| from .base import SequentialDataPipe | |
| from .common_pipes import LoadAudio, Phonemize, SetOutputKeys | |
| class Speech2PhonemePipe(SequentialDataPipe): | |
| """ | |
| each item in the input dataset should have: | |
| wav_path: str | |
| transcription: str | |
| """ | |
| def __init__(self): | |
| output_keys = dict( | |
| x="wav", | |
| x_len="wav_len", | |
| labels="phonemized_text", | |
| class_ids="tokenized_text", | |
| unique_name="id", | |
| ) | |
| super().__init__( | |
| LoadAudio(), | |
| Phonemize(), | |
| SetOutputKeys(output_keys=output_keys), | |
| ) | |