HIGH-EFFICIENCY ENCODING METHOD.

摘要

A high-efficiency method for encoding data on the frequency axis obtained by dividing an input audio signal in a block unit and converting it to the frequency axis, wherein a band BVH having the highest center frequency among V (voiced) bands is sought when the number of change point of discrimination data of V (voiced sound)/UV (unvoiced sound) of all the bands on the on-frequency-axis data is judged greater than 1, the number NV of the V bands up to the band BVH is determined, whether or not their proportion is greater than a predetermined threshold value Nth is judged, and one V/UV division position is determined. In this way, the V/UV judgement data for each band is converted to the data of one division position inside all the bands, so that the data quantity and the bit rate can be reduced. The quantity of computation of the code book search is reduced by using two-stage hierarchical vector quantitization when the on-frequency-axis data is quantitized, and the memory capacity of the code book is reduced.