Using super resolution pitch in waveform speech coders

Abstract

The author presents an efficient exploitation of a super resolution pitch estimate in waveform speech coders such as LPC (linear predictive coding), CELP (code-excited linear prediction), and TDHS. Despite the compression/quality gain afforded with using super resolution pitch, the computational resources required to interpolate the speech signal to a much finer resolution are significant and therefore practical use is limited to a rather moderate resolution increase (factor of ×2 is typical). One of the main contributions of this work is the ability to conserve computational resources by using a hybrid signal interpolation scheme composed of a second-order digital IIR (infinite impulse response) filter in tandem with a continuous linear interpolator which can be discretized to any desirable time resolution and thus take advantage of a super resolution pitch estimate.