A new 1.2 kbit/s speech coding method based on a sinusoidal harmonic vocoder

Abstract

A very low bit rate speech coder at 1.2 kbit/s is proposed in which a speech signal is synthesized by using sinusoids whose frequencies are multiples of the fundamental frequency, and whose amplitudes are adaptively modulated in accordance with auditory perceptual characteristics, in order to improve subjective quality. The auditory perceptual characteristics are simulated by using Gammatone filters. The phases of the sinusoids are also controlled with reference to the subjective quality of the synthesized speech. The quality of the synthesized speech was improved by 0.45 in the Mean Opinion Score compared with that of a simple LPC vocoder operating at the same rate, and was comparable to that of a 2.4 kbit/s MELP coder. © 2000 Scripta Technica, Syst Comp Jpn, 31(14): 64–73, 2000