TEXT INDEPENDENT SPEAKER RECOGNITION SYSTEM USING WAVELET FILTERS

This paper depicts a new speech feature extraction technique for use in automatic speaker recognition (ASR). Wavelets have been shown to be successful front end processors for speaker recognition. Front-end or feature extractor is the first component in an automatic speaker recognition system. Feature extraction transforms the raw speech signal into a compact but effective representation that is more stable and discriminative than the original signal. Since the front-end is the first component in the chain, the quality of the later components (speaker modeling and pattern matching) is strongly determined by the quality of the front-end. In other words, classification can be at most as accurate as the features. Wavelet Transform coefficients can be utilized to feature parameters in various forms. An appropriate transformation base is also important for the feature extraction. To use the coefficients directly as the feature is the simplest way to exploit the wavelet transform characteristics. We thus introduce a simple feature extraction model based on the result of DWT. In order to parameterize the speech signal, we should first decompose the signal in the dyadic form using the Mallat algorithm Speaker Recognition System in text independent mode has been developed and tested on set of 100, 50 and 25 speakers. The performance of system was tested with 2 sec, 3 sec and 5 sec speech samples which were different from the text used for training (i.e., text independent mode) respectively. A speaker was said to be identified if the Euclidean distance is minimum for VQ.