Phone-based filter parameter optimization for robust speech recognition using likelihood maximization

Abstract

Accuracy of speech recognition systems decreases when the distance between talker and microphone increases. By the using of microphone arrays and appropriate filtering of received signals, the accuracy of recognizer can be increased. Many different methods have been proposed. These methods can be classified in two main approaches: Systems that perform in two independent stages of array processing and recognition, and systems that use the likelihood acoustic information of recognition stage to calibrate the parameters of array processing stage on an utterance-based manner. In this paper a new approach to microphone array processing is proposed in which the parameters of array processing are adjusted based on phones used in language. Optimized filter parameters are stored and used during recognition phase. Persian language is used to find any improvement in speech recognition accuracy.