On Reducing the Complexity and Storage of CELP Speech Coding using Self-orthogonal Codebook

In this paper, we propose an algorithm to reduce the complexity and the storage requirement for CELP speech coding substantially. The new algorithm is based on the concept of designing a special Gaussian codebook so that the autocorrelations of each codeword with its shifting entities are orthogonal. Segmental SNR and informal listening both showed that the new algorithm did not have any perfomance degradation compared to the original CELP coding.