Abstract:
Voice assistants are becoming more and more commonplace in real-world applications. Voice recognition technology is currently being incorporated into a diverse assortment of products, such as mobile applications and smart speakers found in consumers' homes. Additionally, voice assistants are rapidly turning into an important component of our day-to-day lives. A significant number of people who speak Bengali as their native language are illiterate, and thus have trouble using computers because the controls are in English. People who have trouble communicating in English may have an easier time using a computer or smartphone if they are able to give instructions in their native language of Bengali. A Bengali virtual assistant may be the solution to this problem. In this article, a Bengali virtual assistant known as "Saathi" is constructed. In order to understand the commands given in Bengali, "Saathi" makes use of the CNN model. The CNN employs a spectrogram to determine the nature of the orders and then carries out the corresponding responses.
