Pattern Recognition of Speech Signals Using Wavelet Transform and Artificial Intelligence

摘要

Feature extraction in speech processing is one of the main phases to develop speech processing applications. A large set of feature extraction methods is available to implement on speech processing approaches, however the decomposition through Wavelet packets is one of the most popular nowadays for its robustness. This paper describes the development and implementation of the WPD technique using speech samples of the utterances of /cero/ and /uno/. The characteristic coefficients that result of the WPD are entered in a pattern recognition based on neural networks to classify data and recognize between the uttered words. The results show a classification above 75%, which demonstrates the suitability of the method for recognition.