Newsgroups: comp.speech
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!agate!msuinfo!harbinger.cc.monash.edu.au!news.cs.su.oz.au!metro!extro!brucet
From: brucet@extro.ucc.su.OZ.AU (Bruce Tulloch)
Subject: Re: Pitch changing?
Message-ID: <brucet.778807515@extro>
Sender: news@ucc.su.OZ.AU
Nntp-Posting-Host: extro.ucc.su.oz.au
Organization: Information Services, Sydney University, Sydney, NSW, Australia
References: <34ejga$e8m@acorn.acorn.co.uk>
Date: Mon, 5 Sep 1994 23:25:15 GMT
Lines: 30

> Is there a way to change the pitch of a sample without changing its
> playback speed?
>
> --Philip
 
It seems this (and the one about time dilation) are perennial questions in
comp.dsp and comp.speech. This is a reply I gave a couple of weeks ago in
comp.dsp on this question.
 
: Time dilation is a process whereby the temporal evolution of a signal is 
: modified - and by analogy the pitch can be shifted without stretching its 
: temporal evolution, ie the time axis (or pitch axis) of a time/frequency 
: representation of the signal is re-scaled. The best way of achieving it is 
: through time/frequency processing algorithms such as the phase vocoder, 
: wavelet transforms or wigner-ville techniques. The idea is simple - in the 
: case of the phase vocoder for example, one splits the signal into it's 
: constituent partials with a bank of bandpass filters, interpolates the 
: amplitude, phase and frequencies (in time) of the output of each bandpass filter 
: and uses the interpolated results to drive a bank of oscillators to recreate 
: the dilated signal. Short-time Fourier transforms using the FFT can be used to
: reduce the computational requirements to make the algorithm tractable. If the
: signal being dilated is purely harmonic then simplifications can be found 
: allowing the entire process to be performed in the time domain. Indeed some 
: of the simple pitch shifting techniques previously described in this group 
: can be analysed in this way.
 
Trust it helps.
-- 
             bruce tulloch sydney australia - brucet@ucc.su.oz.au
***complex problems have straight forward, easy to understand wrong answers***
