[PD] Sigmund~ parameters

Benoît Fortier benoitfortier at yahoo.ca
Thu Mar 13 17:48:48 CET 2014


Hi Miller,

Here's a more detailed example. I've found that this "problem" is particularly obvious with minpower. The sound file I'm playing through sigmund has peaks varying mostly between 60 and 80, with occasional peaks around 90 (according to sigmund's env output). To my understanding the help file suggest that a value of minpower around 50 should detect pretty much all the notes, and a value around 90 would detect almost none. But in practice I find that there's almost no difference. I have to use values of minpower around 100 and 130, with 130 having the effect of effectively making sigmund output no notes at all. Is it a normal behaviour? Is there something I don't get in the way sigmund's work? I feel the growth parameter have this same issue, it feels to me that it's working quite differently in practice than what you would expect from reading the help file.

Thanks for this wonderful object!

Benoît Fortier





Le jeudi 13 mars 2014 0h37, Miller Puckette <msp at ucsd.edu> a écrit :
 
I presume you found the (very short) mention in the subppatch of the
help wnidow ("pd setting-parameters").  

Basically, if the measured signal power is less than "minpower" (expressed
in dB) then the reported instantaneous potch is zero and no new notes will
be reports (and this can give rise to repeated notes if the input power
drifts below this value and then rises again.)

"growth" is a threshold whereby sigmund~ may report repeated notes at teh
same pitch, even if that pitch s continuously present in the signal, just
because of a fall and subsequent rise in the measured power.

This all needs a picture :)

M


On Wed, Mar 12, 2014 at 04:10:23PM -0700, Benoît Fortier wrote:
> Ok folks, never mind, I did some more testing with sigmund~ and I do now see that minpower and growth seems to be working. It works in a way I'm not quite sure to understand so I find it hard to use those parameters in a meaningful, predictable way, but I guess that's because of my limited understanding of what's going on exactly under the hood.
> 
> Benoît Fortier
> 
> 
> 
> 
> Le mercredi 12 mars 2014 16h50, Benoît Fortier <benoitfortier at yahoo.ca> a écrit :
>  
> Thanks for your reply. Yes I did some testing with fiddle and Helmholtz, If I need to I'll get to that in an other question.
> 
> My question wasn't clear enough sorry about that. I would like first to clarify some aspect of the sigmund "notes" mode which seems very straight forward in the help file, namely growth and minpower parameter, but which in practice I find it doesn't quite work as expected. As anyone ever managed to get significantly different output by changing those two parameters? Are they working at all, and if so can anyone explain for the record how they actually work? Or maybe I'm missing something obvious?
> 
> I'll ask a clearer question regarding other specific strategies for refining pitch tracking result soon, thank you for your help.
>  
> Benoît Fortier
> 581 995-5622
> 
> 
> 
> Le mercredi 12 mars 2014 16h02, Alexandros Drymonitis <adrcki at gmail.com> a écrit :
>  
> Haven't tried [sigmund~], but [fiddle~] and [helmholtz~] (the latter by Katja Vetter) and I found [both fine but [helmholtz~] a bit better for my taste. I think it's a bit more responsive. What if you combine this with [env~] for example and this way you get pitch and amplitude...
> 
> 
> 
> 
> On Wed, Mar 12, 2014 at 9:42 PM, Benoît Fortier <benoitfortier at yahoo.ca> wrote:
> 
> Dear list,
> >
> >
> >I've been working on a patch for quite a few months now which involves pitch tracking. I've explored many possible solutions, considering the pros and cons of every approach, made different tests, etc. I've also done a lot of reading (help files, forums, etc), but I couldn't find satisfying answers to all my questions. I am now in front of 4 or 5 different approach, each of which requires to do a certain number of compromise at different level. My feeling is that I am now at the point where I could possibly eliminate some of those compromise with your help.
> >
> >
> >I have a lot of questions, but I'll start with one regarding Sigmund's notes parameters. I've done a lot of tests, changing one parameter at a time. I found that only the stable time and vibrato parameters seems to be affecting the output the way it should. The other parameters - growth and minpower - doesn't seem to be working the way it's described in the help file. Am I the only person who noticed that? When I looked in the list archives, I found someone else mentioning this issue but nobody replied (and i lost the link to the thread).
> >
> >
> >I'm using my patch to track the pitch of a violin. What I'm trying to do is to convert the pitch of the violon into the most accurate "note on" message possible (with midi note and velocity). I need this patch to be as responsive as possible but also with as few insignificant output as possible.
> >
> >
> >Thanks for your help!
> >
> >
> >Benoît Fortier
> >581 995-5622
> >_______________________________________________
> >Pd-list at iem.at mailing list
> >UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
> >
> >
> 
> 
> 
> 
> _______________________________________________
> Pd-list at iem.at mailing list
> UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list

> _______________________________________________
> Pd-list at iem.at mailing list
> UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puredata.info/pipermail/pd-list/attachments/20140313/9ce9d982/attachment-0001.htm>


More information about the Pd-list mailing list