Optimizing high quality audio coding: advantages of full system observability

Perceptual audio coders rely on the efficient reduction of perceptually irrelevant components of the audio signal as well as on the removal of statistical signal redundancies to achieve good coding gains. In order to reach high compression ratios without reducing the subjective quality of the encoded audio signal, it is necessary to identify critically interdependent functional units of the encoding algorithm and to jointly optimize their performance. A flexible and interactive simulation and analysis environment has been programmed to assist the development and optimization of a new perceptual audio coder. The main features of this environment will be explained and the most relevant aspects that were found to limit the encoding performance will be presented