[linux-audio-user] extreme time stretching

[Steve Harris]

On Mon, Mar 15, 2004 at 05:13:13 -0500, Eric Dantan Rzewnicki wrote:
> On Mon, Mar 15, 2004 at 09:48:05PM +0000, Steve Harris wrote:
> > On Mon, Mar 15, 2004 at 10:21:00 +0100, Frank Barknecht wrote:
> > > Hallo,
> > > Erik de Castro Lopo hat gesagt: // Erik de Castro Lopo wrote:
> > > 
> > > > On Mon, 15 Mar 2004 15:21:01 -0500
> > > > Eric Dantan Rzewnicki <rzewnickie at rfa.org> wrote:
> > > > > For the purposes of time stretching or
> > > > > compressing sound, how does it compare to the way Erik's SRC or
> > > > > SoundTouch work? 
> > > >
> > > > SoundTouch is a true pithc independant time stretcher.
> > > 
> > > I think, Soundtouch uses a Fourier-based approach. Granular synthesis
> > > isn't really the best method of clean time stretching, I just
> > > mentioned it as one example of a technique, that can be used to do
> > > extreme streches. 
> > 
> > Soundtouch uses WSOLA, which is similar to granular synthesis, its not 
> > fequency domain like FFT or wavelets.
> > 
> > - Steve
> 
> Waveform Similarity Overlap Add?
> 
> http://bmrc.berkeley.edu/courseware/cs298/spring00/Ponceleon/Dulce/index4.htm

Yep, thats the one. Its just about slicing up the waveform into sections
and trying to paste them back together in overlapping ways, but so that
the features in the original sections line up.

- Steve