You still have to encode the length. Asymptotically, this will cost you O(log_2 log_2 N) when you write N in binary.
Practically, if N is probably small, you may want to just use base one to represent the length (thus 1111110 101010 for an overhead of 7 bits = 1+ceil(log_2 N)). For larger values of N are likely, then either recursing ( 1110 110 101010 to encode the fact that the length is 3 bits long, the length and then the data for an overhead of
2 ceil(log_2 log_2 N) + 1
Recursing further only helps if very large values of N are likely and doesn't really help the asymptotic result, but only reduces the constant.
See Williams and Zobel's paper for more practical information, Greg Chaitin's, work for more theory.
The upcoming revision to MP3 should answer these quality concerns and include backward and forward compatibility. Stuff encoded with the new methods will still play on older decoders, but won't sound terribly good. Stuff encoded with old encoders will sound just as good on the new decoder. And the real kicker is that stuff encoded with the new encoder and played with the new decoder sounds fabulous at very low bit-rates.
Slashdot Mirror
You still have to encode the length. Asymptotically, this will cost you O(log_2 log_2 N) when you write N in binary.
Practically, if N is probably small, you may want to just use base one to represent the length (thus 1111110 101010 for an overhead of 7 bits = 1+ceil(log_2 N)). For larger values of N are likely, then either recursing ( 1110 110 101010 to encode the fact that the length is 3 bits long, the length and then the data for an overhead of
2 ceil(log_2 log_2 N) + 1
Recursing further only helps if very large values of N are likely and doesn't really help the asymptotic result, but only reduces the constant.
See Williams and Zobel's paper for more practical information, Greg Chaitin's, work for more theory.
http://www1.oup.co.uk/computer_journal/hdb/Volu
http://www.cs.auckland.ac.nz/CDMTCS/chaitin/
Justin Zobel's home page appears unavailable at the moment, unfortunately, but here it is.
http://www.cs.rmit.edu.au/~jz
You mean DMCA. You probably wouldn't know what it stands for, so it isn't surprising that you got it wrong.
The upcoming revision to MP3 should answer these quality concerns and include backward and forward compatibility. Stuff encoded with the new methods will still play on older decoders, but won't sound terribly good. Stuff encoded with old encoders will sound just as good on the new decoder. And the real kicker is that stuff encoded with the new encoder and played with the new decoder sounds fabulous at very low bit-rates.