A summation algorithm for MPEG-1 coded audio signals: a first step towards audio processing in the compressed domain

In this paper, we propose a new summation method for MPEG-1 encoded audio signals in the compressed domain. This method operates directly in the subband domain in order to reduce the delay and the implementation complexity. The main problem to be solved deals with bit allocation and subsequent quantization of the resulting summation signal. The bit allocation of encoded MPEG signals is based on the signal-to-mask ratios of the individual signals. To estimate the signal-to-mask ratios of the combined signal, the algorithm simply needs the information contained in the individual encoded frames. Along with a description of the proposed algorithm, variations and applications of the algorithm are detailed. Additionally, a performance evaluation of this method compared to direct summation in the time domain is given.