vctk / README.md
lhoestq's picture
lhoestq HF staff
Enrich card with number of hours and TTS use (#2)
bb50fbe
metadata
annotations_creators:
  - expert-generated
language_creators:
  - crowdsourced
language:
  - en
license:
  - cc-by-4.0
multilinguality:
  - monolingual
pretty_name: VCTK
size_categories:
  - 10K<n<100K
source_datasets:
  - original
task_categories:
  - automatic-speech-recognition
  - text-to-speech
  - text-to-audio
task_ids: []
paperswithcode_id: vctk
train-eval-index:
  - config: main
    task: automatic-speech-recognition
    task_id: speech_recognition
    splits:
      train_split: train
    col_mapping:
      file: path
      text: text
    metrics:
      - type: wer
        name: WER
      - type: cer
        name: CER
dataset_info:
  features:
    - name: speaker_id
      dtype: string
    - name: audio
      dtype:
        audio:
          sampling_rate: 48000
    - name: file
      dtype: string
    - name: text
      dtype: string
    - name: text_id
      dtype: string
    - name: age
      dtype: string
    - name: gender
      dtype: string
    - name: accent
      dtype: string
    - name: region
      dtype: string
    - name: comment
      dtype: string
  config_name: main
  splits:
    - name: train
      num_bytes: 40103111
      num_examples: 88156
  download_size: 11747302977
  dataset_size: 40103111

Dataset Card for VCTK

Table of Contents

Dataset Description

Dataset Summary

This CSTR VCTK Corpus includes around 44-hours of speech data uttered by 110 English speakers with various accents. Each speaker reads out about 400 sentences, which were selected from a newspaper, the rainbow passage and an elicitation paragraph used for the speech accent archive.

Supported Tasks

  • automatic-speech-recognition, speaker-identification: The dataset can be used to train a model for Automatic Speech Recognition (ASR). The model is presented with an audio file and asked to transcribe the audio file to written text. The most common evaluation metric is the word error rate (WER).
  • text-to-speech, text-to-audio: The dataset can also be used to train a model for Text-To-Speech (TTS).

Languages

[More Information Needed]

Dataset Structure

Data Instances

A data point comprises the path to the audio file, called file and its transcription, called text.

{
  'speaker_id': 'p225',
  'text_id': '001',
  'text': 'Please call Stella.',
  'age': '23',
  'gender': 'F',
  'accent': 'English',
  'region': 'Southern England',
  'file': '/datasets/downloads/extracted/8ed7dad05dfffdb552a3699777442af8e8ed11e656feb277f35bf9aea448f49e/wav48_silence_trimmed/p225/p225_001_mic1.flac',
  'audio':
    {
      'path': '/datasets/downloads/extracted/8ed7dad05dfffdb552a3699777442af8e8ed11e656feb277f35bf9aea448f49e/wav48_silence_trimmed/p225/p225_001_mic1.flac',
      'array': array([0.00485229, 0.00689697, 0.00619507, ..., 0.00811768, 0.00836182, 0.00854492], dtype=float32),
      'sampling_rate': 48000
    },
  'comment': ''
}

Each audio file is a single-channel FLAC with a sample rate of 48000 Hz.

Data Fields

Each row consists of the following fields:

  • speaker_id: Speaker ID
  • audio: Audio recording
  • file: Path to audio file
  • text: Text transcription of corresponding audio
  • text_id: Text ID
  • age: Speaker's age
  • gender: Speaker's gender
  • accent: Speaker's accent
  • region: Speaker's region, if annotation exists
  • comment: Miscellaneous comments, if any

Data Splits

The dataset has no predefined splits.

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

The dataset consists of people who have donated their voice online. You agree to not attempt to determine the identity of speakers in this dataset.

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

Public Domain, Creative Commons Attribution 4.0 International Public License (CC-BY-4.0)

Citation Information

@inproceedings{Veaux2017CSTRVC,
    title        = {CSTR VCTK Corpus: English Multi-speaker Corpus for CSTR Voice Cloning Toolkit},
    author       = {Christophe Veaux and Junichi Yamagishi and Kirsten MacDonald},
    year         = 2017
}

Contributions

Thanks to @jaketae for adding this dataset.