espnet2.train.dataset.label_loader

source

espnet2.train.dataset.label_loader(path)

Loads label data from a specified file and returns it in a structured format.

This function reads a label file and prepares it for further processing by returning an adapter that converts the raw data into a structured format, specifically designed for handling time-label pairs.

• Parameters:path (str) – The path to the label file to be loaded.
• Returns: An instance of AdapterForLabelScpReader that contains the loaded label data.
• Return type:AdapterForLabelScpReader

Examples

>>> label_data = label_loader('path/to/label_file.txt')
>>> sample_time, sample_label = label_data['utterance_id_A']
>>> print(sample_time)
[[0.0, 1.0], [1.0, 2.0]]
>>> print(sample_label)
['phone_1', 'phone_2']

NOTE

The label file should be formatted as follows: utterance_id_A start_1 end_1 phone_1 utterance_id_B start_2 end_2 phone_2 …