Development of online isolated digit recognizer using hidden Markov model

Abstract

The objective of this work is to develop an online isolated digit recognizer that takes small footprint and also runs faster than that using the conventional Hidden Markov Model (HTK) toolkit. Such system will find applications on devices with limited computing power and memory capacity. The different modules in the online digit recognizer are explained first. This is followed by the development of offline digit recognizer in MATLAB. Later the online digit recognizer is implemented in C language. The performance evaluation of the digit recognizer is made in both offline and online mode. The digit recognizer works well except for digit two and eight where the confusion is high. However, the online system works well in speaker independent scenario.