pre-processing - Githubissues

zhaojw1998 / Beat-Transformer

Codes for ISMIR 2022 paper: Beat Transformer: Demixed Beat and Downbeat Tracking with Dilated Self-Attention

MIT License

90 stars 17 forks source link

pre-processing #13

Open LeeYaNam opened 1 year ago

LeeYaNam commented 1 year ago

I tried to use your pre-processing (demixing.py) to convert the audio file into a spectrogram, but I got an error: 'dict' object has no attribute 'append'.

It seems that When you override the _separate_librosa method, out is a dictionary type and you cannot use the append method

zhaojw1998 commented 1 year ago

I tried to use your pre-processing (demixing.py) to convert the audio file into a spectrogram, but I got an error: 'dict' object has no attribute 'append'.

It seems that When you override the _separate_librosa method, out is a dictionary type and you cannot use the append method

Could you uncomment line 28, comment out lines 29 and 46-49, and try again, please? I was intending to change the "out" object from a dict to a list. But it seems I mistakenly preserved the original unmodified code. Please kindly let me know if it works :)

LeeYaNam commented 1 year ago

I tried to use your pre-processing (demixing.py) to convert the audio file into a spectrogram, but I got an error: 'dict' object has no attribute 'append'. It seems that When you override the _separate_librosa method, out is a dictionary type and you cannot use the append method

Could you uncomment line 28, comment out lines 29 and 46-49, and try again, please? I was intending to change the "out" object from a dict to a list. But it seems I mistakenly preserved the original unmodified code. Please kindly let me know if it works :)

It just got stuck

LeeYaNam commented 1 year ago

sorry，it was just running code. Wait a minute,please

LeeYaNam commented 1 year ago

sorry，it was just running code. Wait a minute,please I don't understand what's this meaning, I'm really sad...

zhaojw1998 commented 1 year ago

It looks like an error happening when reading the waveform data. Can you check that "/home/lyn/code/train_data/GTZAN/GTZAN_audio/" is the correct data directory on your computer? If no problem with the directory, then have you installed ffmpeg on your computer? There are some dependencies (ffmpeg & libsndfile) necessary for using Spleeter, see more details here.

LeeYaNam commented 1 year ago

It looks like an error happening when reading the waveform data. Can you check that "/home/lyn/code/train_data/GTZAN/GTZAN_audio/" is the correct data directory on your computer? If no problem with the directory, then have you installed ffmpeg on your computer? There are some dependencies (ffmpeg & libsndfile) necessary for using Spleeter, see more details here.

I checked that my path was correct and that the dependency packages were installed. I would like to know if it is possible to not use tensorflow? It feels like it's the reason for the error.

LeeYaNam commented 1 year ago

I found that the npy file was successfully generated after these errors were ignored! So the solution of uncommenting line 28, commenting lines 29 and 46-49 is valid.