Acoustic Modelling of Speech Code

Contemporary processing of speech signals has relied primarily on statistical methods. Recent advances in modelling techniques allow sophisticated methodology to be applied to the characterization of short term as well as long term dependence inherent in the speech signals, thus making many specific tasks such as speech coding, enhancement and recognition ready to be accomplished. In this paper, we elaborate on a particular integrated framework that is consistent, theoretically rigorous, and practically effective in providing solutions to the speech modelling problem.