speechtotext.datasets.Dataset

class Dataset(path_to_dir, name, file_ext='.wav')

Bases: DatasetBare

Class to extract data from the dataset folder.

Creates dataset object. There needs to be an transcripts.txt directly in the dir. :type path_to_dir: str :param path_to_dir: Path to dir ending with “/”. :type path_to_dir: str :type name: str :param name: Name of dataset. :type name: str :type file_ext: str :param file_ext: Extention of files. :type file_ext: str

Methods

get_n_samples	Get n random samples.
get_path_of_fragment	Gets path of fragment.
get_text_of_id	Get text of fragment id.
load_transcript	Loads transcript.
number_of_samples	Get number of samples in dataset.
validate_samples	Validate if samples have a corresponding file.

get_n_samples(number_of_samples)

Get n random samples.

Parameters:

number_of_samples (int) – Number of random samples.

Returns:

Dataset with the samples.

Return type:

SampleDataset

get_path_of_fragment(audio_id)

Gets path of fragment.

Parameters:

id (str) – Id of file.

Raises:

FileNotFoundError – If id doesn’t exist.

Returns:

Path to fragment.

Return type:

str

get_text_of_id(audio_id)

Get text of fragment id.

Parameters:

id (str) – Id of fragment.

Returns:

String of spoken text.

Return type:

str

load_transcript()

Loads transcript.

number_of_samples()

Get number of samples in dataset.

Returns:

Number of samples in dataset.

Return type:

int

validate_samples()

Validate if samples have a corresponding file.

Returns:

True if all samples have a file.

Return type:

bool