Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!uknet!festival!leeds.ac.uk!news
From: een6njb@gps.leeds.ac.uk (N J Bailey)
Subject: Re: Change playback speed without changing
Reply-To: een6njb@gps.leeds.ac.uk
Organization: University of Leeds, U.K.
Date: Mon, 14 Feb 1994 12:03:11 GMT
Message-ID: <1994Feb14.120311.18299@leeds.ac.uk>
References: <1994Feb3.135104.20554@ucl.ac.uk>
Sender: nntp@leeds.ac.uk
Lines: 14

There is a piece of software called CSound which is widely used in the computer
music community, and which contains a phase vocoder module.  By taking concurrent
FFT frames and strectching them further apart in time, then resynthesising,
you achieve a very high quality extension.  You need to learn the CSound command
language, but it's not too bad.  I know of music postgrads who have used it to 
notate Jazz and pieces of ethnic music by streching a sequence of really rapid
notes by factors in excess of 20!

The down side is that the processing power involved in messing around in the 
frequency domain is enormous compared with algorithms like SOLA.

Look also at the paper by Keith Lent: "An Efficient Method for Pitch Shifiting
Digitally Sampled Sounds", Computer Music Journal, 13(4) Winter 1989.

