AGROASSAM: A Web Based Assamese Speech Recognition Application for Retrieving Agricultural Commodity Price and Weather Information

This paper presents a speech-based web application developed for retrieving the price of agricultural commodities and weather related information in Assamese language. The price of agricultural commodities are retrieved from AGMARKNET website while the weather related informations are extracted from the IMD website. Both these websites are updated on a daily basis by the Government of India. The back-end of the application consists of automatic speech recognition (ASR) modules developed using state-of-the-art acoustic modeling approaches. Word error rates (WERs) of 7.79% and 4.98% are achieved for commodity and district names respectively.