Optimization and quality measurement for the layer 2 compression scheme of MPEG audio coding

Abstract

The layer 2 compression scheme of MPEG audio coding is a very common audio coding standard for CD quality audio (music). However, its encoding process requires a large amount of computations. In this paper, we propose some optimization techniques to make the encoding program much faster while allowing to impair the audio quality a little. In order to measure how much quality loss, we define criteria to quantify the quality loss rate of an encoded audio stream. From the definition, we can say that the encoder encoding streams with a quality loss rate equivalently wastes a number of bits, if the encoder that completely conforms to the layer 2 standard can save the same number of bits while keeping the same quality loss rate. In our experiments, we optimize the layer 2 compression programs such that audio streams can be encoded real time on a Pentium Pro 180 with only 0.0003% quality loss and equivalently 1.5% bits wasted and on a Pentium 180 with only 0.0009% quality loss and equivalently 4% bits wasted.