[linux-audio-dev] Pitchshift/Timestretch project..

[Steve Harris]

On Tue, Apr 06, 2004 at 11:28:28 +0200, Florian Schmidt wrote:
> On Tue, 6 Apr 2004 07:33:50 +1000
> Erik de Castro Lopo <erikd-lad at mega-nerd.com> wrote:
> 
> > On Tue, 06 Apr 2004 00:59:39 +0200
> > Marek Peteraj <marpet at naex.sk> wrote:
> > 
> > > Hi Erik,
> > > 
> > > On Mon, 2004-04-05 at 22:25, Erik de Castro Lopo wrote:
> > > > 
> > > > There are other people working on something like this.
> > > 
> > > Could you reveal some details? :)
> > 
> > Well me. I've been working on this since the start of the year, but
> > been thinking about the problem for over 10 years. 
> 
> I've only been thinking about how this is done for very short periods of
> time. My naive approach to timestretching would be to transform the
> signal into the frequency domain [either by windowe fourier or by
> wavelet transform]. and then afterwards retransform, but with a changed
> time base. Actually i rather think of it as synthesizing the
> timestretched material from the frequency information..

Theres info about this on Stephan's website:
http://www.dspdimension.com/start.html

Its not that simple - but thats the basic idea.

- Steve