Момчеджиков, Михаил Благоев; Momchedzhikov, Mihailo lagoev; Попов, Румен Н.; Popov, Rumen N.,
(2003)
In this paper are reviewed the disadvantages of currently most ed methods for spectral analysis in Speech Recognition — bank of filters, ear Predictive Coding and Fast Fourier Transform. Also, Frequency Warped st Fourier Transform is analysed and a new method, based on Wavelet nsform, is proposed, which surmounts to a great extent these disadvantages.