Skip to content

Dataset · RidheshBhati

100,000-hour multilingual speech corpus

RidheshBhati/Complete_Data_Source_100K_HOURS

A collection of audio recordings and texts in 6 languages, including English, Hindi, Bengali, and Tamil.

What's inside

The repository collects between 1 and 10 million records (claimed 100 thousand hours of audio) in the form of high-quality audio waves and text transcripts. The data spans six languages: English, Hindi, Bengali, Assamese, Kannada, and Tamil, divided into corresponding language subsets.

What you can do with it

  • Training and fine-tuning automatic speech recognition (ASR) models for Indian languages and English
  • Developing multilingual text-to-speech (TTS) systems
Languages
Assamese, Bangla, English, Hindi, Kannada, Tamil, Telugu, Gujarati and 4 more
Size
1M–10M rows
Downloads this month
23.1K
Likes
4

How to load it

from datasets import load_dataset

ds = load_dataset("RidheshBhati/Complete_Data_Source_100K_HOURS")