Olti Qirici*, Ridi Qirici
Faculty of Natural Sciences, University of Tirana
Abstract: Hereafter it will be shown a comparative approach toward the speech analysis, between different methods of characteristics extraction which would lead to a better understanding and simpler implementation of a automatic speech recognition system. Different methods are shown till now and all these methods try to rich in a better description of the signal information to be simple enough for implementation and big enough to really show the signal content. Hereby will be shown side by side some of these methods with a better view on the MFCC method which seems to be also the most used till know.
Keywords: MFCC, Cepstrum, frame, windowing, DFT, ASR, LFCC
