Polyphonic Aftertouch (or Better) should be standard

[ThomasHelzle]

Not sure why what it's called should be such a big deal? Even the established variations are rather clear and straight forward.

I would agree that the actual implementation is often the biggest problem so far. Many keyboards have a pretty small range of pressure that can be used, so you often can't really modulate it delicately - which is the whole point.

I'm unsure if the classical keyboard really is the best place for Polyphonic Aftertouch anyway. When I tested the Qunexus, I was amazed by how sensitive it is. You can touch it very softly and already get it to recognize the touch and then have a nice range of pressure for modulation. But it's not really a "classical" keybed.
Since the classical keyboard needs to be hit rather hard already for velocity, somehow I think it's adverse by nature to finely modulated touch action after. Or the pressure sensor would have to be off for the initial press and only become active with a slight ramp afterwards?
Would be actually interesting to come up with some new concepts here.

But basically: the piano-style keys were originally created for more or less on/off kind of action with only velocity as variable parameter. For fine modulations something less rigid may be better anyway.

Interesting times

Tom

[zerocrossing]

People are not aware of why and how to use channel-aftertouch, how can they tackle poly-aftertouch? Nobody is going to seek it, and therefore create a buzz for manufacturers to take notice.

Perhaps, it only leaves synth anoraks and geeks, those who know what CS80, SQ80, T8, MK88 are capable of, and those who have an unhealthy obsession with Vangelis' early music (me! lol). But nobody is going to cater to us...or will they?

So true. When biggies like Korg (I'm looking at you KingKORG) offer a $1300 synth without aftertouch, and the new Nord Leads not only do not have aftertouch enabled keyboards but won't even respond to aftertouch... what hope is there for poly aftertouch (or poly pressure, whatever you want to call it) to come back in style? The truth is that most keyboard players these days probably sequence 90% of their stuff and use knobs or automation to modulate the sound in real time. I don't think much traditional playing goes on that would require aftertouch so that's why manufacturers don't include it.

Sad really. My old Ensoniq TS-10 had it and it was lovely. I'm not even a good keyboard player but as a guitarist I could really appreciate it in the same way I can sculpt the sound with my fingers on a fretboard.

[mztk]

a lot of modern synth don't even have sophisiticated keyscaling options
like those available on 80s FM synths. you're tweaking something and
you get to the bit where you want a bit of...aaaw, none on here...

[zerocrossing]

Hey, here's something interesting...

https://www.youtube.com/watch?v=DjqA_fio23E#t=25

Sadly, I can't seem to find this Max patch...

Still, if I could find a TS-10 at a good price I think I'd ditch my Novation Remote 61 for it.

[thetechnobear]

his website seems to 'having issues' but you can download using: http://sadfactory.com/assets/downloads/ ... 1.2.maxpat

if you just want poly pressure playback (normal), then PXT-General has this.

[himalaya]

a lot of modern synth don't even have sophisiticated keyscaling options
like those available on 80s FM synths. you're tweaking something and
you get to the bit where you want a bit of...aaaw, none on here...

You know? You have touched on a subject matter which is very important to me. In fact I was planning to start a petition for all synth plugin devs to start looking at key-scaling, to highlight how the traditional implementation is absolutely archaic and insufficient, and how to move forward. Your mention of the Yamaha FM synths is a great example, since these synths had good key-scaling features.

This is OT, but I feel like starting a separate thread to highlight this issue.

[himalaya]

So true. When biggies like Korg (I'm looking at you KingKORG) offer a $1300 synth without aftertouch, and the new Nord Leads not only do not have aftertouch enabled keyboards but won't even respond to aftertouch... what hope is there for poly aftertouch (or poly pressure, whatever you want to call it) to come back in style?

And imagine this...back in the late 1980s Kawai released the K1 synth which retailed at £595, and it had....channel aftertouch. Same with Casio's more expensive VZ-1 synth.

Today, none of the sub £1000 synths from the three main Japanese synth companies offer aftertouch.

[JCJR]

Since the classical keyboard needs to be hit rather hard already for velocity, somehow I think it's adverse by nature to finely modulated touch action after. Or the pressure sensor would have to be off for the initial press and only become active with a slight ramp afterwards?
Would be actually interesting to come up with some new concepts here.

But basically: the piano-style keys were originally created for more or less on/off kind of action with only velocity as variable parameter. For fine modulations something less rigid may be better anyway.

Tom

Those are good points. I doubt that any alternate keyboard layout will get traction. As I mentioned earlier, inventors have been trying to popularize "improved keyboard layouts" for at least 200 years. Lots of inventors. But its not impossible that one might eventually get traction.

You are spot-on about the pressure threshold problems, for mere mortals to exploit poly AT.

The only keyboard I had with poly AT was the EPS. On that one, the poly AT sensing was a global keyboard setting. Aftertouch response scaling was a synth modulation destination in the patches, so if you are playing a gig you could leave Poly AT turned on in the keyboard globals, then program inappropriate patches to ignore the messages. Though poly AT can really clog up a midi pipe if you are only ignoring the generated data at the receiving end.

And some synths, as best I recall, wanted to do stuff with the data if received. In summary, it was better to turn off Poly AT globally for piano, organ, just about all meat and potatoes patches except novelty rare patches. After awhile it is too much trouble to visit the global page to toggle the Poly AT on for any patch at all. Especially on live gigs.

Nowadays with typical users midi controller and softsynths, it would be similar. Either leave the PolyAT always on and filter it out at each destination in the sequencer, or at the global or patch level in each synth. Or selectively turn it on/off from a button on the controller, as needed.

If leaving either poly AT or channel pressure always turned on, as you say it needs to be set rather insensitive unless a person has extraordinarily even touch. Unless one wants constant accidental random notes getting modulation on piano or organ or keybass playing.

Once set insensitive enough that it is unlikely to trigger accidentally, the weak fingers might not be strong enough to trigger the Poly AT with the same control range as the strong fingers. On my old KX-88, the channel pressure works pretty good, and the channel pressure is unlikely to accidentally trigger even playing dang aggressive, because you REALLY have to lean on the keyboard to trigger the channel pressure.

But if I'm playing a mono lead line and want to add channel pressure to a note under the pinky finger, I turn my hand sideways and use the whole hand to help the pinky finger. The pinky isn't strong enough to push that hard all by its lonesome. Well, the little finger can do it all by itself, but it hurts.

In cases wanting to add expression to a whole chord with channel pressure, it doesn't matter if the pinky isn't strong enough to get the job done without pulling a muscle-- The thumb and middle finger are in there helping mash the keys down hard enough to trigger the aftertouch.

But in the same scenario with poly AT, the weak finger notes will just tend to get little or no modulation, and the strong finger notes might get too much modulation.

Increasing the after touch sensitivity would minimize that problem, but for playing piano or keybass or whatever, the after touch would either need to be turned off, or alternately be good enough to develop an inhumanly even fingering technique, to avoid accidental modulation.

[aciddose]

a lot of modern synth don't even have sophisiticated keyscaling options
like those available on 80s FM synths. you're tweaking something and
you get to the bit where you want a bit of...aaaw, none on here...

You know? You have touched on a subject matter which is very important to me. In fact I was planning to start a petition for all synth plugin devs to start looking at key-scaling, to highlight how the traditional implementation is absolutely archaic and insufficient, and how to move forward. Your mention of the Yamaha FM synths is a great example, since these synths had good key-scaling features.

This is OT, but I feel like starting a separate thread to highlight this issue.

You should, as I'm interested in what you're talking about and it definitely is not intuitive in the least.

From my point of view what you are asking is to turn a key-tracking knob into a set of five or more controls. I'm not sure how you can justify this massive increase in complexity with the rare use of such features.

You should be able to accomplish the same with any synthesizer with a powerful matrix and "modulation functions". The modulation functions work like a matrix, but each has multiple inputs (usually two or more). For example using a function like pow() it is possible to translate key-tracking in pitch (exponential) back to linear by applying a root. Various scaling functions (a combination of limiting, abs and scaling) can then apply whatever complicated transform you desire.

If you look at the issue this way what remains is really a trade-off between in-your-face complexity in a grid-of-knobs sort of way versus hidden complexity in a more generic matrix format.

Ideally you would be gifted with a synthesizer also implementing "meta controls" which are able to act as inputs to many functions and matrix rows at once. This would allow you to set up a system with any configuration of knobs you like, exactly like the dx7 if desired.

In addition to that, "modulation presets" as "building blocks" would enable you to design and implement such a mod routing once, then reuse it by dropping it into place.

What you're asking for is, as far as I can see, a massive increase in the internal complexity of the synthesizer.

[pdxindy]

I'm unsure if the classical keyboard really is the best place for Polyphonic Aftertouch anyway. When I tested the Qunexus, I was amazed by how sensitive it is. You can touch it very softly and already get it to recognize the touch and then have a nice range of pressure for modulation. But it's not really a "classical" keybed.
Since the classical keyboard needs to be hit rather hard already for velocity, somehow I think it's adverse by nature to finely modulated touch action after. Or the pressure sensor would have to be off for the initial press and only become active with a slight ramp afterwards?
Would be actually interesting to come up with some new concepts here.

But basically: the piano-style keys were originally created for more or less on/off kind of action with only velocity as variable parameter. For fine modulations something less rigid may be better anyway.

Interesting times

Tom

Hey Tom

The AT on both my Z1 and Analog Keys works fines... and I have good control over velocity as well... There is no reason a standard keyboard cannot have PolyAT

As for new concepts, I wonder why poly AT on a QueNexus is so cheap but so expensive on a regular keyboard? I really enjoy playing my Push and also my regular keyboards. They have different strengths. The AT works better on the keyboards than the Push.

[pdxindy]

a lot of modern synth don't even have sophisiticated keyscaling options
like those available on 80s FM synths. you're tweaking something and
you get to the bit where you want a bit of...aaaw, none on here...

You know? You have touched on a subject matter which is very important to me. In fact I was planning to start a petition for all synth plugin devs to start looking at key-scaling, to highlight how the traditional implementation is absolutely archaic and insufficient, and how to move forward. Your mention of the Yamaha FM synths is a great example, since these synths had good key-scaling features.

This is OT, but I feel like starting a separate thread to highlight this issue.

You should, as I'm interested in what you're talking about and it definitely is not intuitive in the least.

From my point of view what you are asking is to turn a key-tracking knob into a set of five or more controls. I'm not sure how you can justify this massive increase in complexity with the rare use of such features.

You should be able to accomplish the same with any synthesizer with a powerful matrix and "modulation functions". The modulation functions work like a matrix, but each has multiple inputs (usually two or more). For example using a function like pow() it is possible to translate key-tracking in pitch (exponential) back to linear by applying a root. Various scaling functions (a combination of limiting, abs and scaling) can then apply whatever complicated transform you desire.

If you look at the issue this way what remains is really a trade-off between in-your-face complexity in a grid-of-knobs sort of way versus hidden complexity in a more generic matrix format.

Ideally you would be gifted with a synthesizer also implementing "meta controls" which are able to act as inputs to many functions and matrix rows at once. This would allow you to set up a system with any configuration of knobs you like, exactly like the dx7 if desired.

In addition to that, "modulation presets" as "building blocks" would enable you to design and implement such a mod routing once, then reuse it by dropping it into place.

What you're asking for is, as far as I can see, a massive increase in the internal complexity of the synthesizer.

All you need is the regular keytracking control... and then in a separate tabbed area, a graph to draw the curve you want with the default at linear... some synths already allow this...

[EvilDragon]

a lot of modern synth don't even have sophisiticated keyscaling options
like those available on 80s FM synths. you're tweaking something and
you get to the bit where you want a bit of...aaaw, none on here...

You know? You have touched on a subject matter which is very important to me. In fact I was planning to start a petition for all synth plugin devs to start looking at key-scaling, to highlight how the traditional implementation is absolutely archaic and insufficient, and how to move forward. Your mention of the Yamaha FM synths is a great example, since these synths had good key-scaling features.

This is OT, but I feel like starting a separate thread to highlight this issue.

I would say that keyscaling is quite advanced on Kronos... And pretty much akin DX7 keyscaling (since it can import all DX7 patches, too).

[EvilDragon]

From my point of view what you are asking is to turn a key-tracking knob into a set of five or more controls. I'm not sure how you can justify this massive increase in complexity with the rare use of such features.

I don't think it needs to use knobs at all. For example, see Zebra 2 and its keyscaling graphs. Much easier to tweak than having a bunch of knobs. That's how keytracking is more easily done. And yes, it is definitely beneficial to have it that way. Just a matrix and mod functions wouldn't quite do it in the same way (not if you want one knee of the multi-line keytrack curve to have a different slope than the other, for example), that would in fact be overcomplicating it.

[aciddose]

All you need is the regular keytracking control... and then in a separate tabbed area, a graph to draw the curve you want with the default at linear... some synths already allow this...

Okay, so now you want a vector envelope and graph on the GUI to take up a large amount of space. Just as much space most likely as the bunch of knobs would have required. You jump to the assumption that tabs/pages would make sense to hide the extra graph but you don't factor in whether the theme of the GUI may be "hardware" or similar. Pages just don't make much sense in that case unless the existing envelope controls are turned into a "digital display" along with the page tabs/switch/whatever.

While I don't have a problem with that sort of thing, you need to realize it is not going to make sense to apply to every synthesizer out there.

Not only that, but having written a vector envelope + graph a couple years ago and used it in a recent product myself, you may have no clue exactly how "simple" this sort of thing is.

This is exactly on-topic in fact as it is a great demonstration of exactly the same thinking as "standardize aftertouch". It takes a lot of effort to do this! That is the reason most synthesizers don't have it, much like most keyboards do not have aftertouch. It increases the expense of the product and provides little in the way of benefit. Only in situations where the benefit outweighs the cost would it be implemented.

[aciddose]

From my point of view what you are asking is to turn a key-tracking knob into a set of five or more controls. I'm not sure how you can justify this massive increase in complexity with the rare use of such features.

I don't think it needs to use knobs at all. For example, see Zebra 2 and its keyscaling graphs. Much easier to tweak than having a bunch of knobs. That's how keytracking is more easily done. And yes, it is definitely beneficial to have it that way. Just a matrix and mod functions wouldn't quite do it in the same way (not if you want one knee of the multi-line keytrack curve to have a different slope than the other, for example), that would in fact be overcomplicating it.

It depends how you look at it. It would be slightly more complex for you to input the configuration you want, but only because you are not a programmer.

This could in fact be done with an interpretive language rather than graphical elements, for example:

Code: Select all

filter.cutoff += piecewise2(60, semitone, 1/2, 3/4) * filter.keytrack

This is far more easy for me to specify exactly what I want than fiddling around with a mouse and trying to see a tiny graphical display.

Piecewise2 would be a piecewise split function. You'd specify the split point (semitone 60 = midi middle c), the input value and the below/above fractions.

The cool thing about this (which I already have working btw, it isn't impossible, just not really suitable for a mainstream product) is you could do really anything. There are no limitations as if you don't have a function you need you can write it yourself! Even supports recursive functions, loops and all that lovely / horrible stuff.

It's a question of real ultimate power vs. some very limited interface. The limited interface seems more simple and thus easier to use to you, but if you were the sort of person who thought outside the box in these cases you'd realize this sort of implementation is in fact weighing everything down rather than making it any better.

Just remember when you are looking at my script language for specifying a mod matrix with your own preference for a typical spread-sheet style input... There is someone looking at your keytracking graph with a preference for a single knob.

We don't know everyone's preferences, but we can clearly see the single knob solution is far more commonly implemented. The reasons for this are not really so simple. It is not always simply a matter of "didn't they think of X?".

http://en.wikipedia.org/wiki/Curve_fitting

Also possible to specify a function in the interpretive language by "drawing" what you want and having it generate the coefficients and function automatically for you.

I just find the idea of curve-fitting to something scrawled by a wobbly ape limb a little bit gross.