Hi,
Why are we doing the frequency inversion ie multiplying every even indexed sample by -1, for only even indexed subband outputs. When we do the mdct of first subband output the frequency components will be split into the first subband frequencies right ie in the 0-689 Hz frequncy range. But for second subband this will range from 690 to 689*2. So why the frequency inversion is being done only for even indexed subbands?

Advance thanks for ur reply...

Because the subbands are overlapping. So you invert the samples in the even subbands in order to cancel them out with the indentical samples that exist in odd subbands in overlapped regions