[PD] Re: [PD-announce] pvoc~ 0.1

[Michael Iber]

Hi Ben,
nice work. I have been thinking about doing something like this myself, but 
my C-facilities are probably too basic, I guess.

I am very impressed about the performance advantage to the 
MSPuckette-abstraction (phaselockedvocoder). Your external only takes about 
20% of the abstraction. On the other hand, I find the abstraction sounds 
slightliy better. Can this be true, or is it psychological?

I have a couple of questions:
What is the reason using a fader for the locking like you are doing in the 
help patch?

What does the transient parameter do? What is a transient, how do I 
recognized it? In your example I couldn't hear any difference to using no 
transients.

You are using the FFTW_ESTIMATE-flag of FFTW, what would be the advantage 
using the FFTW_MEASURE one in this particular case?

Thanks and regards,
Michael

At 14:06 10.05.2003, you wrote:
>pvoc~ is a phase vocoder external that provides independent control of
>the playback position within the sound and pitch transposition via two
>control inlets.  It features adjustable phase locking and transient
>de-smearing (given a list of manually determined transient positions).
>In the future it may incorporate other enhancements, such as
>independent processing of peaks and noise.
>
>One advantage over an abstraction-based PV is that the delay of the
>output doesn't depend on the FFT size, so it's more immediately
>responsive.  Hopefully it's also a little more efficient, but I haven't
>tested it yet for efficiency.
>
>pvoc~ requires FFTW 3.  It has not been tested yet on anything other
>than Debian unstable on Intel.  Here's a direct URL, but it will only
>be there temporarily because I'm graduating soon and losing my web
>space..
>
>http://www.macalester.edu/~bsaylor/pvoc~-0.1.tar.gz
>
>Comments, suggestions, and code welcome. :)
>
>Ben
>
>_______________________________________________
>PD-announce mailing list
>PD-announce at iem.kug.ac.at
>http://iem.kug.ac.at/cgi-bin/mailman/listinfo/pd-announce

Michael Iber

http://www.iberspace.de
mailto:mail at iber-online.de