hi,
would it be possible to patch an acoustic hi-hat pedal complete with heel-splash sound, closed sound, open sound, and everything in-between.
i am not so sure where to start, i want a sampler-base solution.
adsr, playing with filters, morphing of sounds, using 2 samples or more (open, middle, close, very close). i would gladly pay a beer for this.
pat
these aren't at all acoustic. and they aren't at all sample based. but they do give you very nice control over the timbre and shape of the sound.
open up d-808cy~-help.pd or d-808hh~-help.pd
chur d
patrick wrote:
hi,
would it be possible to patch an acoustic hi-hat pedal complete with heel-splash sound, closed sound, open sound, and everything in-between.
i am not so sure where to start, i want a sampler-base solution.
adsr, playing with filters, morphing of sounds, using 2 samples or more (open, middle, close, very close). i would gladly pay a beer for this.
pat
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
it'd help if i attached them eh
Damian Stewart wrote:
these aren't at all acoustic. and they aren't at all sample based. but they do give you very nice control over the timbre and shape of the sound.
open up d-808cy~-help.pd or d-808hh~-help.pd
chur d
patrick wrote:
hi,
would it be possible to patch an acoustic hi-hat pedal complete with heel-splash sound, closed sound, open sound, and everything in-between.
i am not so sure where to start, i want a sampler-base solution.
adsr, playing with filters, morphing of sounds, using 2 samples or more (open, middle, close, very close). i would gladly pay a beer for this.
pat
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Hey Patrick,
It's the "everything in-between" bit that is hard. A hihat is one of those instruments that seems really simple, but on deeper analysis you see it is an amazingly complex and subtle device (which is obvious when you hear a really good drummer playing). With samples you can get the usual opening and closing strike sounds, and the cup/clash.
But if you want a really good model you probably need to look at physical modelling and granular methods.
The two parts can rattle against each other causing new excitations.
Sometimes the top cymbal rotates around the bottom one, creating a rolling 'clatter'.
The pressure of the pedal pushes them together harder increasing the frequencies (speeding up the roll - like if you push down in the middle of a rolling dinner plate) and damping both parts more.
You can hit it on the edge or at any radius from the centre. As a simple rule the spectral complexity increases as you move outwards, hitting the bell in the middle produces a purer, shorter tone.
You can use the tip of the stick, or brush, or hit lower down to get a less elastic impact with more energy transferrence (louder and noiser).
So for the input vector; for pedal you probably want two continuous ranges, position and pressure (where they are touching), and for excitation you probably want two more impact event parameters, energy (0.5 * mass * velocity^2) and duration (impulse).
You can also model leaving the stick connected, which damps the vibration at one point and reduces the modes.
A full physical model of two interacting centre supported circular plates is certainly too expensive. FM methods can give you the raw spectra to blend according to interaction rules. (In fact - because the relationship of spectra in both FM and the disc modeal model both derive from Bessel functions you can get very natural evolution if you choose the right synth model - complex FM with multiple modulator sources works extremely well.)
The trick is mapping the control params onto the synthesis params. As they come together you can tap off a little bit of the lowest and strongest vibrational mode (lop~ -> max~) and use it to make impulse spikes that amplitude modulate the other plate. Of course this is a reciprocal relationship (the lower plate also rattles against the upper one) - so it's ring modulation of the two maximum excusions to get a 'rattle modulator'. Moving the lop~ up and bringing the clip down will create the impression of the plates being forced together.
The other approach is to make a parametric map from samples into a great big wavetable to use as grains. I have never tried it but I think that would work well for hi-hat.
a.
On Wed, 26 Nov 2008 01:27:23 -0500 patrick puredata@11h11.com wrote:
hi,
would it be possible to patch an acoustic hi-hat pedal complete with heel-splash sound, closed sound, open sound, and everything in-between.
i am not so sure where to start, i want a sampler-base solution.
adsr, playing with filters, morphing of sounds, using 2 samples or more (open, middle, close, very close). i would gladly pay a beer for this.
pat
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Someday I hope you make a big ole' physical modeled jazz drum kit, Andy.
Genius!
~Kyle
On Sat, Nov 29, 2008 at 10:31 AM, Andy Farnell padawan12@obiwannabe.co.ukwrote:
Hey Patrick,
It's the "everything in-between" bit that is hard. A hihat is one of those instruments that seems really simple, but on deeper analysis you see it is an amazingly complex and subtle device (which is obvious when you hear a really good drummer playing). With samples you can get the usual opening and closing strike sounds, and the cup/clash.
But if you want a really good model you probably need to look at physical modelling and granular methods.
The two parts can rattle against each other causing new excitations.
Sometimes the top cymbal rotates around the bottom one, creating a rolling 'clatter'.
The pressure of the pedal pushes them together harder increasing the frequencies (speeding up the roll - like if you push down in the middle of a rolling dinner plate) and damping both parts more.
You can hit it on the edge or at any radius from the centre. As a simple rule the spectral complexity increases as you move outwards, hitting the bell in the middle produces a purer, shorter tone.
You can use the tip of the stick, or brush, or hit lower down to get a less elastic impact with more energy transferrence (louder and noiser).
So for the input vector; for pedal you probably want two continuous ranges, position and pressure (where they are touching), and for excitation you probably want two more impact event parameters, energy (0.5 * mass * velocity^2) and duration (impulse).
You can also model leaving the stick connected, which damps the vibration at one point and reduces the modes.
A full physical model of two interacting centre supported circular plates is certainly too expensive. FM methods can give you the raw spectra to blend according to interaction rules. (In fact - because the relationship of spectra in both FM and the disc modeal model both derive from Bessel functions you can get very natural evolution if you choose the right synth model - complex FM with multiple modulator sources works extremely well.)
The trick is mapping the control params onto the synthesis params. As they come together you can tap off a little bit of the lowest and strongest vibrational mode (lop~ -> max~) and use it to make impulse spikes that amplitude modulate the other plate. Of course this is a reciprocal relationship (the lower plate also rattles against the upper one) - so it's ring modulation of the two maximum excusions to get a 'rattle modulator'. Moving the lop~ up and bringing the clip down will create the impression of the plates being forced together.
The other approach is to make a parametric map from samples into a great big wavetable to use as grains. I have never tried it but I think that would work well for hi-hat.
a.
On Wed, 26 Nov 2008 01:27:23 -0500 patrick puredata@11h11.com wrote:
hi,
would it be possible to patch an acoustic hi-hat pedal complete with heel-splash sound, closed sound, open sound, and everything in-between.
i am not so sure where to start, i want a sampler-base solution.
adsr, playing with filters, morphing of sounds, using 2 samples or more (open, middle, close, very close). i would gladly pay a beer for this.
pat
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management ->
http://lists.puredata.info/listinfo/pd-list
-- Use the source
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
Useful post, thank you very much Andy. Indeed, it's pretty complex dsp, at least i have the technical terms ;)
pat
YVW, and FWIW, there's some tuts to look at
mine using a familiar additive/modulation method here http://www.obiwannabe.co.uk/html/toys/cymbal-FMAM/cymbalfmam.html
and here http://www.mcld.co.uk/cymbalsynthesis/
Dan Stowell does a great job of using subtractive methods in SC
A nice thread followed on music-dsp
http://music.columbia.edu/pipermail/music-dsp/2007-December/thread.html
as we compared the merits of some approaches.
a.
On Mon, 01 Dec 2008 19:20:22 -0500 patrick puredata@11h11.com wrote:
Useful post, thank you very much Andy. Indeed, it's pretty complex dsp, at least i have the technical terms ;)
pat
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
ah yeah, i've seen that dan stowell page before. he does make a really nice cymbal sound.
he mentions using supercollider's "ringz resonators". what might be a close equivalent in pd?
Hallo, hard off hat gesagt: // hard off wrote:
ah yeah, i've seen that dan stowell page before. he does make a really nice cymbal sound.
he mentions using supercollider's "ringz resonators". what might be a close equivalent in pd?
ringz is a two pole (+ maybe two zero?) resonant bandpass filter. Attached is something similar build with Pd. It's not really finished, just something I'm working on from time to time. Set "norm" to 1 and play with the ring time parameter to test. Be careful with the volume of your speakers, there isn't much protection build in.
Ciao
Frank Barknecht Do You RjDj.me? _ ______footils.org__
i just followed the basics from dan stowell's page, and built a reasonable sounding hihat in pd.
it has 100 resonating bandpass filters connected, and so for some reason needs a couple of 'hits' until all those filters let it work properly. also needs a couple of hits to get things working right when changing between presets.
pd's cpu meter tells me it's only using 1-2% here, but i am having trouble believing that. anyway, runs ok here.
patches attached.
Hallo, hard off hat gesagt: // hard off wrote:
i just followed the basics from dan stowell's page, and built a reasonable sounding hihat in pd.
it has 100 resonating bandpass filters connected, and so for some reason needs a couple of 'hits' until all those filters let it work properly. also needs a couple of hits to get things working right when changing between presets.
pd's cpu meter tells me it's only using 1-2% here, but i am having trouble believing that. anyway, runs ok here.
patches attached.
Very nice!
To all those running a normal Pd: Replace [initbang] with [loadbang] in diy3-ringz.pd to make it work. (Why are you using initbang here?)
Ciao
Frank
Lovely job! Hit position seems jolly effective too. And some cool presets.
I found a great effect by very slightly changing the spectrum jitter.
cheers,
Andy
On Wed, 3 Dec 2008 01:38:38 +0900 "hard off" hard.off@gmail.com wrote:
i just followed the basics from dan stowell's page, and built a reasonable sounding hihat in pd.
it has 100 resonating bandpass filters connected, and so for some reason needs a couple of 'hits' until all those filters let it work properly. also needs a couple of hits to get things working right when changing between presets.
pd's cpu meter tells me it's only using 1-2% here, but i am having trouble believing that. anyway, runs ok here.
patches attached.
I wonder if the "warmup problem" (initial state for causal filter) can be solved by keeping a very low level noise signal into them at all times?
On Wed, 3 Dec 2008 01:38:38 +0900 "hard off" hard.off@gmail.com wrote:
i just followed the basics from dan stowell's page, and built a reasonable sounding hihat in pd.
it has 100 resonating bandpass filters connected, and so for some reason needs a couple of 'hits' until all those filters let it work properly. also needs a couple of hits to get things working right when changing between presets.
pd's cpu meter tells me it's only using 1-2% here, but i am having trouble believing that. anyway, runs ok here.
patches attached.
actually, the 'warmup problem' was easily solved by removing the [switch~] object. that object was also causing strange things to happen when changing presets too, so it's better off to just not use it. cpu usage seems kinda minimal anyway, which baffles me cos there are 100 filters running.
i replaced the initbang with a loadbang construct.
oh and andy, i did try letting a really low level of noise into the filters, and wow! it really makes things sound a lot better! so now i added a 'noisiness' parameter, and the realism of the sounds just increased 200% i reckon.
updated version attached.
-
Andy Farnell -
Damian Stewart -
Frank Barknecht -
hard off -
Kyle Klipowicz -
patrick