A type of the split-band coding process, where the incoming audio signal is split into a number of independent frequency sub-bands, and the accuracy of the quantization is varied in each band according to the input signal energy. Critical regions of the audio spectrum can, therefore, be coded more accurately, with quantizing energy being biased toward the high-sensitivity, low-frequency regions. High-energy regions are also coded more accurately than in PCM techniques, yielding a lower coding noise platform. See transform coding.