I am a Computer Science student at Hermannstadt University (Sibiu), I have recently came across the implementation of the synthesis filterbank used in most of the xine-lib. I have also read the article regarding the Fast IMDCT written by Mr. Lee, and I still  miss the main idea of avoiding the computation as is described in the standard ( the shifting, matrixing, and building the U vector) while computing the Synthesis Filter (synth_filter.c file)  as it looks like it is using a different indexing . Any article or guidance on how the Synthesis Filter was improved and by avoiding the standard implementation and how this was performed, is highly appreciated.  

                                                  Thank you in advance!
                                                     Yours faithfully,
                                                Lazar Constantin-Ariton,