)
Telugu ASR Benchmark Dataset for Challenging Speech Recognition (Kathbath hard Telugu)
Telugu ASR (Automatic Speech Recognition) benchmark challenging speech dataset from Bhashini for supporting the development of robust regional speech recognition systems.
About Dataset
This is a Telugu ASR benchmark dataset specifically designed to evaluate and improve Automatic Speech Recognition (ASR) systems in challenging scenarios, particularly in the news and general domains. The "hard" variant of the dataset includes audio samples with complex linguistic patterns, background noise, or overlapping speech, making it an ideal resource for stress-testing ASR models. Submitted by AI4Bharat, it supports advancements in speech recognition technologies for regional languages, fostering the development of robust and accurate ASR systems.
Dataset Metadata
Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
NA
Sector Agnostic
Microsoft
Unstructured
NA
NA
N.A.
24/02/25 13:22:22
Open
Hosted
Activity Overview
0
17 
341.17 MB 
276
Tags
- NLP Dataset
- Benchmark
- News Domain
- Telugu
- General Domain
- Automatic Speech Recognition
- Speech Technology
- AI4Bharat
- ASR
- Regional Languages
- Audio Processing
- Hard Dataset
License Control
Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
844424930563393-1123-m.wav ( 263.45 KB )
To preview this file, you need to be a registered user. Please complete the registration process to gain access and continue viewing the content.
Data Quality Score Beta
Version Control
Version 1(341.17 MB)
- admin·1 year(s) ago
844424930563393-1123-m.wav
844424930563394-1123-m.wav
844424930563395-1123-m.wav
844424930563397-1123-m.wav
844424930563398-1123-m.wav
844424930563399-1123-m.wav
844424930563400-1123-m.wav
844424930563401-1123-m.wav
844424930563404-1123-m.wav
844424930563405-1123-m.wav
- more_horiz 1484 more
© 2026 - Copyright AIKosh. All rights reserved. This portal is developed by National e-Governance Division for AIKosh mission.