[LAU] [LAA] Qtractor 0.1.3 (frugal damsel) released!

Mark Knecht markknecht at gmail.com
Sat May 3 14:04:34 EDT 2008


On Sat, May 3, 2008 at 7:12 AM, Rui Nuno Capela <rncbc at rncbc.org> wrote:
>
> Pieter Palmers wrote:
>
> > Rui Nuno Capela wrote:
> >
> > > Mark Knecht wrote:
> > >
> > > > On Fri, May 2, 2008 at 12:21 PM, Rui Nuno Capela <rncbc at rncbc.org>
> wrote:
> > > >
> > >
> >
> > [...]
> >
> >
> > >
> > > > As for edge dragging, in Acid this allows you to play a portion of a
> > > > loop. If I place a loop in a session the tempo is adjusted
> > > > automatically and I hear the whole loop. If I drag the left edge to
> > > > the right I lose the front part of the loop. If I drag the right edge
> > > > to the left I lose the last part of the loop. There are ways to drag
> > > > the loop to an out-of-sync point if you want but I don't use it much,
> > > > really only for beat slicing where they have better tools in the
> > > > editor.
> > > >
> > > >
> > >
> > > when you import (or drop) a new audio file into the qtractor time-line,
> it's always assumed that the new clip is in tempo with the current session.
> if you find it is not the case, you can trim and/or (time-)stretch the clip
> until you see fit. take a tour:
> > >
> >
> > [...]
> >
> > As far as I can remember, the "automagic-ness" of ACID relies on some sort
> of extra metadata. I think it was embedded in the WAV file, but it might
> have been somewhere else.
> >
> > The point is, if there is such metadata in a clip (e.g. wav), why no use
> it? Have an option like "strech clip according to tempo metadata" or the
> likes. If I remember your LAC presentation correctly, all resampling is done
> "on-the-fly" (modulo some cacheing) so isn't is simply initializing the
> 'stretch factor' to a proper value?
> >
> > 2cents for the masses,
> >
> >
>
>  that's a good point indeed. i do remember that Mark has aslo referred to me
> about the meta-data acid do stuff into wavs, but i'm afraid i've lost the
> hint at the bottomless qtractor todo list ;)
>
>  question that occurs to me atm. might be: does libsndfile have the means to
> read or access that meta-data ?
>
>  as said, implicitly, i don't nor ever used acid, but it might be of some
> help if someone would hand out one of the acid-stuffed wavs for me to try?
>
>  cheers
>  --
>  rncbc aka Rui Nuno Capela
>  rncbc at rncbc.org
>

Short answer, yes, sort of...

I don't remember what app it is and I'm not in Linux at the moment but
I think it might be sndfileinfo has the ability to read a lot of the
meta-data. Problem was whomever I worked on this with 5 years ago (I
think it was Erik in those days but please don't let my bad memory
tarnish his reputation) ran out of time/interest and only got part of
the work done. Fortunately what he did get done was apparently
accurate so at least it's a good start. I do see that libsndfile lists
reading loop data as added in Nov., 2004 but I know I worked on it
earlier than that.

Anyway, when you find the right app and you download a few 'Acidized'
loops from the ACIDPlanet site for free you'll find that the info
tells you not only tempo & key but it also gives the author, library
name and probably most interesting, trimming information. Apparently
many Acid loops have a bit of a leader and trailer in the file. It's
my suspicion that the loop author suggests where to start and stop
playing, thus synchronizing the loop's natural beats with the measure
bars in the session. A few samples before and after thus allow for
some automatic fade-in/fade-out, etc., without everything being so
abrupt. That's all guessing on my part though.

I think the important issue here is that not ever Acid loop has
exactly the right number of samples to equal tempo. Doing some
calculations:

60 BPM is 1 beat/sec. Assuming a 4/4 time signature and a 1 measure
loop then the loop's sound be 4 seconds long. 4 seconds of 44.1KHz
audio would suggest 176,400 16-bit samples. (Per channel) When you
look at the actual length it may be 178,000 or 180,000 with trim
information to suggest where to play from and to.

The issue (I think) is that if you just played the whole loop, from
first sample to last, then the playback wouldn't be identical to the
way the loop author hears it on a system that honors this start/end
data. Close, but not identical.

Hope this helps, and again, all credit to those who build these wonderful toys.

Cheers,
Mark



More information about the Linux-audio-user mailing list