|
From: Josh G. <jg...@us...> - 2003-06-24 11:29:13
|
On Tue, 2003-06-24 at 00:40, Steve Mokris wrote: > > The term often used is interpolation. There are different methods for > > this that trade off between quality and speed. Linear interpolation is > > probably the simplest (besides no interpolation :) There is cubic > > interpolation and other more complex ones as well, each increases the > > number of points used in approximating (interpolating) samples. > > to throw something else out there into the possible-feature-pool (and > supporting the general opensource premise that "everything is better than > something"), i think i might be interesting/useful to have a conveniently > abstracted interface for converting "input wave + note data" into "output > wave"; that is, pitch scaling. > > ...and also be able to select which algorithm is used, as a parameter > intrinsic to each instrument. that way, one might (eventually) implement > all of the above-mentioned algorithms, plus some others such as > formant-preserving pitch scaling, time-preserving pitch scaling, and > maybe, for retro effect, emulations of a few of those aggressively clunky > samplers from the 1980s. > > steve > Perhaps libsamplerate (http://www.mega-nerd.com/SRC/index.html) might be something similar to what you mentioned. I just recently heard about it, so I'm not real familiar with its capabilities. Cheers. Josh Green |