Improving Speech Recognition for Indic Languages using Language Model

Abstract: We study the effect of applying a LLM (LM) on the output of Automatic Speech Recognition (ASR) systems for Indic languages. We fine-tune wav2vec $2.0$ models for $18$ Indic languages and adjust the results with LLMs trained on text derived from a variety of sources. Our findings demonstrate that the average Character Error Rate (CER) decreases by over $28$ \% and the average Word Error Rate (WER) decreases by about $36$ \% after decoding with LM. We show that a large LM may not provide a substantial improvement as compared to a diverse one. We also demonstrate that high quality transcriptions can be obtained on domain-specific data without retraining the ASR model and show results on biomedical domain.