Fractal excitation signals for CELP speech coders

A class of random signals is presented as a new excitation codebook for stochastic predictive speech coders. These signals, known as fractional noises, depend on a single parameter, 0<H<1, that controls the low- or high-pass trend of their power spectrum. Preliminary experiments have shown their performance improves when H is limited to a subinterval of (0,1) and becomes better than that of the standard Gaussian codebook as the codebook bit rate decreases from 0.25 down to 0.1 b per excitation sample. Their parametric nature allows efficient coding by quantizing H. Further, with no search of the codebook but at a certain loss of speech quality, a suboptimum excitation can be estimated with an H that matches the fractal characteristics of the speech residual