QUOTE(manuelator @ Feb 4 2006, 07:26 PM)
But you know how i can decode a ogg vobis or LAME to 8 bits, the source it' s 8 bits, but the decode it's 16 (double size).
Well, you have to know that
every input, whatever resolution it has (8,16,24 or 32) is encoded into floating point with pretty much every lossy codec. The coded file (mp3,ogg,...) doesn't really have a bit resolution, because it may round something to save space. So how should the decoder know that the input file to the encoder was 8bit? It can't. And it doesn't make any sense to feed a lossy encoder with 8bit (except the original is already at that bit depth).
To decode to 8bit (for every lossy file), you have to set output resolution to 8bit in the decoder.