 year ago

QHCV commented 1 year ago

I find a problem when training my model:

soundfile.LibsndfileError: Error opening 'xx/xx/xx.wav': File contains data in an unknown format.

I run this in Ubuntu

I have tried many methods. What can I do about this problem? Think you!

Plutoisme commented 1 year ago

ffmpeg is all you need, i meet the same question in other projects, you can try conda install ffmpeg.

QHCV commented 1 year ago

ffmpeg is all you need, i meet the same question in other projects, you can try conda install ffmpeg.

image The problem is the same.

MonkeyDNoppu commented 1 year ago

ffmpeg cannot solve my problem (fyi, I'm trying to asr a mp4)

QHCV commented 1 year ago

I'm very sorry, but it turned out that the audio files in the dataset were corrupted, which caused this error. The dataset I used is https://github.com/microsoft/DNS-Challenge/tree/interspeech2020/master. I switched to other data sets and there was no error, thank you for your help.

Plutoisme commented 1 year ago

when you used the DNS-Challenge dataset, did you use the shell file 'download-dns-challenge-4.sh ' to download the data?

QHCV commented 1 year ago

when you used the DNS-Challenge dataset, did you use the shell file 'download-dns-challenge-4.sh ' to download the data? image

No, I use the DCN 2020 dataset.

Plutoisme commented 1 year ago

when you used the DNS-Challenge dataset, did you use the shell file 'download-dns-challenge-4.sh ' to download the data? image

No, I use the DCN 2020 dataset.

Sometimes the similar bug is relate to the encoding format，I usually use the same function to deal with different .wav files, sometimes show bug, ffmpeg may solve a part of the problem.

QHCV commented 1 year ago

Sometimes the similar bug is relate to the encoding format，I usually use the same function to deal with different .wav files, sometimes show bug, ffmpeg may solve a part of the problem.

Thank you for sharing your experience.