cnmat resonators~, sdif, spear & public models
hi everybody!
i have some questions regarding the cnmat resonators~ object, which i want to use in some patch:
my main question concerns the list format that resonators~ accepts: its stated as requency/volume/decay triplets, whereas the RES1 format specified in the cnmat documentation consists of frequ/vol/ec/phase - that would be the easiest part, just cutting away every forth number - unfortunately i didn't encounter any resonance models in this format in the wwww (apart from the aLib-res-models, which also seem to need some re-formatting due to the coll-like textfile).
are there any more public resonance models out there? some database? are there any easy converters which free me from the need of doing this manually?
second question is about SPEAR, which is a nice tool, but i can't get it to output this RES1 format --- am i doing something wrong? are there tricks to do this?
i'd be happy to eg. choose some time frame from an sdif file and use it as a resonance model to treat other audio material...
any help would be highly appreciated!
thanks in advance!
clemens
> my main question concerns the list format that resonators~ accepts: its stated as requency/volume/decay triplets, whereas the RES1 format specified in the cnmat documentation consists of frequ/vol/ec/phase -
You've got it. A message to SDIF-tuples "tuples columns 1 2 3" will get you the data you need from that
>
> are there any more public resonance models out there?
That would be great, and we're working on it. There are some sprinkled in the CNMAT MMJ Depot, but not enough. The only software I know that does this analysis is ResAn, which is part of IRCAM's not-free Diphone Studio.
> second question is about SPEAR, which is a nice tool, but i can't get it to output this RES1 format --- am i doing something wrong? are there tricks to do this?
SPEAR doesn't do this, AFAIK. Just 1TRC.
> i'd be happy to eg. choose some time frame from an sdif file and use it as a resonance model to treat other audio material...
>
Somewhere I have a patch that does this; I'll try to find it. All you need is some made-up decay-rate data... frequency * 0.1 might be a good place to start.
mz
> You've got it. A message to SDIF-tuples "tuples columns 1 2 3" will get you the data you need from that
> >
ah, but which format is the one tuples accepts? i'm quite new to the sdif-stuff, so sorry for stupid questions :)
> > are there any more public resonance models out there?
>
> That would be great, and we're working on it. There are some sprinkled in the CNMAT MMJ Depot, but not enough. The only software I know that does this analysis is ResAn, which is part of IRCAM's not-free Diphone Studio.
>
i read about it. pity that it is not free.
> > i'd be happy to eg. choose some time frame from an sdif file and use it as a resonance model to treat other audio material...
> >
>
> Somewhere I have a patch that does this; I'll try to find it. All you need is some made-up decay-rate data... frequency * 0.1 might be a good place to start.
>
thanks for this tip - i'll try that!
btw. the mmj depot is missing a file that is claimed to be included in the archive on the cnmat website-
it's called Resonance Model Workspace and seems to contain some really useful parts - i wrote to the author but still no response- is it possible to somehow get this or the one you're writing about?
thanks alot in advance!!1
clemens
Quote: vogt wrote on Thu, 03 April 2008 07:32
----------------------------------------------------
>
> > You've got it. A message to SDIF-tuples "tuples columns 1 2 3" will get you the data you need from that
> > >
>
> ah, but which format is the one tuples accepts? i'm quite new to the sdif-stuff, so sorry for stupid questions :)
>
? SDIF-tuples will suck data out of anything you can get into SDIF-Buffer
If you've found the Spectral Tutorials in the CNMAT MMJ Depot, look at number 26.
> btw. the mmj depot is missing a file that is claimed to be included in the archive on the cnmat website-
> it's called Resonance Model Workspace and seems to contain some really useful parts - i wrote to the author but still no response- is it possible to somehow get this or the one you're writing about?
----------------------------------------------------
Unfortunately, the author pulled that patch because it wasn't ready for prime time. I'll see if I can't get him back on the case.
mz
I have a quick question about resonators~
What is the unit for the decay? Is it the -60db ring time?
Thanks.
According to CNMAT's Standard Frame Types page (http://cnmat.berkeley.edu/library/sdif/sdif_standard_frame_types), the decay is specified as follows:
The decay curve of a resonance should be the same as that of a two-pole filter with bandwidth equal to decay rate divided by pi. This formula gives the amplitude of each sinusoid over time:
amp(t) = initial_amp * e ^ (- decay_rate * t)
So decay = bandwidth * pi. But I can't figure out what the decay rate means in the time domain (e.g. -dB/sec or similar).
Anyone?
On 27 Oct 2008, at 19:11, Anthony wrote:
>
> According to CNMAT's Standard Frame Types page (http://
> cnmat.berkeley.edu/library/sdif/sdif_standard_frame_types), the
> decay is specified as follows:
>
> The decay curve of a resonance should be the same as that of a two-
> pole filter with bandwidth equal to decay rate divided by pi. This
> formula gives the amplitude of each sinusoid over time:
> amp(t) = initial_amp * e ^ (- decay_rate * t)
>
> So decay = bandwidth * pi. But I can't figure out what the decay
> rate means in the time domain (e.g. -dB/sec or similar).
according to this relation, if bw is in Hz, decay should be in Hz,
too. don't know if that's what you were looking for.
but you can solve the above equation for decay_rate and thus control
the decay of the filter (sinusoid) by specifying a time in seconds.
decay_rate = -ln( amp(t) / initial_amp ) / t
does that help?
volker
Yes, that's perfect... thanks!