datasets/librispeech.py [19:61]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    }

    sample_rate = 16000

    def __init__(self, data_path, preprocessor, split, augment=False):
        augmentation = []
        if augment:
            augmentation = [
                torchaudio.transforms.FrequencyMasking(27, iid_masks=True),
                torchaudio.transforms.FrequencyMasking(27, iid_masks=True),
                torchaudio.transforms.TimeMasking(100, iid_masks=True),
                torchaudio.transforms.TimeMasking(100, iid_masks=True),
            ]

        super(Dataset, self).__init__(
            data_path,
            preprocessor,
            split,
            self.splits,
            augmentation=augmentation,
            sample_rate=self.sample_rate,
        )


if __name__ == "__main__":
    import argparse
    import torch

    parser = argparse.ArgumentParser(description="Compute data stats.")
    parser.add_argument("--data_path", type=str, help="Path to dataset JSON files.")
    parser.add_argument(
        "--save_text", type=str, help="Path to save parsed train text.", default=None
    )
    parser.add_argument(
        "--save_tokens", type=str, help="Path to save tokens.", default=None
    )
    parser.add_argument(
        "--compute_stats",
        action="store_true",
        help="Compute training data statistics.",
        default=False,
    )
    args = parser.parse_args()
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



datasets/wsj.py [19:61]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
    }

    sample_rate = 16000

    def __init__(self, data_path, preprocessor, split, augment=False):
        augmentation = []
        if augment:
            augmentation = [
                torchaudio.transforms.FrequencyMasking(27, iid_masks=True),
                torchaudio.transforms.FrequencyMasking(27, iid_masks=True),
                torchaudio.transforms.TimeMasking(100, iid_masks=True),
                torchaudio.transforms.TimeMasking(100, iid_masks=True),
            ]

        super(Dataset, self).__init__(
            data_path,
            preprocessor,
            split,
            self.splits,
            augmentation=augmentation,
            sample_rate=self.sample_rate,
        )


if __name__ == "__main__":
    import argparse
    import torch

    parser = argparse.ArgumentParser(description="Compute data stats.")
    parser.add_argument("--data_path", type=str, help="Path to dataset JSON files.")
    parser.add_argument(
        "--save_text", type=str, help="Path to save parsed train text.", default=None
    )
    parser.add_argument(
        "--save_tokens", type=str, help="Path to save tokens.", default=None
    )
    parser.add_argument(
        "--compute_stats",
        action="store_true",
        help="Compute training data statistics.",
        default=False,
    )
    args = parser.parse_args()
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -