The aim of the project was to develop new or modify existing Speech Recognition algorithm to improve the accuracy for Hindi Language. We made use of MFCC and DTW to create a limited domain automated system. A total of 250 audio samples were used training and 50 audio samples for testing. The final product was on RaspberryPi 3 as a standalone application making use of several audio hardware’s implementations with the Pi.
Working under trained professionals from Centre for Development of Advanced Computing.
Date: July 2016
Category: Digital Speech Processing