Acoustic modular sounds

[dougcl]

A-117 stuff here (with help of QMMG)... uh holy shit virb just imploded. Goodbye virb! So long! Screw you!

[AndreasKrebs]

thank you andreas for your very detailed and nice old school demos!
somewhere, in the first track mostly, it seems you are playing objects more than acoustic instruments, that is interesting anyhow

Thanks! The tracks were not intended to be demos, but I'm happy if they are useful as such.
I like it to be able to use the synth as some kind of "percussive instrument" as well as a tonal one (and everything in between). The BBDs are quite helpful for this.
Andreas

[en.]

Thanks! The tracks were not intended to be demos,

i thought you did them for me

ahaha sorry i used the wrong word, not meant to decry your works
thanks again for the links to your tracks!

[frijitz]

Holy cow I had no idea the finite element method (and finite differences) were being applied to sound synthesis.

The original work was:

"On the oscillations of musical instruments", McIntyre, Schumacher and Woodhouse, J. Acoust. Soc. Am 74(5), 1983.

This was the basis for the Yamaha VL1 Virtual Acoustic Tone Generator and other physical modeling synths. (Physical modeling originally refered to machines running physics models of instruments, not the current meaning of emulating synth circuits.) Julius Smith (Stanford CCRMA) probably still has a lot of his work up at his site.



Ian

[dykesh]

After researching all this stuff, I'm still not sure which to get between the 256 and 512. It seems like most of the demos of the 256 sound a little more like Karplus Strong to me and the 512 sounds a little more like a delay. Any opinions on this? I was set on the 512 but now I'm not so sure....

[selfoscillate]

i prefer the 256 stages bbd for karplus strong synthesis.
the highest possible pitch in a karplus strong patch is
determined by the shortest delay time possible with the bbd chip.
a 256 stages bbd goes one octave higher than a 512 stages bbd.

[dykesh]

Does that mean the 256 cannot go as low as the 512 or is the limitation just on the higher octaves? Thanks...

[AndreasKrebs]

Does that mean the 256 cannot go as low as the 512 or is the limitation just on the higher octaves? Thanks...

When you go lower in pitch, you will sooner hear the 256's internal clock oscillator than the 512's (i.e. you will sooner need to apply a steep lowpass filter like the A-108 to its output).
Andreas

[dykesh]

Great, now I got it. Thanks

[selfoscillate]

Does that mean the 256 cannot go as low as the 512 or is the limitation just on the higher octaves? Thanks...

the highest and lowest possible pitches are directly coupled
to the number of stages in a bbd. a bbd has a maximum and
a minimum delay time, limiting the frequency range on both
ends. bbd's with different size play in different octaves, to say so.
my personal impression is that the shorter bbd's give better results
than longer bbd's, when used in a karplus strong patch. the smaller
bbd's sound cleaner and less distorted. or in other words,
less "delay" and more "oscillator".

[AndreasKrebs]

You may also take a look at the 128 stage BBD: very usable as a sound source, quite clean. (I've bought it as a complement to my A-188-2, which is still the best allround-superversatile BBD)
Andreas

[dykesh]

It looks like the 188-2 goes down to only 396? Am I looking at this correctly? I'm reading the manuals but I'm not sure what I would lose with a 188-2? Reading the description makes me think the clock noise may be a little too much??? Oh, and Andreas, I loved those songs you put up!!!

[Umcorps]

I think just for KP applications, the 256 stage version of 188-1 is the best compromise in terms of usable range and simplicity. With that one the clock noise isn't normally an issue when you are trying to produce pitches above 250 Hz or so.

Just to muddy the waters though, you don't always have to lower the clock rate to the point where it is audible to get lower frequency pitches. Putting a filter in the feedback loop can change the pitch and timbre of the output quite dramatically. I had a weird kind of baritone sax thing going on the other day doing just this at a clock rate that, unfiltered, produced tones an octave higher.

It was a bastard to get it to track properly though. So ideal for microtonal free jazz.

[darenager]

Doepfer A-137-1 excellent for string, brass and woodwind stuff, as well as all the weird stuff it can do. The trick is to animate the CV inputs at differing rates and amounts, amazing module IMHO

[selfoscillate]

It looks like the 188-2 goes down to only 396? Am I looking at this correctly? I'm reading the manuals but I'm not sure what I would lose with a 188-2? Reading the description makes me think the clock noise may be a little too much??? Oh, and Andreas, I loved those songs you put up!!!

the a188/2 uses a 3328 stages bbd, it only has an output tap for 396 stages,
which is not the same as if it would have 396 stages in total.
the a188/2 is much noisier than f.e. the a188/1 with 256 stages, not
only regarding the clock noise, but in general.
i like the a188/2 too, but not for karplus strong stuff.

[dykesh]

Thanks so much for the info. I'm really trying to simulate many of the songs I've heard that were done with the 256 and maybe try to get some of the banjo/guitar type stuff; so it sounds like the 256 will do all this.

[AndreasKrebs]

The A-188-1X (128 stages only) can reach quite low regions. Here's a short sound demo:
http://www.andreaskrebs.de/assets/media/A-188-1X.mp3

Patch: A-155 Sequencer triggers an A-142 VC Decay envelope which controls an A-132-3 VCA (short decay time). VCA input is some noise from an A-117. VCA out goes into an A-188-1X BBD. The BBD frequency is controlled by the A-155 and an A-185-2 (to switch octaves - something like octaves: 1 oct down every 30 seconds).
In the feedback loop, an A-108 is inserted with bandpass out, no emphasis and manually changed filter frequency. BBD output is sent into another A-108 (filter frequency is controlled by BBD frequency CV out), the 48dB out is then sent into DAW.

Andreas

[AndreasKrebs]

Same patch as above, but this time with an A-188-2 (396 stages tap used):
http://www.andreaskrebs.de/assets/media/A-188-2.mp3

Around 3:00 you can hear that there's a little bit more noise / crosstalk involved with the tapped BBD. (Nothing a Stratocaster player would even care...)

Andreas

[Thonk Support]

The thread that keeps on giving

[dykesh]

Andreas, thanks for those comparisons, that helps a great deal. I think a 256 will be the best choice for me.

[KNYST]

yesterday i made a recording that sounds like a mixture of marimba and contra bass with my digital vcos and the qmmg

Gates sind die besten

Damn that's great!

-EDIT-
Lovely thread!

[julianop]

I like those modular sounds that tread that edge between sounding electronic and sounding somehow acoustic in nature.

Not sure exactly what you are looking for, but I did a lot of stuff like this at one time:
http://home.comcast.net/~ijfritz/sf_sc1.mp3
http://home.comcast.net/~ijfritz/sf_sc1.mp3



Ian

Hey, Ian, you still around??

[Xomrys]

There are basically two classes of instruments: instantaneously excited (plucked or struck) and continuously excited (bowed or wind instruments). Hard to think of another category right now. The Macintyre-Woodhouse paper cited above seems to focus on the continuously excited instruments with the feedback loop. But here you need the right kind of controller because you don't just hit something and let it go. The performer has to be actually in feedback with the instrument -- all 'expression' comes from this. Other than having the performer actually use a bow or wind controller, I'm not sure how you would model this aspect.

But for the instantaneously excited instruments, I think the most important ingredient is simply a fast envelope, so you get that surface fidelity and crispness. The problem is everything today is marketed as fast, despite almost nothing being actually so. Even the Minimoog and SEM envelopes are too slow, the acoustic element only comes out when you record a sound and speed it up 2x or 4x. Minimoog also has a deliberately 'blunt' attack, with a nonzero constant region. 2600 or Serge DSG (when used as an AD generator, with feedback for exponential shape) perhaps are fast enough. But this kind of 'crisp' attack is otherwise very hard to find. When it's there and you have a complex timbre or even with a simple pulse wave, the result can really fool you sometimes.

[ferran]

10 years-old thread

[noisewreck]

10 years-old thread

Indeed. But, it’s amazing to see how much things have changed isn’t it?