IndicVoices-R
IndicVoices-R is a large-scale, multilingual, multi-speaker speech dataset for Text-to-Speech (TTS) research in Indian languages.
About Dataset
IndicVoices-R is a comprehensive, multilingual speech dataset designed for Text-to-Speech (TTS) research, covering 22 Indian languages. It includes over 1,700 hours of high-quality, spontaneous speech from more than 10,000 speakers. The dataset is processed with advanced techniques to enhance speech clarity and remove background noise, making it ideal for training TTS models and evaluating speaker generalization. It supports zero-shot, few-shot, and many-shot evaluation metrics for robust TTS model development.
Dataset Metadata
Attribution-No Derivatives 4.0 International (CC BY-ND 4.0)
India
Sector Agnostic
IIT, Madras Team
-
Unstructured
Static
Static
N.A.
09/05/25 02:55:08
Open
Hosted
Activity Overview
0
38 
150 
0
Tags
- Multilingual
- AI4Bharat
- Indian Languages
- Machine Learning
- Text to Speech
- Speech Recognition
- Speech Synthesis
- natural language processing (NLP)
- linguistic diversity
License Control
Attribution-No Derivatives 4.0 International (CC BY-ND 4.0)
© 2026 - Copyright AIKosh. All rights reserved. This portal is developed by National e-Governance Division for AIKosh mission.