The Kathbath-Hindi-Valid dataset is an essential Hindi ASR benchmark dataset curated to validate Automatic Speech Recognition (ASR) systems under general scenarios. It consists of 1684 hours of labeled speech data, covering 12 Indian languages, offering a diverse and comprehensive resource for researchers and developers. Submitted by Tahir Javed, this dataset plays a crucial role in enhancing the performance and reliability of speech recognition technologies for Hindi and other regional Indian languages.
© 2026 - Copyright AIKosh. All rights reserved.