Author |
Message |
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Sun Feb 20, 2011 6:45 am Post subject:
Velocitizer: A different approach to adding velocity to ADSR |
|
|
I was reading this conference paper:
"Expressive Articulation for Synthetic Music Performances"
by Tilo Haehnel and Axel Berndt:
http://www.educ.dab.uts.edu.au/nime/PROCEEDINGS/papers/Paper%20P1-P5/P277_Haehnel.pdf
They look at Leopold Mozart's book on early classical/baroque articulations, and do some measurements of instruments. So I thought I'd have a stab at a circuit in SPICE. Here is it, in case anyone is interested.
Basically, we run the ADSR through a multiplier and then into the same kind of sine shaper diode circuit that VCOs use. The result is that at low velocity, a linear ADSR input comes out err linear. But at higher velocity, the input is bent into a curved shape, filling out the note without altering the timing.
Anyway, attached is the circuit, a sound simulation (synthedit), and an eg waveshape.
NOTE: The circuit digram and simulation has a wrong part. For building, the vanilla 1N4148 diode is supposed to be used. And R12 should be halved to 47k, to fit in with Eurorack typical signsl levels. The circuit assumes the EG input has a peak at at least 5V (up to 8V or even 10V) and that the Vel input is from 0V to 5V.
Description: |
|
Filesize: |
27.8 KB |
Viewed: |
686 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
Description: |
|
Filesize: |
254.68 KB |
Viewed: |
1081 Time(s) |
This image has been reduced to fit the page. Click on it to enlarge. |
|
Description: |
|
Download |
Filename: |
swell.mp3 |
Filesize: |
119.25 KB |
Downloaded: |
1158 Time(s) |
Last edited by Ricko on Fri Jun 07, 2019 1:58 am; edited 3 times in total |
|
Back to top
|
|
|
emdot_ambient
Joined: Nov 22, 2009 Posts: 667 Location: Frederick, MD
|
Posted: Mon Feb 21, 2011 3:48 pm Post subject:
|
|
|
It's a really simple circuit...would be interesting to hear it built and tested on a synth voice articulating it with something other than veolcity (like a sequencer output). _________________ Looking for a certain ratio since 1978 |
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Fri Feb 22, 2019 12:06 am Post subject:
Velocitizer |
|
|
Just to say that Elby Systems now have released this circuit as a kit and built, as Velocitizer, see http://www.elby-designs.com/contents/en-us/p1057_ASM322_-_Velocitizer.html It is a simple build, less than an hour all up: through hole, 4HP and my build worked first time.
Not only can it be used for adding velocity dynamics, where the shape as well as the level changes, it also works on audio signals, as a waveshape animator/chorus effect, as it turns out! UPDATE: video at https://youtu.be/EjKQMTWaeyY
Laurie from Elby has tweaked some components and values in his BOM. For example, in the schematic above I used 1N148 germanium diodes (by mistake!), while Laurie uses BAT42 Schottky diodes. In my built module, I just use vanilla 1N4148. (I think Laurie alters the value of a resistor too. Lots of room for variation.)
I have some demos and info up on Muff at
https://www.muffwiggler.com/forum/topic-214038.html Last edited by Ricko on Fri Feb 22, 2019 7:15 am; edited 3 times in total |
|
Back to top
|
|
|
MusicMan11712
Joined: Aug 08, 2009 Posts: 1082 Location: Out scouting . . .
|
Posted: Fri Feb 22, 2019 6:36 am Post subject:
|
|
|
Interesting article! One of the staples of electronic music experimentation has long been timbral variation which can be created in many different ways. This article provides insights into ways that timbral variations have been achieved by non-electronic musicians. The idea of simulating those techniques without relying on samples/multisamples of non-electronic instruments is intriguing! Thanks for posting the link.
--Steve
Quote: | Conclusion:
Articulation, if not reduced to a mere tone duration feature, offers a big potential for synthetic music performances. |
|
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Fri Mar 08, 2019 1:09 am Post subject:
|
|
|
OK, so now I have tried it with different diodes: Silicon (1N4148), Germanium (1N60), and Schottky (BAT85) which have progressively lower voltage drops so different responses in this fixed circuit.
Works fine with all of them, and each have slightly different feels and output levels, but I cannot really characterize them, or pick a favourite. I think the Schottky gives a sharper attack or something, but I cannot say. I will probably stick with the Germanium, merely because I am tired of unsoldering. It would be neat to have a switch to select between different diode types.
I have put up another little audio demo on Muff Wigger Eurorack forum (search Velocitizer), with the Velocitizer pinging a Low Pass Gate. That demo uses the Germanium. The LPG does not get excited by low CV signals, so it works well with the Velocitizer, giving more of a range than a normal VCA does. |
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Thu Mar 28, 2019 3:28 pm Post subject:
|
|
|
Here is an audio example of the Velocitizer played from a keyboard, to show how the velocity alters dynamics. (The LPG makes everything sound like a banjo, it seems?)
KCV ->VCO -> my pulse forming waveshaper -> LPG in -> out (slight reverb added)
Gate -> ADSR -> Velocitizer (germanium) ->passive mix -> LPG cv
Velocity -> Velocitizer
In this example, you can hear a variety of different note shapes: no knobs were twiddled all the variation is from playing.
One effect that is noticeable is that the sustain is set to 50%: from the action of the Velocitizer, it means that low velocity notes have an effective sustain of 25%, while notes played loud have an effective sustain of 75% (very roughly).
Another effect, sightly more subtle, is that the high velocity envelopes are fatter than the low velocity notes, not just louder, which is the primary intent of the circuit (distinguishing it from a normal 1Q multiplier.) This is not just that I am holding the notes longer, but that the louder velocity envelopes are soft-limited by the Velocitizer to give a fatter envelope. (The LPG adds another layer of dynamic variation on this, of course.)
Description: |
|
Download |
Filename: |
VelocitizerBanjo.wav |
Filesize: |
442.8 KB |
Downloaded: |
897 Time(s) |
|
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Tue Jun 04, 2019 7:34 am Post subject:
|
|
|
I see Elby has its version back in stock, and that Laurie has further adjusted the circuit.
He now has the diodes in a ladder, each in parallel with a resistor. This would give a more gradual effect, rather than the sharp knee. (You could use the same pcb to get the original design, if wanted.) |
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Wed Jun 05, 2019 8:07 pm Post subject:
|
|
|
And here is the graph for a "hotter" EG signal, of 6.6V, which is what my CEMvelope module provides.
It shows the wide range of responses possible by varying the different input levels.
I think it shows a couple of things:
First: the BAT43 Shottky diode kicks in early and fast (as expected), but the result is a poor dynamic range of 2 to 1 (loudest is only twice the volume of the softest), the velocity will not be smooth transition from soft to loud but more a velocity switch, and that change happen quite low in the velocity curve. So the effect will be that soft notes are dampened, while medium velocity notes have the same level as high velocity notes (though the fatness of the EGs curves shapes will be different between medium and high, but not much.)
A pretty interesting webpage in diode responses is at
https://jeelabs.org/2012/05/14/forward-voltage-drop-on-a-diode/
Consequently, I don't think the BAT43 is a good choice for this circuit, unless you have a very specific use in mind, such as using a gate to provide the Velocity level.
Second: there is no much difference between the Germanium and 1N4148 responses. So it seems the 1N4148 wins on price and availability (and consistent operation). They both give a much better dyamic range of 3:1 which is, IMHO, a good range for playing.
Third: the level of the EG clips the peak level of the EG (by intention). The rule of thumb is that the maximum level (volts) of the input EG should be greater than the maximum level of the velocity input. If the EG input is relatively lower, the effect will a very compressed signal, with all the dynamics squeezed down to low-velocity and less a range. For example, I have a MIDI/CV converter that produces a 10V maximum velocity signal: I need to pad it down to 5V (or even 4V) to get the widest response.
The Eurorack norms are "Control voltages, as produced by modulation sources like the LFO and ADSR, are typically from -2.5 V to +2.5 V (5 Vpp) for LFOs, and from 0 V to +8 V for ADSRs" and the Velocitizer (with the 1N4148 and 47k feedback) will operate with those, and should be useful just plugged directly in after the EG. You may care to put the CVs though attenuators first, if you want to vary the response and the range of control more finely, but you may find your keyboard controller has different velocity response curves which will save you the trouble.
Description: |
|
Filesize: |
28.96 KB |
Viewed: |
9780 Time(s) |
|
|
|
Back to top
|
|
|
cloudberry
Joined: Jun 30, 2009 Posts: 51 Location: uk
|
Posted: Sat Jun 05, 2021 8:34 am Post subject:
|
|
|
Thought I'd try this out. But, in the schematic, at the bottom of the string of diodes, we have "R8 1" Is that really 1 ohm?
I can't find an Elby schematic either. Do they not publish theirs? |
|
Back to top
|
|
|
Ricko
Joined: Dec 25, 2007 Posts: 251 Location: Sydney, Australia
Audio files: 27
|
Posted: Sat Oct 01, 2022 7:13 pm Post subject:
|
|
|
Yes, that 1R resistor can be left out. It was probably there as a holdover from playing with different curves. |
|
Back to top
|
|
|
cloudberry
Joined: Jun 30, 2009 Posts: 51 Location: uk
|
Posted: Sun Oct 02, 2022 1:16 am Post subject:
|
|
|
Thanks! |
|
Back to top
|
|
|
|