Aliasing Reduction for Gain Control with Critically Sampled Filter Banks

Abstract

Applications of audio playback and communication often require the possibility to shape the signal spectra. This can easily be implemented in conjunction with audio coding techniques which are based on a spectral decomposition. Here, this can be accomplished by simply multiplying the decoded spectral components with different gain factors. However, since the transforms or filter banks used in coding produce a critically sampled representation, this can lead to undesirable non-compensated aliasing components in the resulting output signal. The gain factor dependent aliasing compensation technique presented in this paper reduces these effects in a very efficient way. The remaining aliasing components are almost as low as for the approximation of a non-critically sampled complex filter bank, while the new technique is of lower computational complexity and introduces less delay.