On a low bit rate speech coder using multi-level amplitude algebraic method
Chan-Joong Jung, Young‐Ho Park, Myung-Jin Bae · 2003
This paper analyzes the dynamic sparse algebraic codebook used to model a residual signal for ACELP. It proposes an improved algebraic codebook structure with a new searching process. The proposed algorithm improves the disadvantage of an algebraic codebook without computation increment. First, this paper makes it possible to select various pulse amplitudes differently from the conventional method which looks up the sign bit simply. In addition, two pulses are selected on the same track and the delay of the LP is minimized at the transient period when the LSF interpolation is performed using a transient detector from unvoiced to voiced speech without additional computation. For telephone line speech quality, a 5.6 kbps speech coder using the proposed algorithm was equivalent to the 6.3 kbps MP-MLQ from the viewpoint of subjective speech quality. For objective quality, however, the degradation was MNRU Q=15 dB compared to MP-MLQ where MNRU Q=15 dB.