espnet2.fileio.rand_gen_dataset.IntRandomGenerateDataset

source

class espnet2.fileio.rand_gen_dataset.IntRandomGenerateDataset(shape_file: Path | str, low: int, high: int = None, dtype: str | dtype = 'int64', loader_type: str = 'csv_int')

Bases: Mapping

Generate integer arrays from a shape definition file.

This class generates random integer arrays based on the shapes defined in a specified text file. The text file should list utterances and their respective shapes in a comma-separated format. The generated integers will be within the specified range defined by low and high.

low

The lower bound for the random integers.

• Type: int

high

The upper bound for the random integers.

• Type: int

dtype

The data type of the generated integers.

• Type: np.dtype

utt2shape

A mapping from utterance identifiers to their shapes.

• Type: dict

• Parameters:

  • shape_file (Union *[*Path , str ]) – The path to the shape definition file.
  • low (int) – The minimum value of the random integers.
  • high (int , optional) – The maximum value of the random integers. If not specified, defaults to None.
  • dtype (Union *[*str , np.dtype ] , optional) – The data type of the output integers. Defaults to “int64”.
  • loader_type (str , optional) – The method used to load the shape file. Defaults to “csv_int”.

• Returns: A random integer array of the specified shape.

• Return type: np.ndarray

Examples

shape.txt uttA 123,83 uttB 34,83

>>> dataset = IntRandomGenerateDataset("shape.txt", low=0, high=10)
>>> array = dataset["uttA"]
>>> assert array.shape == (123, 83)
>>> array = dataset["uttB"]
>>> assert array.shape == (34, 83)

NOTE

The high parameter must be greater than low if specified.

• Raises:
  • ValueError – If high is not specified and low is greater than or equal
  • to high. –