Frequency domain speech coding

Abstract From the Canadian patent

A sub-band coder for speech signals has a filter for dividing the frequency spectrum of speech signals input thereto into a plurality of sub-bands, from each of which is generated a respective series of samples of the filtered signal corresponding to that band. An encoder digitally encodes each series. A bit allocation unit varies the number of bits used for encoding the respective series in dependence on the relative energy contents of the sub-bands and is operative in accordance with a fixed set of n numbers, where n is the number of sub-bands to, allocate to the sub-band having the largest energy content the number of bits indicated by the largest number of the set and to allocate to …