VALL-E-X_Dataset / data_dir /metadata.csv
Kremon96's picture
Update data_dir/metadata.csv
7924ca8 verified
raw
history blame contribute delete
521 Bytes
from customs.make_custom_dataset import create_dataset
'''
How should the data_dir be created?
Place the necessary audio files in data_dir.
Transcription, tokenization, etc. of the audio files are done by the create_dataset function.
data_dir
├── bpe_69.json
├── utt1.wav
├── utt2.wav
├── utt3.wav
......
└── utt{n}.wav
'''
data_dir = "VALL-E-X_Dataset/data_dir"
create_dataset(data_dir, audio_path, filename, transcription, duration, sample_rate, asr_model, dataloader_process_only=True)