Adaptive wavelet based phoneme recognition

The phoneme recognition systems are part of Automatic Speech Recognition (ASR) systems. These systems are used in telecommunication to provide automatic services such as "call collect", "directory assistance", etc. to the subscribers. The elimination of assistance of human operators saves operating costs for the telecommunication companies in millions of dollars. In this paper, a phoneme recognizer based on adaptive wavelets is described. The experimental results of this recognizer using TIMIT database is provided. This experimental results also include the comparative study with a Hidden Markov Model (HMM) based phoneme recognizer.