Spectral characteristics of speech are known to be particularly useful in describing a speech signal such that it can be efficiently reconstructed after coding or identified for recognition. The wavelets are considered one of such efficient methods for representing the spectrum of speech signals. Wavelets are used to model both production and perception processes of speech. Wavelet-based features prove a success in a widespread area of practical applications in speech-processing realm.