Subband coder study

This report gives a detailed treatment of the theory and implementation of a speech waveform coding technique known as subband coders whose operations include the partitioning of the input speech band into subbands and the quantization of subbands individually. Both the integer-band sampling and the quadrature mirror filtering (QMF) approaches are studied and the result indicates that the latter method yields much improved speech quality as compared to the former one. In addition to the design of bandpass filters and subband quantizers, practical aspects in the implementation of subband coders are considered and they include (1) a more efficient digital interpolation scheme, and (2) a high speed multiplier-accumulator.