Does anyone else not find the fun in programming...?

[Dave Benjamin]

First of I don't use realtime...

I figured... =)

I create lists of notes::

class Note:

def __init__(self, time=0, pitch=64, velocity=64, duration=96):
self.time = time
self.pitch = pitch
self.velocity = velocity
self.duration = duration

def __str__(self):
r = []
a = r.append
a('time %s' % self.time)
a('pitch %s' % self.pitch)
a('velocity %s' % self.velocity)
a('duration %s' % self.duration)
return '\n'.join(r)

That are then converted into midi files by a very simple wrapper layer.

This simple structure makes it extremely simple to create
transformations on a list of notes. I considder each list a "part" like
you see it in Cubase/Logic.

The idea is then to create a personal library of transformations and
generators that expres your own musical style. I also have a few
routines for repeating/extending/sequencing these parts.

So, a data-flow style more or less... that seems to be a useful model for
music software. Have you ever played around with Max?

I import these midi files into software like Cubase, or Reason or Orion.
Where they drive either hardware or software synths.

I like to fiddle around with the sounds manually by twiddleling the knobs.

But I don't change the mnusic manually in the sequencer software. Rather
i change the software and genereate a new midi file, that I reload.

It is a bit like writing code generators. And it is completely flexible,
creative and fun due to the ease of Python.

Yeah, makes perfect sense to me. Do you have any Python-generated songs
available? Sounds really cool.

The thing I was working on was a probabilistic beat generator, based on
statistics of when a certain drum would hit in a beat for a selection of hip
hop songs. I was trying to capture the "feel" rather than the exact sequence
of sounds. I still have the (tcl) source laying around here somewhere. It
worked pretty well on Linux (I could actually get it to beat-match) but in
Windows the performance was terrible. Which is probably because I'd have to
be out of my mind to think I could get realtime performance out of Tcl.

Thanks, Max...

 

[Max M]

So, a data-flow style more or less... that seems to be a useful model for
music software. Have you ever played around with Max?

No. I tried c-sound, and while it was interesting, it was really
cumbersome at the level I am interested in.

Yeah, makes perfect sense to me. Do you have any Python-generated songs
available? Sounds really cool.

Nothing I will release. Mostly pre-studies.

The thing I was working on was a probabilistic beat generator, based on
statistics of when a certain drum would hit in a beat for a selection of hip
hop songs. I was trying to capture the "feel" rather than the exact sequence
of sounds.

That is a good way to go. Bayes networks for rythmical patterns.


regards Max M

 

[Dave Benjamin]

No. I tried c-sound, and while it was interesting, it was really
cumbersome at the level I am interested in.

Yeah, CSound is theoretically amazing but practically useless for me. But
I've had a few friends swear by it. I played around with Max a little bit in
an electronic studio class, and found it very interesting. It's the only
thing I've ever seen that was worthy of the name "graphical programming
languaeg". A language you could program in entirely with a mouse, and it was
actually useful. My teacher used it to do MIDI routing and later to create
an algorithmic music installation in a musem.

For the sake of keeping this on-topic, and because it actually sounds like a
neat idea, I think it'd be cool to develop something like Max (or jMax - is
that project still alive?) for Python. What's the best way of interfacing
Python with real-time MIDI?

Nothing I will release. Mostly pre-studies.
Hehe...


That is a good way to go. Bayes networks for rythmical patterns.

Interesting idea.