audio_check / automatic-speech-recognition /hw_asr /datasets /custom_dir_audio_dataset.py
tgritsaev's picture
Upload 198 files
affcd23 verified
raw
history blame contribute delete
899 Bytes
import logging
from pathlib import Path
from hw_asr.datasets.custom_audio_dataset import CustomAudioDataset
logger = logging.getLogger(__name__)
class CustomDirAudioDataset(CustomAudioDataset):
def __init__(self, audio_dir, transcription_dir=None, *args, **kwargs):
data = []
for path in Path(audio_dir).iterdir():
entry = {}
if path.suffix in [".mp3", ".wav", ".flac", ".m4a"]:
entry["path"] = str(path)
if transcription_dir and Path(transcription_dir).exists():
transc_path = Path(transcription_dir) / (path.stem + '.txt')
if transc_path.exists():
with transc_path.open() as f:
entry["text"] = f.read().strip()
if len(entry) > 0:
data.append(entry)
super().__init__(data, *args, **kwargs)