I'm just finishing off my project report for a 3rd year BSc project (in Matlab building something based on the MP3 encoder)

What exactly is put into the MP3 bitstream? So you have the output from the MDCT (36 bins for each of the 32 bands). These are then quantized and etc according to the PsychoAcoustic model. Then you have the header for each frame, and inside the data part, what goes in? simply the information for each band?
so lets say the bits allocated to the first 4 bands are
1: 16
2:12
3: 0
4: 5
you would get 16 bits for band 1, 12 for band 2, NOTHING would be there for band 3, then 5 bits for band 4. Would the huffman encoding sort this out so that no further information is needed to say "16 bits have been used for band 1", and so the decoder realises that no bits have been allocated for band 3?

If anyone could help me out this you would very much help my project report as its SUCH an important part of the encoding!

Thanks,
Dan Gent

After 65 views no reply. Maybe my question didn't make sense or it was so obvious people didn't want to explain it again!
If anyone even knows of any good papers/websites they could refer me to it would be helpful! I don't want to hand in a dissertation with some glaringly obvious fault with my analysis of MP3 compression ;^)

ISO/IEC 11172 Part 3

Cheers,
SG

absolute hero SebastianG!
I think I might put your username on my "acknowledgements" page of my dissertation.
if only i'd Had this paper a month ago! :^)