Accurate representation of time-varying signals using mixed transforms with applications to speech

摘要

Basis functions from two transforms, the discrete Fourier transform (DFT) and the Walsh-Hadamard transform are used. An iterative algorithm to implement the mixed representation is developed. A finite number of DFT components, either fixed or variable, is used to represent the selective (narrowband) dominant spectrum of the signal on a frame-by-frame basis. The whitened (broadband) residual signal is represented using a small number of Walsh components. The mixed partial sets are applied to speech representation. Preliminary results and informal listening using this model produced good quality speech.