Division of Research DIU

Scopus Indexed Publications

Author: S.M. Saiful Islam Badhon, Farea Rehnuma Rupon, Md. Habibur Rahaman, Sheikh Abujar,

Abstract: Do we want AI as our future working partner or as our personal assistant? if the answer is yes than we must think about the communication between human and machines. Undoubtedly most comfortable communication media is the verbal communication. When it's about human speech it goes to Natural Language processing (NLP) and there are already many advanced works done in this sector specially in English. And some renown big companies already made some operating system-based assistant and robots those or really working near to human, but the concern is almost everything is very rich in English, English is international language but there are some popular languages as well where we need to focus, Bangla is one of them. Bangla is world's 8 ^th most popular language almost 163 million people talk in this language which is not ignorable number at all [1]. There are very few quality works in this language even big companies are not that much advanced in Bangla like English or other languages. Here we muster some works in Bangla where researchers tried to solve Bangla speech recognition problems by their own methodologies. We tried to gather most accurate and recent work in this sector.

Keywords: Speech to text , Bengali voice recognition , MFCCs , bangla speech dataset

Journal or Conference Name: 11th International Conference on Computing, Communication and Networking Technologies, ICCCNT 2020