Hi dear pd freaks,
I'm currently using [env~] for measurement purpose (room's sound
monitoring and soundfiles analysis) and I have a few questions/remarks
...that unfortunately may have already been discussed here (I have
been unsubscribed from the list since a couple of years).
However :
1)
Just noticed : the [env~] help patch (from PDDP) states that default
analysis window is 1024 samples, while it links to Miller's example
C07.envelope.follower.pd where you can read that the default window is
256 samples.
2)
AFAIK there is no way to dynamically (message) change the analysis
window's size, at least without dynamic patching.. what I "painfully"
managed to do (see attached patch "ph_env~.pd").
Ah, I can see on sourceforge that it is a request open since 2012...
http://sourceforge.net/p/pure-data/feature-requests/109/
3)
Last but not least, the question I can't answer myself !
When using multiple [env~] it isn't very clear for me which one will
output first.
So it confuses me when I try to do very simple things like comparing
(difference) two signals' amplitude : doing a substraction requires to
input the [- ] object in correct order (right inlet before left's).
While doing it the wrong order may seem to work, I realized that in
fact I was comparing two different windows.
The actual order of output between different [env~] seems to be
related to the objects' order of creation. I think that order of
creation is a trouble since you cannot "read" it in the patch, so it
isn't "the diagram is the program" anymore. Moreover, as far as i can
deduce from what i experimented empirically (means : i'm not sure at
all) the first to output is the last that was created.
My experiments with order can be found in "order_env~.pd" attached file.
Finally, maybe all this mess is just me not being very clear with how
message/DSP are scheluded/interfaced... but I feel that [snapshot~] is
way more easy to understand and control, since it outputs value "on
demand" (bang) and order can be easily stated with [trigger]. I think
i'd feel much more comfortable with a kind of [env~] object that
computes the enveloppe of the last N audio blocks or last N samples,
"on demand", when triggered by a bang.
Maybe someone will answer me that I'm really confused and that my
problems are false problems... In case, I'd be glad to be taught the
right way !
Cheers,
Raphaël
_______________________________________________
Pd-list@lists.iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list