i always just multiply the signal by 10 and then use clip~ to bring it back down to size.

[inlet~] | [*~ 10] | [clip~ -0.05 0.5] | [outlet~]

not clever or authentic, but will destroy the soundwaves.

Frank Barknecht wrote:

I also like the delay based total destroytion of rrad.erosion~.pd

It works by writing the incoming signal into a vd~-delayline, and then modulating the delay time with bandpass filtered noise~ or sine waves.

Another fun delay trick is using the comb-filter/karplus-strong/pluck model. Make a few of them, and set them somewhere between where you get a "plucked" frequency and where it turns into a distinct echo, and try different intervals between the timing of each, such as setting them to octaves or setting one only a few Hz off of another for a great "detuned" effect. Something like a sustained, overdriven guitar sound comes out. No guitar needed, as Josh Steiner mentioned replying to Orm Finnendahl's "karplus strong" thread.

d.

On Thu, 13 Apr 2006, hard off wrote:

i always just multiply the signal by 10 and then use clip~ to bring it back down to size.

[inlet~] | [*~ 10] | [clip~ -0.05 0.5] | [outlet~]

not clever or authentic, but will destroy the soundwaves.

[clip~ -0.5 0.5] gives a more balanced sound and its output doesn't have much DC compared to -0.05 0.5.

for a more analog sound, replace the clip~ by some kind of [expr~ tanh($v1)], which is clearly my favourite. It's CPU-expensive so it could deserve a [tabread4].

_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801 - http://artengine.ca/matju | Freelance Digital Arts Engineer, Montréal QC Canada

On Thu, 13 Apr 2006, derek holzer wrote:

In PD you could use waveshaping, either by table-lookup or by using [expr~] if you're one of those mathematician types that haunt this list ;-)

Most notably Miller Smith Puckette, whose doctoral thesis was a Probability Theory paper titled:

"Shannon Entropy and the Central Limit Theorem"

http://www-crca.ucsd.edu/~msp/Publications/thesis.ps

And about waveshaping, let me say that, for "big distortion", in addition to the tanh(x) and atan(x)*2/pi functions that I may have previously mentioned, x/sqrt(1+x*x) and erf(x) are quite cool. They aren't particularly modeling the behaviour of analog amps nor speaker membranes, but still, they're rather simple functions that appear everywhere in science.

_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801 - http://artengine.ca/matju | Freelance Digital Arts Engineer, Montréal QC Canada

Is there a spectral theory of waveshapers? Generally speaking, distortion produces different harmonics of signals.... Another way of generating harmonics is to use convolution in the fourier domain... for example, the nonlinear transfer function y=x^3 can be represented in the fourier domain:

FT(y)= FT(x) convolved with FT(x) convolved with FT(x) and curiously....we can compute and convolve fractional powers of the FT... I haven't gone too far in depth for this hypothesis....just a thought for you mathematician types, heh heh

Chuck

On 4/16/06, Tim Blechmann TimBlechmann@gmx.net wrote:

On Sun, 2006-04-16 at 16:16 -0400, Mathieu Bouchard wrote:

...

And about waveshaping, let me say that, for "big distortion", in addition to the tanh(x) and atan(x)*2/pi functions that I may have previously mentioned, x/sqrt(1+x*x) and erf(x) are quite cool. They aren't particularly modeling the behaviour of analog amps nor speaker membranes, but still, they're rather simple functions that appear everywhere in science.

well, most cumulative distribution functions can be adapted as wonderful wave-shapers .... with the right preamplification and paramerers they make really powerful tools ...

t

-- TimBlechmann@gmx.de ICQ: 96771783 http://www.mokabar.tk

Life is really simple, but we insist on making it complicated. Confucius

-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQBEQrS0NDZZF/Yk3sURAqajAKCnHJvkd059yD3yYVGvFwP5pOZLxQCePwGC l8ghW8+29SJV0Z4X7ib4B7g= =YL76 -----END PGP SIGNATURE-----

---

PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list

-- Charles Zachary Henry

anti.dazed.med Med student who needs a Mickey's

Mathieu Bouchard wrote:

...

and curiously....we can compute and convolve fractional powers of the FT...

Yeah, I tried fractional convolutions last week. I tried applying this effect on video, and it gives some *really* cool results if tuned appropriately. With Pd of course.

ok so i don't understand but i want to. what's a good primer on fft stuff? i know the basics but don't really know how it can be used or dealt with interestingly...

vincent rioux wrote:

Dear all,

hi. my mail-reader seems to think that this is in the same thread as "mathematician types".

i am looking for a simple way to record gem pixes. is there a flag to properly compile Gem with pix_record support on Linux? (Gem i got compiled says it does not support it).

you need a recent version of libquicktime (NOT quicktime4linux!) a version that is up-to-date enough is already in debian/sid. if you are on ubuntu you will have to wait for the next release. i cannot speak for other distros.

btw i noticed that it was very cpu consuming to use gem2pdp and then pdp_rec~ to record gem pixes -

if you only want to record gem _pixes_ then you are better off with pix2pdp.

mfg.asdr.- IOhannes

vincent rioux wrote:

thanks IOhannes, are you talking about libquicktime available here: http://sourceforge.net/project/showfiles.php?group_id=46254

yes

isn't it possible to compile it on ubuntu?

i guess it is possible fairly well. (there is no reason why it should not)

i was rather refering to pre-build packages of the distro of your choice ubuntu breezy's libquicktime-packages are too old; so you have to wait for dapper if you don't want to compile the packages yourself; or use debian; or...

sorry, at times i am of the impatient type.

nevermind. however, i didn't know you were using ubuntu (how should i?)

mfg.as.dr IOhannes

...

...

isn't it possible to compile it on ubuntu?

i guess it is possible fairly well. (there is no reason why it should not)

btw libquicktime 0.9.8 doesn't work at all with GridFlow, crashes on startup, and it seems to be a bug in libquicktime that I can't get around.
Anyone got similar experiences with libquicktime 0.9.8 (but e.g. through GEM) ?

hi matju,

well, libquicktime 0.9.8 compiled on ubuntu. Gem did not crash but then i could not get Gem to display movies anymore (i got just a black screen) (pix_record got compiled however apparently). I did not try to recompile gridflow with that libquicktime version.

with previous version of libquicktime i tried to use (but did not manage to get them working):

• #pix_export of Gridflow (i will report more precisely the error

message i received)

• pix2pdp (just crashed when connected, especially in my patches - not

the example)

i finally opted for a a double patch (two pd instances launched) with one part written with Gem/pdp2gem (a visual control interface) and one in pure pdp (for actual display) that i could record with pdp_rec~.

best regards, vincent