Advancements and Challenges in Automatic Speech Recognition for Low-Resource Languages: A Review with a Focus on Assamese Brajavali

This review paper discusses various difficulties faced by different low-resource languages like Assamese Brajavali when recognition is performed by automatic speech recognition (ASR) techniques. There are many such languages in the world, and commonly it is seen that recognition is a challenging task for this type of language. The article discusses issues raised globally and for Indian low-resource languages when developing any ASR system. The corpus or sample collection is the performance-determining factor for any ASR, and authors have discussed how volume and different algorithms of artificial intelligence influence the efficiency of the models used. The paper mainly focuses on proposing a fruitful automated neural network-based model for the Assamese Brajavali language, which is based on Maithili but looks like Bengali. Research gap as well as technical gaps is also discussed in this paper with respect to the automatic recognition of other low-resource languages.