7 Apr
2004
7 Apr
'04
7:08 a.m.
On Wed, 2004-04-07 at 09:46, Cournapeau David wrote:
Erik de Castro Lopo wrote:
To do it well is difficult, but the basic stuff is really simple, if you
have a library to do sample-rate conversion: doing the time scale
compression/expansion
with the simplest method (overlapp add) is just a matter of buffering.
It will sound awful, of course !
you can do this (or perhaps it's some other even simpler method) in a
tracker - all you have to do is retrigger the sample at progressive
intervals, then you can change the pitch without changing the playback
speed (or vice versa). it's very very dirty but it can be taken to great
extremes :) - I implemented the same functionality in ssm's poshsampler
cheers,
dave
--
................................. www.pawfal.org/nebogeo
But I cant
find out how to deal with the missing data that is removed - to
keep the original length.
Changing the pitch without changing the length is true pitch shifting. The
code and concepts required to do this well is extremely complicated.