mrq/DL-Art-School
1
3
Fork 5
Code Issues Pull Requests Packages Projects Releases Wiki Activity

audio prep misc

Browse Source
This commit is contained in:
James Betker 2022-04-28 10:08:38 -06:00
parent f02b01bd9d
commit ab8176b217
3 changed files with 10 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
codes/scripts/audio/preparation/phase_1_split_files.py
View File

@ -47,9 +47,9 @@ def process_file(file, base_path, output_path, progress_file):
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('-path', type=str, help='Path to search for files', default='Y:\\sources\\big_podcast')
parser.add_argument('-progress_file', type=str, help='Place to store all files that have already been processed', default='Y:\\sources\\big_podcast\\already_processed.txt')
parser.add_argument('-output_path', type=str, help='Path for output files', default='Y:\\split\\big_podcast')
parser.add_argument('-path', type=str, help='Path to search for files', default='Y:\\clips\\red_rising')
parser.add_argument('-progress_file', type=str, help='Place to store all files that have already been processed', default='Y:\\clips\\red_rising\\already_processed.txt')
parser.add_argument('-output_path', type=str, help='Path for output files', default='Y:\\clips\\red_rising_split')
parser.add_argument('-num_threads', type=int, help='Number of concurrent workers processing files.', default=4)
args = parser.parse_args()

10
codes/scripts/audio/preparation/phase_2_sample_and_filter.py
View File

@ -72,7 +72,7 @@ def process_folder(folder, output_path, base_path, progress_file, max_files):
stft = torch.stft(clip, n_fft=22000, hop_length=1024, return_complex=True)
stft = stft[0, -2000:, :]
return (stft.real ** 2 + stft.imag ** 2).sqrt()
no_hifreq_data = get_spec_mags(clips).mean(dim=1) < .15
no_hifreq_data = get_spec_mags(clips).mean(dim=1) < .01
if torch.all(no_hifreq_data):
continue
@ -104,11 +104,11 @@ def process_folder(folder, output_path, base_path, progress_file, max_files):
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('-path', type=str, help='Path to search for split files (should be the direct output of phase 1)',
default='Y:\\split\\youtube')
parser.add_argument('-progress_file', type=str, help='Place to store all folders that have already been processed', default='Y:\\filtered\\big_podcast\\already_processed.txt')
parser.add_argument('-output_path', type=str, help='Path where sampled&filtered files are sent', default='Y:\\filtered\\youtube')
default='Y:\\clips\\red_rising_split')
parser.add_argument('-progress_file', type=str, help='Place to store all folders that have already been processed', default='Y:\\clips\\red_rising_filtered\\already_processed.txt')
parser.add_argument('-output_path', type=str, help='Path where sampled&filtered files are sent', default='Y:\\clips\\red_rising_filtered')
parser.add_argument('-num_threads', type=int, help='Number of concurrent workers processing files.', default=6)
parser.add_argument('-max_samples_per_folder', type=int, help='Maximum number of clips that can be extracted from each folder.', default=1000)
parser.add_argument('-max_samples_per_folder', type=int, help='Maximum number of clips that can be extracted from each folder.', default=999999)
parser.add_argument('-classifier_model_opt', type=str, help='Train/test options file that configures the model used to classify the audio clips.',
default='../options/test_noisy_audio_clips_classifier.yml')
args = parser.parse_args()

4
codes/scripts/audio/preparation/phase_3_generate_similarities.py
View File

@ -92,7 +92,7 @@ def process_subdir(subdir, options, clip_sz):
else:
for i in range(1,n): # The first entry is always the file itself.
top_ind = top3.indices[i]
simpaths.append(os.path.relpath(paths[top_ind], root))
simpaths.append(str(os.path.relpath(paths[top_ind], root)).replace('\\', '/'))
simmap[rel] = simpaths
torch.save(simmap, output_file)
@ -107,7 +107,7 @@ if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('-o', type=str, help='Path to the options YAML file used to train the CLIP model', default='../options/train_voice_voice_clip.yml')
parser.add_argument('--num_workers', type=int, help='Number concurrent processes to use', default=4)
parser.add_argument('--root_path', type=str, help='Root path to search for audio directories from', default='Y:\\filtered\\youtube')
parser.add_argument('--root_path', type=str, help='Root path to search for audio directories from', default='Y:\\clips\\red_rising_filtered')
parser.add_argument('--clip_size', type=int, help='Amount of audio samples to pull from each file', default=22050)
args = parser.parse_args()