Hi,


I noticed that the MDCT / IMDCT, being a Lapped Ortogonal Transform, needs to be be 50% overlapped with the previous frame in order to cancel our the non-diagonal terms which contributes to aliasing artifacts.. whereas, the Lapped Biortogonal Transfrom doesn't have the non-diagonal terms..

I am wondering, if the frames aren't evenly quantized, would the LOT failed to cancel out the aliasing terms and thus lead to a greater degradation of audio quality?

What about the LBT ? Is it used in audio coding at all ??



wkwai