|Ph.D. thesis:||Speech coding by contourizing an ear-adapted spectrogram
application to data reduction, in german.
Technical University of Munich, 1997. Published 1998 as
VDI ser. 10 no. 522 at VDI-Verlag,
Contours bearing relevant information correspond, among other, to part tones perceptible by the ear. This work treats audio representations based on contours, defined as `ridges' of an ear-adapted spectrogram. Starting off from a known representation additional ridges and a new signal reconstruction procedure are introduced. A classification of ridges allows to separate tonal from noise-like signal components. Applying these foundations, speech codecs at data rates down to 4 kbps are presented.
Keywords: speech coding - spectrogram - short time Fourier spectrum - time contours - tonal/noise separation- part tone time pattern - sinusoidal representations - Fourier-t-transform - wavelets - phase reconstruction
Availiable online resources (small font: not translated from german yet):
Mail: Emanuelstr. 27
Phone: +49 - 89 - 300 44 70
$Date: 2015/07/27 21:12:13 $