Subject: Re: [linux-audio-user] Pitch shifting
From: Steve Harris (S.W.Harris_AT_ecs.soton.ac.uk)
Date: Mon Jan 28 2002 - 19:17:45 EET
On Mon, Jan 28, 2002 at 05:28:44 +0100, Guenther Sohler wrote:
> hallo steve,
>
> thank you very much for all the information!
No problem.
> > The hard part is maintaining the ratio between the partials.
> Ya, but you can just wight them
> lets say i=number of partial
> and j=i*factor, which is some double number
> you can weight the amplitude of i to floor(i) and ceil(i) correctly
Yes, but there isn't a 1:1 relation between the bins and the partials.
> what do you exacly mean by zero padding ? can you excacly describe the way it
> should be done ?
As the FFT works on circular data you have to add an equal ammount of
zeros to the end of the time data. There is a good description of FFT
pitch scaling at http://www.dspdimension.com/
> > I think that the current best hardware uses LPC (Linear Predictice Coding).
> > I haven't tried this technique though, it is a bit of a specialisation.
> where can i find indofrmation on this ?
Sorry, don't know a relaible source. Try a google search. LCP for pitch
scaling is a specialisation, most applications of LPC are in compression.
- Steve
This archive was generated by hypermail 2b28 : Mon Jan 28 2002 - 19:09:00 EET