VoxCeleb1Identification

class torchaudio.datasets.VoxCeleb1Identification(root: Union[str, Path], subset: str = 'train', meta_url: str = 'https://www.robots.ox.ac.uk/~vgg/data/voxceleb/meta/iden_split.txt', download: bool = False)

VoxCeleb1 [Nagrani et al., 2017] dataset for speaker identification task.

Each data sample contains the waveform, sample rate, speaker id, and the file id.

Parameters:

• root (str or Path) – Path to the directory where the dataset is found or downloaded.

• subset (str, optional) – Subset of the dataset to use. Options: [“train”, “dev”, “test”]. (Default: "train")

• meta_url (str, optional) – The url of meta file that contains the list of subset labels and file paths. The format of each row is subset file_path". For example: ``1 id10006/nLEBBc9oIFs/00003.wav. 1, 2, 3 mean train, dev, and test subest, respectively. (Default: "https://www.robots.ox.ac.uk/~vgg/data/voxceleb/meta/iden_split.txt")

• download (bool, optional) – Whether to download the dataset if it is not found at root path. (Default: False).

Note

The file structure of VoxCeleb1Identification dataset is as follows:

└─ root/

└─ wav/

└─ speaker_id folders

Users who pre-downloaded the "vox1_dev_wav.zip" and "vox1_test_wav.zip" files need to move the extracted files into the same root directory.

__getitem__

VoxCeleb1Identification.__getitem__(n: int) → Tuple[Tensor, int, int, str]

Load the n-th sample from the dataset.

Parameters:

n (int) – The index of the sample to be loaded

Returns:

Tuple of the following items;

Tensor:

Waveform

int:

Sample rate

int:

Speaker ID

str:

File ID

get_metadata

VoxCeleb1Identification.get_metadata(n: int) → Tuple[str, int, int, str]

Get metadata for the n-th sample from the dataset. Returns filepath instead of waveform, but otherwise returns the same fields as __getitem__().

Parameters:

n (int) – The index of the sample

Returns:

Tuple of the following items;

str:

Path to audio

int:

Sample rate

int:

Speaker ID

str:

File ID