Comparison of filterbanks for high quality audio coding

In the last years perceptual coding techniques have been applied successfully to high quality coding of digital audio signals. The basic perceptual coding system uses an analysis/synthesis system to map the time domain data into a number of frequency domain channels. A perceptual model is used to estimate the amount of noise shaping needed in order to avoiud any audible noise due to the quantization of the frequency domain data. Different filterbanks havbe been applied to perceptual coding. A comparison of different filterbanks shows that there is no performance penalty for hybrid filterbanks compared to other solutions.