Function get_data_module

train.py:87–136

r"""Create data_module. Mini-batch data can be obtained by: code-block:: python data_module.setup() for batch_data_dict in data_module.train_dataloader(): print(batch_data_dict.keys()) break Args: workspace: str config_yaml: str

(
    config_yaml: str,
    num_workers: int,
    batch_size: int,
)

 
def get_data_module(
    config_yaml: str,
    num_workers: int,
    batch_size: int,
) -> DataModule:
    r"""Create data_module. Mini-batch data can be obtained by:

    code-block:: python

        data_module.setup()

        for batch_data_dict in data_module.train_dataloader():
            print(batch_data_dict.keys())
            break

    Args:
        workspace: str
        config_yaml: str
        num_workers: int, e.g., 0 for non-parallel and 8 for using cpu cores
            for preparing data in parallel
        distributed: bool

    Returns:
        data_module: DataModule
    """

    # read configurations
    configs = parse_yaml(config_yaml)
    sampling_rate = configs['data']['sampling_rate']
    segment_seconds = configs['data']['segment_seconds']
    
    # audio-text datasets
    datafiles = configs['data']['datafiles']
    
    # dataset
    dataset = AudioTextDataset(
        datafiles=datafiles, 
        sampling_rate=sampling_rate, 
        max_clip_len=segment_seconds,
    )
    
    
    # data module
    data_module = DataModule(
        train_dataset=dataset,
        num_workers=num_workers,
        batch_size=batch_size
    )

    return data_module


def train(args) -> NoReturn: