[PD] best method for non fft spitch shiting

nullpointer nullpointer at odessadesign.co.uk
Wed Oct 31 09:57:53 CET 2001


Hi ,Pix, Miller

Thanks for the pointer.. I have already looked at the implementation in the
pd patch ((same as the one in the abstrus harmoniser) and I have tried an
algo based on the same/ I was advised that you could try other algos like
reversing the direction of the read head at the end of the delayline (so it
bouncs back an amount based on the delay/transpo time) and then plays
forward, so it is ready for the next pass.. Also someone mentiond a method
using two or more chorus (i.e. modulated delaylines) effects with a
crossfade.. It seems though without using fft the result will always have a
slight wobbble in it?... I'll try making up another version of the pd unit
with more taps and maybe a different amp envelope for each scratch(tap) to
calm down the wobble...

Thanks

Tom

http://www.nullpointer.co.uk
http://www.dividebyzero.org
http://www.orphanrecords.co.uk
________________________________________________________
C:\AUTOEXEC.BAT
C:\REM [Header]
C:\>
C:\>PATH=C:\PERL\BIN;C:\WINDOWS\COMMAND;\C:jdk1.2.2\bin
________________________________________________________

(all suffixes enabled)

-----Original Message-----
From: Miller Puckette <mpuckett at man104-1.ucsd.edu>
To: nullpointer <nullpointer at odessadesign.co.uk>
Cc: pd-list at iem.kug.ac.at <pd-list at iem.kug.ac.at>
Date: 30 October 2001 22:30
Subject: Re: [PD] best method for non fft spitch shiting


>Hi,
>
>It's not perfect, but there's an example in 56.delay.pitchshift.pd in the
>audio examples of the Pd distribution.
>
>cheers
>Miller
>
>On Tue, Oct 30, 2001 at 10:11:05PM -0000, nullpointer wrote:
>> Hi all,
>>
>> THis is slightly off topic but the algo should hold the same for both a
pd
>> patch and a stand alone implementation.. I asked on music-dsp but no-one
>> knew (or could be bothered to answer).
>>
>> What is your reccomendation for the best method of implementing a non fft
>> style delayline based pitchshifter/harmoniser in dsp. I have set up two
>> readpoints 50% offset from each other and crossfading at the end and
>> beginning of the line as an attempt to eliminate discontinuities and
>> transients but i still get cips and flicker at the frequency of the delay
>> buffer size...I understand that it is impossible to get a perfect shifter
>> using this method but i am sure there must be methods to improve it?
>>
>> Tom
>> http://www.nullpointer.co.uk
>> http://www.dividebyzero.org
>> http://www.orphanrecords.co.uk
>> ________________________________________________________
>




More information about the Pd-list mailing list