An Overview of Bengali Speech Recognition: Methods, Challenges, and Future Direction

In the subject of human-computer interactions, speech recognition is an appealing technique that gives users the opportunity to interact with and control the machine. Currently, automatic speech recognition (ASR) systems are being utilized to flawlessly convert speech to text. The implementation of ASR systems in Bengali has not yet achieved an acceptable standard, despite the fact that they are being used extensively in other languages. So far, various ASR models have been implemented for speech recognition like LSTM (Long Short-Term Memory), Transformer-based models like RNN (Recurrent Neural Network) and CNN (Convolutional Neural Network) are also quite popular for speech recognition. The Bengali language is more grammatically and structurally diverse than English. Therefore, it is difficult for researchers to use the same language model as English or any other language. So, the Bengali language is difficult to work with. Different studies have been carried out on Bengali speech recognition. We want to enlist the numerous models that have been used for Bengali speech recognition from 2009 to 2022. In this paper, we will discuss the challenges that were faced and the scope of future research in this field. This survey paper also provides datasets utilized in numerous research studies.