Efficient Stereo Bitrate Allocation for Fully Scalable Audio Codec

Joint stereo coding has been adopted to improve quality of compressed digital audio. Mid and side stereo coding technique is the most popular technique of stereo coding and it is used in MPEG-1 and MPEG-4. According to ISO 14496-3 published in the year2005, MPEG-4 scalable lossless coding is among the currect MPEG audio coding, which has the potential to integrate functions of perceptual audio coding to a single framework. Eradication of complication associated with the truncators and encoders of the SLS is the main objective of designing the efficient stereo bitrate allocation.

Structurally  MPEG-4 SLS RS comprises of the encoder, truncators, and a decoder. Rahardja, Koh and Licontributed significantly to the enhancement of SLS structure. Particularly Y. Kim and Y. Seo and S. Park  identified the MPEG-4 AAC as the key layer in a perpetual audio code. Besides the core layer, the SLS has a non-core mode in which production of scalable LLE bistreams takes place. A fully SLS can be achieved by using this mode. Transformation of the input audio losslessly in SLS RS encoder leads to generation of integer MDCT coefficient. To generate core bitstream, coefficients are introduced to core layer encoder.

According to R. Yu et. al incorporation of CBAC with BPGC increases the efficiency of BPGC to capture statistical dependencies of a data.  When  SLS is non-core a bit-plane coding is applied directly. Its accuracy depends greatly on present bitrate assigned to various channels.  In this scheme, high efficiency is experienced when M channel is of higher amplitude than the S channel.  By use of truncator, SLS bitstreams can be easily truncated. In an effort to develop efficient stereo bitrate allocation enhanced encoder and truncators have been used. To improve the accuracy recovery of the L/R channel in enhanced encoder more resources should be allocated to M channel. The result obtained by J.Johnston and A. Ferreira on perceptual allocation logarithms was inversely applicable in non-core. Adjusting of the bit allocation plays a significant role in improving operation of pure bit-plane coding. Bit allocation in SLS RS truncator is given by . For the enhanced truncators the   and  cases are considered respectively. In the secong case perceptual core bitstream was first allocated before proportional enhancement was done. To test  formances of an enhanced encoder and truncator two tests were carried out. The audio sequence in test two were losslessly encoded by a RS encoder. Results were measured using OPERA interms of ODG which ranges from -4 to zero.

From the two test carried out it was noticed that both the enhanced truncator and encoder worked perfectly in improving the quality of the first four sequence at different bitrates. Due to lower correction for L and R channels in the last two test an improvement was not obvious. The last two sequences are limited from the pro of M/S stereo coding.

For fully scalable and non-core SLS efficient stereo bitrate allocation algorithms can be implemented to increase their efficiency. Additionally, the knowledge of M/S is crucial in understanding and designing the stereo bitrate.

Though the article comprehensively analyzes the concept, it excludes some of the limitation associated with the conceptualization of the idea. Additionally, it does not portray some of the areas which require further study.  The author has provided summarized information thus understanding some technical section like enhancement of the truncators and encoders has been a challenge.