1000-hr Gigaspeech corpus

The 1000-hr Gigaspeech corpus is a large dataset used to evaluate automatic speech recognition (ASR) systems, specifically focusing on the effectiveness of contextual representations in improving performance.

Papers using 1000-hr Gigaspeech corpus (2)

Towards Effective and Compact Contextual Representation for Conformer Transducer Speech Recognition Systems2023Exploring SSL Discrete Speech Features for Zipformer-based Contextual ASR2024