Multi-objective optimisation of wavelet features for phoneme recognition

VIGNOLO, L.D.; RUFINER, H.L.; MILONE, D.H.

Bs. As.

Simposio; XVII Simposio Argentino de Inteligencia Artificial - 45 Jornadas Argentinas de Informática; 2016

One of the most important issues in speech applications involves the pre-processing stage, which is meant to produce a manageable set of significant features, exploiting the capabilities of the classification phase. The most widely used features for speech recognition, and also applied for different tasks involving speech and music signals, are the mel-frequency cepstral coefficients (MFCCs). For the purpose of obtaining appropriate features for state of the art speech recognizers, a classifier based on hidden Markov models (HMM) is used to estimate the capability of candidate solutions, using on a set of English phonemes. The proposed method, which we refer to as evolutionary wavelet packets (EWP), exploits the benefits provided by multi-objective evolutionary optimisation in order to find a better speech representation.