espnet2.train.preprocessor.AbsPreprocessor

source

class espnet2.train.preprocessor.AbsPreprocessor(train: bool)

Bases: ABC

Abstract base class for audio data preprocessing.

This class serves as a blueprint for implementing various types of preprocessing tasks for audio data, such as speech enhancement, speech recognition, and others. It provides the structure for initializing preprocessors and defining the processing method.

train

A flag indicating whether the preprocessor is being used in training mode.

• Type: bool

• Parameters:train (bool) – Indicates if the preprocessor is used for training.

__call__(uid

str, data: Dict[str, Union[str, np.ndarray]]) -> Dict[str, np.ndarray]: Abstract method that processes the input data and returns the preprocessed data.