Abstract: When determining a quantization for digital audio signals having a spectral and temporal structure wherein the audio signals are buffered in frames and decomposed into spectral components, the method is: obtaining a variance and/or a representative of the components of the the frame within a defined frequency interval; computing a necessary bandwidth using psychoacoustic criteria; determining an initial quantization for each frequency interval using an approximate mathematical model; and iteratively increasing the quantization of each of the frequency intervals through a psychoacoustically weighted inverse relationship with the the variance of each of the frequency intervals.

Abstract: Provided is a dynamic bit allocation apparatus and method for audio coding which can be used widely for almost all digital audio compression systems and besides implemented simply with low cost. The bit allocation apparatus and method perform a very efficient bit allocation process, paying attention to a psychoacoustics behavior of the human audio characteristics with a simplified simultaneous masking model. In this process, peak energies of units in frequency divisional bands are computed, and a masking effect that is a minimum audio limit with the use of a simplified simultaneous masking effect model is computed and set as an absolute threshold for each unit. Then, a signal-to-mask ratio of each unit is computed, and then, based on this, an efficient dynamic bit allocation is performed.