End-to-end ASR For Code-switched Hindi-english Speech

Abstract

End-to-end (E2E) models have been explored for large speech corpora and have been found to match or outperform traditional pipeline-based systems in some languages. However, most prior work on end-to-end models use speech corpora exceeding hundreds or thousands of hours. In this study, we explore end-to-end models for code-switched Hindi-English language with less than 50 hours of data. We utilize two specific measures to improve network performance in the low-resource setting, namely multi-task learning (MTL) and balancing the corpus to deal with the inherent class imbalance problem i.e. the skewed frequency distribution over graphemes. We compare the results of the proposed approaches with traditional, cascaded ASR systems. While the lack of data adversely affects the performance of end-to-end models, we see promising improvements with MTL and balancing the corpus.

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keysrivastava2019end

Related papers

• Investigating Target Set Reduction For End-to-end Speech Recognition Of Hindi-english Code-switching Data (2019)5.84
• Balanced End-to-end Monolingual Pre-training For Low-resourced Indic Languages Code-switching Speech Recognition (2021)0.00
• End-to-end Code-switching ASR For Low-resourced Language Pairs (2019)9.76
• Large-scale Multilingual Speech Recognition With A Streaming End-to-end Model (2019)14.97
• Towards End-to-end Code-switching Speech Recognition (2018)0.00
• On The End-to-end Solution To Mandarin-english Code-switching Speech Recognition (2018)12.10
• Constrained Output Embeddings For End-to-end Code-switching Speech Recognition With Only Monolingual Data (2019)7.16
• Costa: Code-switched Speech Translation Using Aligned Speech-text Interleaving (2024)0.00