For my project, I need to break down parts of the Lame algorithm and run them from Matlab, to provide a learning toolbox where aspects of Mp3 can be visualised.

However, does anyone think this is possible? for example, can I take only the pyschoacoustic part and just run that? so far, it appears to me that in order for Lame to work, everything has to be compiled together...............................

Take a look at the mp3x frame analyzer hooks. Well, try mp3x. Maybe it already does want you want?

http://lame.sourceforge.net/screenshots/screenshots.html

ok, i just had a look. This implementation is probably something like what i need to do in Matlab, but on a simpler level. it was originally intended that I could produce something without having an in-depth knowledge of the Lame algorithm, but the web of complexity is looking very overwhelming.

If you could imagine Lame being written as a Matlab script and being able to access workspace variables and data structures from the algorithm at different stages, then thats what needs to be implemented on a simpler level.

You could perhaps modify Lame so that it output psychoacoustic results in a text file, and analyse/display results from Matlab.

Gabriel, that is an interesting idea, very interesting, I will definately look into that one..............but keeping your ideas coming will be greatly appreciated.

If you need a very basic overview:
http://gabriel.mp3-tech.org/lame/diagrams/

ok, thanks, but how do I open the files with the .dia extension?

With Dia.
But those are just the diagram files from which the .png are created.

Tripwire, are you saying that I print out an Mp3 data stream onto paper......and then somehow do something with it.......?

I am not sure want you mean by that....

Good old punched cards?

how about I pay someone to do my project????????????????

How much would it take?

only joking, im skint!

Gabriel....as a Lame developer, is there no way that I can extract the psycho.c code and use that as a stand-alone piece of code?

I mean, apart from fft.c, is the anything else that this code MUST #include in order to work?

I think that it would be way easier to try outputting results to an external text file.

To use psymodel.c you will need l3side.h andutil.h, which in turn needs machine.h and gain_analysis.h,...

This could be possible, but will be painfull.

unfortunately for my sanity, the painful route is what I need to take as the the text file method does not meet my project specs. Also this method is probably limited to pyscho.c, and similar things need to be done for the bit allocation loop.........