Adaptive transform coding (ATC) and Analysis-by-Synthesis (AbS) have been shown to provide high quality speech above 10 kb/s and below 8 kb/s respectively. The application of the discrete wavelet packet transform (DWPT) and the finite interscale basis coefficient sequences are investigated and evaluated for speech coding applications. A fully quantised DWPT based ATC is presented. Objective and subjective measurements prove the superior WT performance in the adaptive quantisation and bit allocation processes in comparison with other transform based ATC algorithms. Moreover, a hybrid Wavelet-Binary pulse excitation (WBPE) model for CELP speech coding is also presented. Results of a WBPE targeted at 6.3 kb/s are also compared with the conventional CELP coder at 8 kb/s using Gaussian, sparse and ternary codebook population.
Bibliographic reference. Ancin, F. J. / Larreategui, M. L. / Burrows, B. L. / Carrasco, R. A. (1995): "Wavelets for low bit rate speech coding applications", In EUROSPEECH-1995, 665-669.