kxngluca/DL-Art-School
1
0
Fork 0
forked from mrq/DL-Art-School
Code Pull Requests Activity

Squelch annoying warning

Browse Source
This commit is contained in:
James Betker 2022-05-22 12:12:40 -06:00
parent 0c60f22197
commit 37640e9759
1 changed files with 9 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
codes/data/audio/unsupervised_audio_dataset.py
View File

@ -38,8 +38,8 @@ def load_audio(audiopath, sampling_rate):
audio = torchaudio.functional.resample(audio, lsr, sampling_rate) audio = torchaudio.functional.resample(audio, lsr, sampling_rate)
# Check some assumptions about audio range. This should be automatically fixed in load_wav_to_torch, but might not be in some edge cases, where we should squawk. # Check some assumptions about audio range. This should be automatically fixed in load_wav_to_torch, but might not be in some edge cases, where we should squawk.
# '2' is arbitrarily chosen since it seems like audio will often "overdrive" the [-1,1] bounds. # '10' is arbitrarily chosen since it seems like audio will often "overdrive" the [-1,1] bounds.
if torch.any(audio > 2) or not torch.any(audio < 0): if torch.any(audio > 10) or not torch.any(audio < 0):
print(f"Error with {audiopath}. Max={audio.max()} min={audio.min()}") print(f"Error with {audiopath}. Max={audio.max()} min={audio.min()}")
audio.clip_(-1, 1) audio.clip_(-1, 1)
@ -178,14 +178,16 @@ if __name__ == '__main__':
params = { params = {
'mode': 'unsupervised_audio', 'mode': 'unsupervised_audio',
'path': ['Y:\\split\\yt-music'], 'path': ['Y:\\split\\yt-music'],
'cache_path': 'Y:\\split\\yt-music\\cache-windows.pth', 'cache_path': 'Y:\\separated\\no-vocals-cache-win.pth',
'endswith': 'no_vocals.wav',
'sampling_rate': 22050, 'sampling_rate': 22050,
'pad_to_samples': 22050, 'pad_to_samples': 200000,
'resample_clip': False,
'extra_samples': 1,
'extra_sample_length': 100000,
'phase': 'train', 'phase': 'train',
'n_workers': 1, 'n_workers': 1,
'batch_size': 16, 'batch_size': 16,
'extra_samples': 4,
'resample_clip': False,
} }
from data import create_dataset, create_dataloader from data import create_dataset, create_dataloader
@ -196,5 +198,5 @@ if __name__ == '__main__':
for b_ in range(b['clip'].shape[0]): for b_ in range(b['clip'].shape[0]):
#pass #pass
torchaudio.save(f'{i}_clip_{b_}.wav', b['clip'][b_], ds.sampling_rate) torchaudio.save(f'{i}_clip_{b_}.wav', b['clip'][b_], ds.sampling_rate)
#torchaudio.save(f'{i}_resampled_clip_{b_}.wav', b['resampled_clip'][b_], ds.sampling_rate) torchaudio.save(f'{i}_alt_clip_{b_}.wav', b['alt_clips'][b_], ds.sampling_rate)
i += 1 i += 1