1.2 kbit/s harmonic coder using auditory filters

Abstract

In this paper, a very low bit speech coder at 1.2 kbps is newly proposed. Like the LPC vocoder, it only requires gain, pitch, and spectral information, but its quality is far superior. The synthesis method is one of harmonic coding, using sinusoids whose frequencies are multiples of the fundamental frequency, where the amplitudes of the sinusoids are adaptively modulated using gammatone filters as a perceptual weighting filter. The sinusoids' phases are also adjusted so as to maximize the perceptual quality. In order to reduce the total bit rate to 1.2 kbit/s, a new segment coder for spectral information (LSP coefficients) using DP matching is also proposed. The quality of the synthesized speech was improved by 0.45 in the mean opinion score (MOS) compared with that of the simple LPC vocoder operating at the same rate, and it was comparable to that of 2.4 kbit/s MELP coder.