[linux-audio-dev] Developing a music editor/sequencer

[Kevin Dahan]

Le lundi 31 janvier 2005 à 09:55 +0100, Frank Barknecht a écrit : 
> Hallo,
> Dave Robillard hat gesagt: // Dave Robillard wrote:
> 
> > Abstraction is good - it makes improvement easy.  The core of your
> > engine shouldn't even know what "audio" or "MIDI" is.
> 
> Maybe such an engine shouldn't even know what a "track" is? I'm
> serious: For some years now I use a sequencing software, which doesn't
> have the concept of a "track" (Pd). All that's needed is "time" and
> maybe not even that.

That's something I have worked on for my (near defended) PhD thesis. It
seems that more and more composers (both classical - electroacoustic and
pop oriented) are thinking of new ways of putting up musical ideas. 
The rigidity of the current 'track-based' approach (while being an
useful representation mechanism) is very difficult to change. 
We made back in 2002 an experiment at getting rid of all the constraints
normally used by computer music software (tracks, strong audio/MIDI
typing, flexible time representation), which was quite interesting -
even if the prototype was ill- and fast-coded ;)

In fact I was intending to work on a software with less typing
orientation (i.e. no difference in the treatment of audio, MIDI or
whatever) and new visual paradigms... When I'll have more free time...

That said, I completely agree with Frank: the track concept is *very*
restricting when doing composition - you want to be able to define
relationships between objects not only by defining time links but
perhaps logical, statistical, etc... (that's why I use non-visual
software à la Csound, SC3), hence the notion of 'time' becomes much less
important.

Just my 0.02€... :)

Kevin 'Unet' Dahan