This thesis is concerned with the design of a speech recognition
system to recognize digits 1, 2, 3 and 4. The system was designed
by using the characteristics of the spectral patterns of amplitude vs.
time at discrete frequencies. Data obtained for digits 0 to 9 are presented.
The outputs...
This dissertation introduces a new idea for frequency
domain deconvolution of speech signals. The idea is to
approximate a unit sample by a Gaussian pulse. The
deconvolution is performed by designing a glottal inverse
filter that will estimate the excitation of the vocal tract.
Then, the excitation will be used...
This thesis investigates an implementation of speech recognition front-end. It is an application specific integrated circuit (ASIC) solution. A Mel Cepstrum algorithm is implemented for the feature extraction. We present a new mixed split-radix and radix-2 Fast Fourier Transform (FFT) algorithm, which can effectively minimize the number of complex multiplications...