Subject: [linux-audio-dev] Re: idea: wavelets, sampler program
From: Tom Weber (np98towe_AT_dd.chalmers.se)
Date: Fri Jun 20 2003 - 20:28:48 EEST
> There may be some problems...
>
> - What about noise-like signals ?
The brain treats all sound in a fourier representation, noise also.
Don't you think noise sounds like thousands of different beeps? The
computer representation would be no different.
> - For a resolution of 0.1 Hz you need 10 seconds of sound. Only a
real
> sine wave lasting the whole 10 s will tranform as a 'peak',
everything
> else will be smeared out.
Hmm, are you sure? I would agree on that for a 0.1 Hz signal, which is
not the case. Assume the sound contains a 100.0 Hz signal, then the
analyzed amplitude (using sin and cos for phase independency) should
be higher for 100.0 Hz than for 99.9 or 100.1, even for short periods.
But I'm a newbie to DSP...
What I actually aim at doing is a "sampler" program/plugin with pitch
scaling, using this "complete fourier transform" approach to overcome
the problems listed in:
http://www.dspdimension.com/html/timepitch.html
Any suggestions?
This archive was generated by hypermail 2b28 : Fri Jun 20 2003 - 20:36:34 EEST