Skip to content

Dataset · CSTR-Edinburgh

English speech corpus with diverse accents

CSTR-Edinburgh/vctk

Audio recordings of speech from 110 English-speaking speakers with various accents for speech synthesis and recognition tasks.

What's inside

The CSTR VCTK corpus contains English speech data read by 110 speakers with different accents. The size of the dataset ranges from 10 to 100 thousand records, and the dataset is intended for ASR, TTS, and text-to-audio tasks.

What you can do with it

  • Training and testing automatic speech recognition (ASR) systems on different English accents
  • Training multi-speaker speech synthesis (TTS) and voice cloning models
Tasks
Automatic speech recognition, Text → speech, Text → audio
Languages
English
Size
10K–100K rows
Downloads this month
4.7K
Likes
57

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("CSTR-Edinburgh/vctk")