[Clam-devel] GSoC Ideas

[Pau Arumí]

On dc, 2008-03-26 at 01:20 +0100, Xavier Amatriain wrote:
> Hello Michael, and thanks for your interest in CLAM...
> 
> Overall I think your proposal sounds really interesting. I think we all 
> agree that, like in most OS projects,
> documentation is an area we need to work on in CLAM. I like the idea of 
> having sound examples in the
> documentation (written as wiki pages for sure) and linking to the 
> doxygen code. I also like the proposal
> of improving code documentation.
> 
> What I am not so sure is whether restricting this to only "musical 
> examples" is a good idea. If I were to
> promote a project like this I'd like to see it cover as much of CLAM as 
> possible maybe including audio
> effects and why not some MIR examples.
> 
> What do other mentors think?
> 
> Btw, although CLAM was formally connected to MTG for a long time, it is 
> not anymore but I am sure
> that was not the only reason you were interested in CLAM ;-)
> 
> Xavier
> 
> 
> Michael Chinen wrote:
> > Hello,
> >
> > I am interested in participating in gsoc through CLAM.  I have heard a
> > lot of good things about UPF MTG projects and the work you folks do
> > overlaps with my own research, so I think CLAM makes the best fit for
> > me out of all the GSoC groups.
> >
> > About me: I am a graduate student in electro-acoustic music at
> > Dartmouth College.  My undergraduate degrees are from the University
> > of Washington and are in computer science and musical composition.  I
> > have been involved in 3 sourceforge audio projects under the handle
> > mchinen.  I've worked for a few small companies on big c++/C#
> > projects, and one big bad one working on audio dsp. On the creative
> > side, I have worked in a game company to do sound effects.  The music
> > I write is experimental and algorithmic computer music in
> > supercollider or lisp.  However in the past few years all my music
> > (including mixing) has been done in C++ and cocoa.  This is why CLAM
> > is intersting to me.
> >
> > I wanted to ask for some feedback on a my idea.
> >
> > -C++ Documentation by Musical Code Samples/Tutorials
> >   While many parts of CLAM are useful for MIR, a lot of it can also be
> > used for music composition.  I propose to write well documented
> > musical pieces in C++ that demonstrate the C++ library aspect of CLAM.
> >  Having well documented samples at several levels of complexity is
> > very helpful for the beginning user, even if he or she is well versed
> > in C++.  It provides a starting point as well as shows off what the
> > library is capable of.
> >   A typical musical code sample will be a completely contained C++
> > program that links to CLAM and outputs a wav file of the composed
> > piece.  It will contain a readme that explains which parts of the
> > library are being used to compose this piece, and the code inside will
> > be well commented.  The music itself will focus on taking advantage of
> > a CLAM component in a clever way.
> >   The intent is to have many small and simple examples that show how
> > the individual CLAM components work, and also to have two or three
> > larger examples that are more complex and explore what is possible
> > with the CLAM framework.
> >
> > I have looked at CLAM and found it to be promising, but I have not yet
> > programmed in it.  To select how many examples that I would make, I
> > think I should discuss with the developers to find out which they
> > believe to be the most interesting.  Please note that the parts of
> > CLAM used in these musical examples don't have to be just synthesis or
> > effects - it could be about some analysis/feature extraction tool as
> > well - the musical output could be based on the analysis features of
> > some other musical input, for instance.
> >
> > While doing this project I can also make sure the CLAM headers are
> > well documented and understandable for those that wish to use CLAM as
> > a library.
> >
> >
> > If this kind of project is not something that fits gsoc, please let me
> > know and I will try to think of another idea.  I do some analysis
> > synthesis noiseband-based modeling using genetic algorithms in a
> > project called genesynth, and maybe some of these ideas could be
> > incorporated into CLAM.

Hi Michael, thanks for your interest in CLAM.

All these ideas fits in of course. Personally I'm more interested in
mentoring *coding* projects (there are more mentors though). 
I like this idea of incorporating genetic algorithms (and maybe making
use of CLAMs spectral data?) for sound synthesis.

Cheers,
Pau