I have to admit that after ignoring halfpedaling on my DP and with the free sample libraries that I've created, I really need to attempt to implement it. I finally sat down at my acoustic and realized that I totally half pedal all of the time.
Before I try to implement it, I need to verify I understand it correctly by bouncing it off the community. I know there are many pedaling techniques, and I think I probably should narrow this down to the following instances:
1. trigger the sample (play the note with no damper pedal), add half pedal to get more resonance, but not as much resonance as a full pedal down. 2. half pedal first, trigger the sample, get resonance somewhere between no pedal and full pedal
For people with a half pedal VST setup and pedals that support it, is that what is going on? Bear in mind, I'm asking about what the VST's are doing, not what a real piano would do....
Incidentally, I found this youtube and plan on making a pedal controller so I can have a try.
Half-pedaling is not so much about resonances, I'd even go as far as to say the resonances don't matter at all. It's about using it for a very fine control of legato between the notes. With a fully pressed pedal, the note sustain long. With half-pedaling the sustain becomes shorter and you can precisely control that up to a point where the notes have almost no sustain and get muted quickly. And this needs to be varied, for instance you start with a fully pressed pedal, then briefly you can raise the pedal to shorten the sustain but then press it again, etc. using your aural feedback. I suggest that you take care of how you program/model the sustain and release of your samples through envelopes, etc. and leave the resonance for a later point.
No digital piano nor VST is able to correctly represent the actual half-pedaling piano technique.
Most VSTs don't care about pedaling. They are single note recordings of pianos with room microphones, which give you the impression of a recording on that piano in that location when fed with a MIDI file. All you need you need for that is sustain your samples on the appropriate MIDI CC (64). "Half-pedaling" usually shortens the sustain if the CC value isn't at full (127).
I recommend to pay attention to repedaling, which is still more important than halph pedaling. Generally, this is a quick pair of pedal off & pedal on events, after which the sustained note(s) remain sustained, but are quieter. In acoustic pianos this produces deep artistic effects, especially pedal pulsations.
No digital piano nor VST is able to correctly represent the actual half-pedaling piano technique.
Most VSTs don't care about pedaling. They are single note recordings of pianos with room microphones, which give you the impression of a recording on that piano in that location when fed with a MIDI file. All you need you need for that is sustain your samples on the appropriate MIDI CC (64). "Half-pedaling" usually shortens the sustain if the CC value isn't at full (127).
That's an oversimplification.
Pianoteq takes some care to modify the tone of a single note, repeated, as the pedal goes from "dampers down" to "dampers up". It's not just the sustain time that changes:
. . . more "bloom" is added, from the resonating harp.
Pianoteq takes some care to modify the tone of a single note, repeated, as the pedal goes from "dampers down" to "dampers up". It's not just the sustain time that changes:
. . . more "bloom" is added, from the resonating harp.
Of course, it's a polyphonic synth. It does something with damper/expression pedal messages.
What aspects based on your experiences are omitted?
This is a good question/observation because we have a knack for generalized omission(s) when the truth is we, the collective, can indeed omit the cream but still add two sugars (metaphor), and universal omissions imply that it’s either all or nothing, black or white, good or evil, etc….
I look forward to a world where omissions are local and not global.
So what you're saying is that the half pedaling affects the envelope more than the timbre? That does sound complicated.
Envelope changes are the most basic implementation of half-pedaling. The envelope change is different for bass notes than for treble notes while the always-undampened notes aren't affected at all. Heavy bass strings are not affected as much by short bursts of half pedal while lightweight strings can be dampened for more clarity, especially on upright pianos.
All that mostly isn't really needed on a digital instrument though (perfect in tune, with shorter sustain and artificial resonance effects), so you can keep the pedal down all the way with a slight lift on harmonic changes.
Working on a new free library right now. I think it will blow your mind (for a free one especially). Considering to add the half pedaling, but need to understand it better.
I don't think I can look at other VST implementations, but I can observe behavior, which is what I'm asking on this thread. Before I get my hardware in and get to soldering and mod'ing arduinos, I figure I can use my current VST's to experiment with CC64 curves.
Any suggestions on CC64 curves I should try? So far I'm not getting anything on Grandeur. Or someone send me a MIDI file so I can observe the behavior?
There is a half pedal option in Grandeur, so it should respond to multiple CC64 messages.
I just tried Garritan, and just drew CC64 curves in the DAW and this is what I detected:
* CC64 above 65 didn't make a difference (at least by my ears - maybe I should do a spectral/gain analysis) compared to full pedal down * CC64 in the range just below 65: the notes trailed off sooner than full pedal down but longer than full pedal up. This must be the half pedal range where the envelope changes. * CC64 in the range above and closer to 0: sounds like full pedal up to me.
I did not notice any discernable timbre changes - but again, maybe I should do a spectral analysis. I would expect the timbre change because certainly my acoustic changes in timbre with the dampers partially/nearly engaged compared to fully engaged.
Pianoteq takes some care to modify the tone of a single note, repeated, as the pedal goes from "dampers down" to "dampers up". It's not just the sustain time that changes:
. . . more "bloom" is added, from the resonating harp.
Of course, it's a polyphonic synth. It does something with damper/expression pedal messages.
I have a few polyphonic synths, and I understand pretty well what they do.
Pianoteq goes well beyond their capabilities, _for emulating an acoustic piano_. It's special-purpose, not general-purpose.
As an example, I don't think you'll find a polyphonic synth (in the normal sense of that word) that emulates "string resonance":
. . . Hold down C3, let the sound die out;
. . . Strike G4 staccato, and listen to the long-continued sustain . . . caused by the open C3 string.
Other VST's emulate the same behavior. They are _not_ simple "sample-playback" engines -- not any more.
My Korg simulates halfpedaling by allowing releases and resonances to sustain a little if the pedal is quickly put back down after note striking. So it’s partially implemented using a time from key release to pedal down method.
Pianoteq takes some care to modify the tone of a single note, repeated, as the pedal goes from "dampers down" to "dampers up". It's not just the sustain time that changes:
. . . more "bloom" is added, from the resonating harp.
Of course, it's a polyphonic synth. It does something with damper/expression pedal messages.
I have a few polyphonic synths, and I understand pretty well what they do.
Pianoteq goes well beyond their capabilities, _for emulating an acoustic piano_. It's special-purpose, not general-purpose.
The Modartt patent about Pianoteq is about an additive synthesis where some partial parameters are computed and afterward, samples are computed as a sum of sinus functions.
So it's a digital synthesizer. Before sampling became mainstream, synths have been used all the time to emulate "natural" sounds with more or less accuracy.
. . . The Modartt patent about Pianoteq is about an additive synthesis where some partial parameters are computed and afterward, samples are computed as a sum of sinus functions.
So it's a digital synthesizer. Before sampling became mainstream, synths have been used all the time to emulate "natural" sounds with more or less accuracy.
You are correct. Of course, under that broad definition, _any_ VST -- or any digital emulation of any instrument, hardware- or software-based -- is a "digital synthesizer". And for piano, they're all polyphonic.
When I hear "polyphonic synth", I naturally think of a microKorg or Roland Gaia or DS88 or Yamaha Motif -- not a Kawai NV10. But that's my mistake.
For people with a half pedal VST setup and pedals that support it, is that what is going on? Bear in mind, I'm asking about what the VST's are doing, not what a real piano would do....
Dore,
I guess you already understood that, if you want to build something serious, this is a wrong strategy. Most existing piano plugins either do not support half-pedaling, or keep their implementation secret, or provide half-built would-be tools for supporting it via velocity curves etc. Only a small number of plugins have a usable implementation: Garritan CFX, Ravenscroft 275 and Synthogy ACD among those, and each of them has its own implementation of half-pedaling, depending on its specific architecture.
I think you should start with understanding the physical phenomena under the string damping, which is not as daunting task as it seems to be.
What is, actually, half-pedaling, re-pedaling, catch-pedaling etc. It's simply a partial reducing of the energy of the oscillating string(s) using some very soft felt pads. How the pianist applies the damper pedal is another question, which is related to his (her) art. The decca resonance is affected post-factum, after the string oscillation energy is reduced.
Given a sounding sample (samples), it is possible to partially reduce their energy according to some envelope changes. The parameters of the envelope changes should depend A) on the note, B) on the depth of the pedal press, and C) on the duration of the press. This is almost all you need to know to start some implementation. Forget about arduino: there exist a host of inexpensive pedals for this, of continuous and on-off type.
I think you misunderstand the purpose of the Arduino. I don't have a pedal or dp that supports continuous cc64. So I cannot experience half damping because I do not have the hardware. Short of buying a new dp or expensive equipment, I've decided to buy a Yamaha pedal and make a "pedal to midi" hardware controller using an Arduino as a microcontroller.
Point taken about the the physics of damper pedals though. I've already observed garritans behavior, and if it's one that people like, then I may like it too when I can actually control it with my foot. Ive written some pseudocode to mimic their behavior . Not sure I can implement it in kontakt or sfz yet.
I think you misunderstand the purpose of the Arduino. I don't have a pedal or dp that supports continuous cc64. So I cannot experience half damping because I do not have the hardware.
So you're developing a virtual piano without actually having the hardware to play it?
I think you misunderstand the purpose of the Arduino. I don't have a pedal or dp that supports continuous cc64. So I cannot experience half damping because I do not have the hardware. Short of buying a new dp or expensive equipment, I've decided to buy a Yamaha pedal and make a "pedal to midi" hardware controller using an Arduino as a microcontroller.-
I was thinking that this cannot be the first time anyone has had a need for a MIDI pedal. And sem found three commercial products to suit that need ... each the size of a thumb drive. Superb, eh?
I think you misunderstand the purpose of the Arduino. I don't have a pedal or dp that supports continuous cc64. So I cannot experience half damping because I do not have the hardware. Short of buying a new dp or expensive equipment, I've decided to buy a Yamaha pedal and make a "pedal to midi" hardware controller using an Arduino as a microcontroller.-
Thanks. Those are helpful. But seeing as I already have Arduinos lying around doing nothing, and I can hack the arduino code to optimize it as I like, and its also another teaching opportunity to get my kids to help with their coding/hardware, I'm choosing the DIY avenue. Definitely if I hit a big roadblock, I'll probably purchase one of those.
I don't post that often, but if you've seen the work that I've done, you'll notice that I take a real DIY approach to everything. I have a goal to create great piano sample libraries investing mostly time rather than money. I could invest more money, but I get the fulfillment out of doing it myself. More than half of my microphones are hand built, and I built them especially for sampling pianos. I could buy more commercial libraries, but I want to build it myself. I want to see what I can achieve with almost nothing.
Dore, Bechstein Digital is Kontakt and supports half pedal. That any help?
Hi Mark-
Since I don't own all of the libraries that support half pedal, anyone who can describe the actual behavior of their supported VST's is really helpful.
You'd have to use a DAW like I did to go through varying levels of CC64 on release and after release to see if there are differences.
I know that this is work, and I don't expect anyone to volunteer this help. There is no published open source code for this as far as I know, but if I can understand the behavior, I can attempt to script it. No guarantees on if I have the talent to write the script though.
In terms of Garritan, the curves that NP pointed me to basically confirmed what I thought. Here is the behavior as I see it:
On release, if CC64 is 64-127, then it's a normal full sustain if CC64 is 0-31, then it's a normal release if CC64 is 32-63 (half pedal range), then the release time will vary depending on the value. Lower value is closer to normal release, higher value is closer to full sustain. Curve is described basically in NP's thread.
Furthermore, if the release happened when CC64 was in the half pedal range, and CC64 then changes: if CC64 is 64-127, then keep the previous gain and continue with full sustain time if CC64 is 0-31, then it's a normal release if CC64 is 32-63 and value increased, do nothing if CC64 is 32-63 and the value decreased, shorten sustain according to curve
So you're developing a virtual piano without actually having the hardware to play it?
With the arduino pedal he will have the hardware.
A Yamaha PF85 is enough to develop a virtual piano instrument? With no action remotely close to any current day piano controller? Is it doing more than 8 steps of velocity sensor resolution?
JoeT: I am sure that Dore appreciates your concern, but if you go back and read a bit about Dore expertise on creating sample libraries, you may understand that you are kind of questioning if Babe Ruth is fit for sports...
A Yamaha PF85 is enough to develop a virtual piano instrument?
Why not? You play a key and you get a MIDI event. It is a weighted and velocity sensitive action. Some people even say that they like the "legendary" Yamaha KX-88 master keyboard with basically the same action. (Plus aftertouch)
Originally Posted by JoeT
With no action remotely close to any current day piano controller?
From the player perspective the Yamaha Action Effect key action is remotely close to a basic level current day digital piano action.
Have you played it yourself?
Originally Posted by JoeT
Is it doing more than 8 steps of velocity sensor resolution?
I have a PF80. I can check some day. But probably yes. Why wouldn't it?
So I just played around with CC64 in Noire, and the behavior is similar to Garritan, but with some notable changes:
* The half pedal range is much wider, not just 32-64 * Once the finger releases the key, the release time is set depending on CC64 value, there is no release time changes after the key is released.
On one hand, this probably makes Garritan better, because after key release you can still cut the release time fractionally if you're pedaling between 32-64. Noire will only respond again with a fast release when CC64 goes below 10. It also confirms to me that you probably cannot change the release time AFTER the release has been triggered by finger release (in Kontakt) - that was bugging me last night. On the other hand, it gives me hope that I can probably write a script or something like Noire's behavior.
Some players, such as VI Labs U and Ravenscroft, allow to adjust the pedal velocity curve to make half pedaling effect configurable by user.
More accurate statement would be that all 5 VI Labs pianos allow to set the minimum and maximum pedal velocity where the half pedal effect is triggered. All Synthogy and VSL Synchron pianos also allow to set the half-pedal min/max velocity range. I assume they all implement the linear velocity curve between min and max values.
I do not know any other sampled library that offers this option.
Does the half pedal velocity minimum setting have a boundary, or is it zero? Does the half pedal velocity maximum setting have a boundary, or is it 127?
Most importantly, do you know what the default is?
Does the half pedal velocity minimum setting have a boundary, or is it zero? Does the half pedal velocity maximum setting have a boundary, or is it 127?
Most importantly, do you know what the default is?
Whatever you want between 0-127. I don't remember what defaults are.
Could a truly continuous (all values between 1-127) result in MIDI timing delays/overload? I wonder if a lot of pianos only send discrete values for pedal as a way of optimizing the MIDI stream.
I'm not sure, but that is exactly why I'm going to make my own arduino controller. Toward the end of the vid that I posted, the guy mentioned that he had to adjust the code so that data send frequency was lower. I want to be able to control that with the hardware.
it's also why I think it would be easier to just lookup the current CC64 on finger release and issue a different decay, ignoring all other data
Maybe take a look at the NI forums for some ideas of how developers deal with this in KSP. Inevitably you will bump into a EvilDragon, a developer who works with several VI makers, including NI; he shares some inside information and tips for pedal scripting, but also noted,
Quote
Yeah for some things they are just not possible with sample libraries, and Pianoteq would be your next best bet. ESPECIALLY if you're a pro pianist.
Maybe take a look at the NI forums for some ideas of how developers deal with this in KSP. Inevitably you will bump into a EvilDragon, a developer who works with several VI makers, including NI; he shares some inside information and tips for pedal scripting, but also noted,
Quote
Yeah for some things they are just not possible with sample libraries, and Pianoteq would be your next best bet. ESPECIALLY if you're a pro pianist.
It's obvious, that with recording sample players like Kontakt, you cannot simulate a piano, you can only simulate a recording, by mixing recording samples together and hope that the result sounds somewhat believable.
That doesn't stop people from trying as hard as they can on their search of the digital piano holy grail.
Thanks. Those are helpful. But seeing as I already have Arduinos lying around doing nothing, and I can hack the arduino code to optimize it as I like, and its also another teaching opportunity to get my kids to help with their coding/hardware, I'm choosing the DIY avenue. Definitely if I hit a big roadblock, I'll probably purchase one of those.
I don't post that often, but if you've seen the work that I've done, you'll notice that I take a real DIY approach to everything. I have a goal to create great piano sample libraries investing mostly time rather than money. I could invest more money, but I get the fulfillment out of doing it myself. More than half of my microphones are hand built, and I built them especially for sampling pianos. I could buy more commercial libraries, but I want to build it myself. I want to see what I can achieve with almost nothing.
In fact I was curious if there was sth wrong with these. I appreciate your DIY spirit. Of course sometimes there's the issue what to spend time on. And sometimes packaged products happen not to do what you want them to and often can't be fixed, unlike if you DIY.
Could a truly continuous (all values between 1-127) result in MIDI timing delays/overload? I wonder if a lot of pianos only send discrete values for pedal as a way of optimizing the MIDI stream.
There would be no need to continuously send the value, if it doesn’t change.
And typically in these Type of things you would build a little dead zone so small movements wouldn’t trigger a new message.
I don't post that often, but if you've seen the work that I've done, you'll notice that I take a real DIY approach to everything. I have a goal to create great piano sample libraries investing mostly time rather than money. I could invest more money, but I get the fulfillment out of doing it myself. More than half of my microphones are hand built, and I built them especially for sampling pianos. I could buy more commercial libraries, but I want to build it myself. I want to see what I can achieve with almost nothing.
I also mount multiple capsules within one mic body, making XY arrays . Jules has a few options that I have made variations on. In fact in the future, I'll use these mics and samples to allow the user to change the mic pattern that I place inside the piano, which changes the sound considerably.
It sends all values from 0 (fully up) to 127 (fully down) for all 3 pedals, sending changed values at a maximum rate of 100Hz. So if you press a pedal down really fast you don't get all the values from 0-127, you just get a new value every 10ms.
Using 'MIDI Loupe' on Mac:
02:56:02:517: Control message: 64, 13 02:56:02:527: Control message: 64, 37 02:56:02:537: Control message: 64, 60 02:56:02:547: Control message: 64, 84 02:56:02:557: Control message: 64, 107 02:56:02:567: Control message: 64, 127
Continual manipulation even of all three pedals at this data rate would consume less than a third of the bandwidth of a standard serial MIDI connection. (Though this data was collected from the A88 USB MIDI connection, not the 5 pin serial).
It sends all values from 0 (fully up) to 127 (fully down) for all 3 pedals, sending changed values at a maximum rate of 100Hz. So if you press a pedal down really fast you don't get all the values from 0-127, you just get a new value every 10ms.
Using 'MIDI Loupe' on Mac:
02:56:02:517: Control message: 64, 13 02:56:02:527: Control message: 64, 37 02:56:02:537: Control message: 64, 60 02:56:02:547: Control message: 64, 84 02:56:02:557: Control message: 64, 107 02:56:02:567: Control message: 64, 127
Continual manipulation even of all three pedals at this data rate would consume less than a third of the bandwidth of a standard serial MIDI connection. (Though this data was collected from the A88 USB MIDI connection, not the 5 pin serial).
It sends all values from 0 (fully up) to 127 (fully down).
Thanks! I love crowd sourcing information
If it is useful for you to know: I tested the Roland DP-10 (in continuous mode) and Kawai F-10H both connected through an Audiofront MIDIExpression (pedal to USB MIDI) interface. For both pedals, I got a very neat 0--127 pedal position range as well.
Audiofront's control software lets you choose the update frequency (the number of updated-value MIDI messages per unit of time).
My brother uses the DP-10 pedal connected directly to his Akai Professional MPK261 MIDI controller and gets the neat 0--127 as well.
Since you belong to the DIY people and want to do something in the piano pedal area, may I suggest to you to enhance the common DP pedal concept as follows.
Damper pedals in real acoustics are essentially based on the =force= (which can be measured in kg or N), not on the =velocity= of the pedal motion. Such a pedal typically works as follows:
A) It lifts a mechanical part of the piano action, which contains about 60 dampers. (The upper register strings typically are not damped);
B): After that, using some additional force, it presses all the dampers onto the strings in order to damper them.
The first force (lifting the dampers) is a constant force of weight lifting. Nevertheless, for mimicking it in the DP pedal, a light spring can be applied, up to the dampers-strings contact position.
The second force (pressing the dampers) should be much stronger, and it should allow a better control by the pianist. For mimicking it in a DP pedal, from the start of the (B) position, a stronger spring can be applied. Its pressure can be measured using a dynamometer. I believe there exist inexpensive arduino dynamometers.
The idea is to mechanically implement phases (A) and (B) and, using the arduino, send to the DAW some measure (in kg or N, normalized to values 0-127) of pressing the second (stronger) spring.
This can be easily implemented as follows. Take some common DP pedal and do not apply its MIDI. Without disassembling it, simply insert some stronger string (or even a rubber ball) and a dynamometer between the metal part of the pedal and its bottom part, leaving some free (unmeasured) motion of the pedal before it touches the dynamometer. Then send the value of the dynamometer into the DAW.
I apologize if anybody did this before (I do not have such info). For me as a pianist, this would look less amateuristic than applying the velocity of the DP pedal motion for controlling the sustain of piano strings.
In your part (A) ... when pressing the pedal, the dampers are lifted. They are NOT pressed onto the strings. Rather, they are lifted AWAY from the strings.
You are right MacMacMac, thank you: (B) should be measured =before= the stage (A) occurs. Anyway, the dynamometer measure seems to better represent how the pianist controls the strings sustain by the foot pedal, and the sustain pedal of real acoustics certainly has 2 distinct stages: (B) sensitive, (A) non-sensitive. My apology for the mistake...
Andrew_G: in my opinion, you are incorrect on several accounts...
Firstly, as Mac noted, there is simply no way to "press further" the damper onto the strings. That could be an interesting 4th pedal for muffled notes using a separate set of dampers, but Mr Cristofori did not think of that; maybe S&Sons is receptive to the idea.
Secondly, the MIDI value of the pedal does not correspond to the velocity but to the position of the pedal (ie, distance of the damper to the strings). To calculate velocity, the sound engine needs to have a continuous pedal sending midi messages at fixed frequency. I doubt that piano engine takes this into account (maybe it should), although many calculate key return speed which is the same as the drop speed of the key damper (if the pedal is not pressed down, of course).
Thirdly, on the way down dampers act by gravity and cannot go faster than what is caused by its own weight. The pianist (Horowitz and his cousins) may be able to reduce speed of descent, but cannot accelerate it by any means.
Fourth, as far as I understand, no springs are actually required in acoustic piano pedals, and they exist simply to add resistance and improve control of pedalling. So, the increasing weightness to which you call stage A and B (???) is simply an effect of gravity cause by the continuous raising of the entire damping mechanism.
There is no stage B. You press the pedal and the dampers rise. You release the pedal and the dampers fall. The applied pedal force need only overcome the weight of the damper mechanism (and of the spring under the pedal, if any).
Originally Posted by Andrew_G
You are right , thank you: (B) should be measured =before= the stage (A) occurs. Anyway, the dynamometer measure seems to better represent how the pianist controls the strings sustain by the foot pedal, and the sustain pedal of real acoustics certainly has 2 distinct stages: (B) sensitive, (A) non-sensitive. My apology for the mistake...
1. First and foremost the position readout of a continuous pedal is important. A sound engine/VSTi should base its 'damper' (or sostenuto, or una corda/soft) behavior on that number.
2. The pedal velocity (the change in pedal position readout over time) is *also* useful. For instance VI Labs Modern U makes good use of this to vary its pedal mechanism sound (very subtle on slow pedal movement; a distinct 'clunk' on a swift movement to full-up or full-down). I love such 'realism' details and love such caring attention to detail.
I agree I confused the order of events. However, I think that you oversimplify this. What is half-pedaling in an acoustic? It's the art of balancing the soft tips of the dampers on the edge of their touch points with the strings, so this is not simply an on-off control, our leg feels and controls the pressure on the pedal, along with its position. Besides, since there can be imperfections in the felt softness, we can achieve unusual sound effects.
Originally Posted by MacMacMac
There is no stage B. You press the pedal and the dampers rise. You release the pedal and the dampers fall. The applied pedal force need only overcome the weight of the damper mechanism (and of the spring under the pedal, if any).
Originally Posted by Andrew_G
You are right , thank you: (B) should be measured =before= the stage (A) occurs. Anyway, the dynamometer measure seems to better represent how the pianist controls the strings sustain by the foot pedal, and the sustain pedal of real acoustics certainly has 2 distinct stages: (B) sensitive, (A) non-sensitive. My apology for the mistake...
Since you belong to the DIY people and want to do something in the piano pedal area, may I suggest to you to enhance the common DP pedal concept as follows.
Damper pedals in real acoustics are essentially based on the =force= (which can be measured in kg or N), not on the =velocity= of the pedal motion. Such a pedal typically works as follows:
A) It lifts a mechanical part of the piano action, which contains about 60 dampers. (The upper register strings typically are not damped);
B): After that, using some additional force, it presses all the dampers onto the strings in order to damper them.
The first force (lifting the dampers) is a constant force of weight lifting. Nevertheless, for mimicking it in the DP pedal, a light spring can be applied, up to the dampers-strings contact position.
The second force (pressing the dampers) should be much stronger, and it should allow a better control by the pianist. For mimicking it in a DP pedal, from the start of the (B) position, a stronger spring can be applied. Its pressure can be measured using a dynamometer. I believe there exist inexpensive arduino dynamometers.
The idea is to mechanically implement phases (A) and (B) and, using the arduino, send to the DAW some measure (in kg or N, normalized to values 0-127) of pressing the second (stronger) spring.
This can be easily implemented as follows. Take some common DP pedal and do not apply its MIDI. Without disassembling it, simply insert some stronger string (or even a rubber ball) and a dynamometer between the metal part of the pedal and its bottom part, leaving some free (unmeasured) motion of the pedal before it touches the dynamometer. Then send the value of the dynamometer into the DAW.
I apologize if anybody did this before (I do not have such info). For me as a pianist, this would look less amateuristic than applying the velocity of the DP pedal motion for controlling the sustain of piano strings.
Just waking up here in California. Wow, I see all of your points here, and from what I gather, making a mechanical damper pedal that actually FEELS like a real damper (as well as reacts like one) sounds like a project I can work on after I finish my sample libar(ies). This one really interests me because I'm a mechanical engineer by training.
One step at a time though - I have to just get the current implementations working (simple potentiometer, arduino, CC64) and see how that feels.
Just an update, Amazon delivered my Yamaha pedal today, I soldered everything together and am playing around with it:
Ardiuno Nano fits right inside the housing, connected to a mini-USB cable directly to my computer. The Arduino firmware worked the first time I plugged it, but because there is a rotational potentiometer inside the pedal, the pedal mechanism prevents the pot to output a range of more than 16-100 (rather than 0-127). I need to go into the Arduino code to calibrate it so that the lowest number is lower than 15, which is I think Noire's threshold for pedal up. I'll have to see if I can get it to work...
I'd assume that the three pedal to USB/Midi converters mentioned earlier in this thread come with some calibration software, but if they don't, well - good thing I'm DIY'ing.
As said before, I am using the MIDIExpression interfaces with control software as found on audiofront.net. That one has excellent calibration and customization provisions.
What I'd recommend for your DIY pedal project, is to get an output range from 0 -- 127. Different VSTis may require the full range, e.g. only an actual 0 meaning no sustain processing at all. Also, I'd recommend to have a tiny bit of travel near the upmost position and a tiny bit of travel near the downmost position both doing nothing.
To (hopefully) illustrate what I mean: in your software re-map/re-spread the pedal's output range of 16 -- 100 (as you apparently derived from the arduino's voltage divider input pin) such that e.g. only 25 -- 90 is proportionally mapped to 0 -- 127.
In my experience, the 'inactive' top and bottom bits of travel have two major advantages: -> It feels more natural playing piano with the pedal that way. -> It prevents 'flutter' of the min. and max. values, e.g. if the pot varies slightly with temperature, moisture, age and thus sometimes goes no lower than 18 instead of 16, and/or gives like 98 or 101, instead of 100, you'll still get that solid 0 -- 127 as the useable range.
Hope my attempt at explaining this works and helps.
Dore: can you confirm that the arduino code accepts the connection of 3 similar pedals to a single board, and output of 3 different CCs through the same usb connection?
I've only connected the one pedal, but the video shows how to connect two additional pedals. However, the two additional pedals are on/off switches and not potentiometers. I can see inside the code that it does support that particular use case - 1 continuous damper with a pot interface, and two switches (digital). I believe that the CC number is also configurable (3 different ones) since the purpose was to create a 3 pedal box in which the 3 pedals were different CC's. I just don't understand C++ well enough to see exactly where it is for now.
The downside is that after calibrating and making a final program for the arduino, it's not ridiculously easy to change the CC number without using a translation software on your computer, or diving into the arduino code and then reuploading the firmware. So the solution is not ridiculously flexible. I've been shuttling mine from my work computer to my piano computer to keep recalibrating.
But I also can theorize that it wouldn't be so difficult to just copy/paste/edit the code to support as many switches and pots that your arduino supports. For my arduino nano, that would mean as many as 8 (?) pots and 12 (?) switches, I guess as power allows.
^^ and ^, Yes, as long as there are enough analogue (voltage divider measurement) input pins on the Arduino, three similar pedals are totally possible.
Hey, maybe I should just buy a bunch of pedals and use all of the inputs to make an organ pedal controller for my piano! I could trigger the lower notes of a Bosendorfer Imperial!
Hey, maybe I should just buy a bunch of pedals and use all of the inputs to make an organ pedal controller for my piano! I could trigger the lower notes of a Bosendorfer Imperial!
You could build a pedal piano! Digital version of Doppio Borgato.
[...]as long as there are enough analogue (voltage divider measurement) input pins on the Arduino[...]
...and if there aren't enough inputs -- or if you decide you want to support 12 pedals or something --then there are some very cheap little breakout boards around with a 74HC4067 analog multiplexer that'll let you scan up to 16 inputs with 1 ADC. (Needs 4 digital outs to drive the select lines though).
Thanks for the info. I am well familiar with arduino but not with its midi implementation. But I had a brain freeze because 3 pedals use only 1 midi channel with messages at 3 different CCs: using 1 CC or 3 CCs does not significantly affect the code.
[...]as long as there are enough analogue (voltage divider measurement) input pins on the Arduino[...]
...and if there aren't enough inputs -- or if you decide you want to support 12 pedals or something --then there are some very cheap little breakout boards around with a 74HC4067 analog multiplexer that'll let you scan up to 16 inputs with 1 ADC. (Needs 4 digital outs to drive the select lines though).
I have a few of those multiplexers because awhile ago I was thinking of figuring out how to incorporate the Cybrid idea directly into my Mason & Hamlin action. Didn't get very far though, and then I got into sampling, and well, now that's all I do.
So after about a week of futzing, I think I have something working.
Here's a question: Why do commercial libraries have a on/off switch in the settings for half-pedaling? Now that I have both types of pedals (continuous and on/off), all of my half-pedal VST's respond properly to both when half pedaling is turned on.
Are there on/off pedals out there that don't send only 0 or 127?
I'm asking because I'm trying to decide if I should code such an option.
Sometimes this exists because the VST player was originally released without half-pedal support, and it was added later so presumably they wanted to keep an option for folks who liked it the old way).
Maybe another reason is for lower resource usage on the PC, or less saturation of the MIDI stream?
I'm also realizing that some people with continuous damper pedals may not like the half pedal ranges, so its better they can just turn it off if desired.
Here's a question: Why do commercial libraries have a on/off switch in the settings for half-pedaling?
Most libraries see piano player only as niche market. Their main market is digital music producers who do not care about half pedal, but at the same time may need to run dozens or sometimes hundreds of VI simultaneously. Turning pedals, half-pedals, release samples, noises, etc., off reduces load on CPU and improves their productivity.
If I understood it correctly, when half-pedalling switch is turned off, the values then just becomes 0 for unpressed damper and any value number other than zero is damper pressed (1-127)?
If I understood it correctly, when half-pedalling switch is turned off, the values then just becomes 0 for unpressed damper and any value number other than zero is damper pressed (1-127)?
I would expect that slight pressure is also treated as 0.
Say, 0-20 = 0, and 21-127=1. Exact ranges are library dependent.
I'm also realizing that some people with continuous damper pedals may not like the half pedal ranges, so its better they can just turn it off if desired.
There are libraries that allow to configure tha half-pedal range. VSL Synchron, VI Labs, Synthogy come to mind.