A novel algorithm for low bit rate speech compression using a hybrid LP-harmonics model

Abstract

We present a new LP-harmonic speech codec. At the coder speech signal is pre-processed, and an LP analysis is performed, together with pitch estimation and voicing decision. At the decoder and when the frame is voiced, the encoded parameters are used to estimate the spectrum envelope, extract and classify the harmonics as either strong or weak depending on their relative distance from multiples of the fundamental frequency. Strong harmonics parameters are then used to generate pure sinusoids. While weak harmonics are used to generate a mixed signal of a pure sinusoid and a random-like signal. For unvoiced frames, the excitation of the LP filter is generated as a white noise signal. The proposed model allows for the mixing of strong and weak periodic signals together with random signals to produce an excitation input that results in natural speech. Informal testing of the coder working at 1.82 kb/s showed that the output speech has high intelligibility, with quality comparable to that of a 4 kb/s sinusoidal codec.