parakeet.datasets package

Submodules

parakeet.datasets.common module

class parakeet.datasets.common.AudioDataset(file_paths: List[pathlib.Path], sample_rate: int, top_db: float = 60)

Bases: paddle.fluid.dataloader.dataset.Dataset

A simple dataset adaptor for the audio files. Read -> trim silence -> normalize

class parakeet.datasets.common.AudioFolderDataset(root, sample_rate, top_db=60, extension='.wav')

Bases: parakeet.datasets.common.AudioDataset

class parakeet.datasets.common.AudioSegmentDataset(file_paths: List[pathlib.Path], sample_rate: int, length: int, top_db: float)

Bases: paddle.fluid.dataloader.dataset.Dataset

A simple dataset adaptor for audio files to train vocoders. Read -> trim silence -> normalize -> extract a segment

parakeet.datasets.ljspeech module

class parakeet.datasets.ljspeech.LJSpeechMetaData(root)

Bases: paddle.fluid.dataloader.dataset.Dataset

Module contents