Way to control tone color of virtual piano. (Page 1 of 2)

First of all,
Thank you k c Paul Li, for your interesting comments. I'm glad we have you in our forum. I’m learning new things all the time, thank you.

What is the tone color of the piano? Was wondering….
I think color, tone, tone quality, sound quality, sonority, and timbre are all the same thing, not much difference between them.

I am not a pianist, and do not understand much about technology. I can only tell things from my experiences, but there is of course more to piano sound and color than just loudness of the tone. (Describing sound is a bit like describing wine….. 

In my opinion there is actually one way more to control the tone color, when playing Pianoteq, and that is using the pedal.
As I see it, feel it, the primary use of the pedals is to enhance the sound/tone color of the piano. And to my ears it does.
Well, that’s what I’m thinking about it.
Best wishes,

Stig

Pianoteqenthusiast, Organteqenthusiast, Harpteqenthusiast, Harpsichordteqenthusiast

Good point, Stig, and I can add one more way to change the tone - very similar to your pedal suggestion, but it's called 'finger pedaling'.  Especially when playing several notes in a phrase, how long your finger lingers on one note, and how quickly it releases will change the coloration by either a soft gradual damper return, or a rapid staccato damper return.  Additionally, which notes you 'keep open' with your fingers change the sympathetic resonances of notes around them that you are playing.

In real time performance you can also use split MIDI keyboards and/or multiple keyboards with different pianos loaded for different sounds. If you really want different textures from your pianos. Of course all three could be presets from the same piano model if you want. Use 37-61 keys for the extra range will limit the annoying frustrations of running out of keys too early as you will quickly with 25 keys.

I'm sure they'll be a few folks using stacked 88s though.

No, I think k c Paul Li is doing something a little more sophisticated.  If you use either layering or filtering, then the same pitch+velocity will have the same tone colour every time.  But by manually editing the midi file and assigning notes to different tracks, you can have the same pitch at the same velocity in two different musical contexts and give them different tone colours.

Given that Pianoteq produces sounds in realtime and therefore reflects reasonably accurately the subtleties of harmonies from held notes, flutter pedalling etc and given that the influence of touch on an acoustic is zero once the hammer is in free flight, I wonder what creates the belief that an acoustic provides more subtle control of tone? (There is a more cogent argument re touch for sampled pianos than modelled ones however) The DP "action" can certainly make it more difficult to make dims and crescs ultra even, but I guess a good hybrid action should resolve this. (I can't afford a hybrid) The sound is still not as good as an acoustic, still some work to do with instrument resonances, but this is not an issue of touch as such. As others have said, careful, precise, use of the pedal, e.g. a staccato chord followed by a rest given a very brief touch of sustain to avoid a sudden, "dead" end to the sound, overlapping notes, holding repeated notes with the pedal, different weight applied to different parts of a chord, not just for "bringing out" a melody, but for the different resonances this creates and therefore the overall sound, even the use of the una corda in non-ppp passages. (I once witnessed an opening movement of a Haydn sonata played with the una corda held firmly down.) As a sound producing system, Pianoteq is getting close to the sound created by acoustics: the perceived deficiency in digital "actions' certainly hinder the player, but this is purely in the area of touch in terms of degrees in volume, not tonal colour.Tonal colouring is a dark art and once digitally modelled sound is virtually indistinguishable from an acoustic I see no reason why a good pianist, with a good hybrid action, shouldn't produce just as varied tones with either physical method. 

These are of course opinions only and I am genuinely puzzled by the views that only acoustics can produce tonal colourings. It's the sound that is the limiting factor at present and Pianoteq continues to make strides in shrinking the gap.

Pianoteq already handles it FWIW.

From the FAQ:
Pianoteq uses all the 127 velocities that the MIDI specification allows, giving you a maximum playability with a superb dynamic sensitivity. Virtual pianos based on sampling technology are usually restricted to at most 16 samples per note which all need to be processed (cross-fading, filtering, etc). The pianist Hugh Sung explains in video tutorial part 1 his observations on this subject.

Does Pianoteq support high resolution MIDI files?
Yes. Thanks to the physical model that Pianoteq is built on, it has the unique capacity of reproducing all the 16,384 velocities by the Hi-Res CC#88 format, and all the 1,023 velocities by the MIDI XP format.

I will be controversial and say I suspect that 127 velocity layers per key is actually already way more than enough resolution to differentiate all the individual tone graduations already, and therefore this is a red herring for anyone hoping to hear some revelation from even more graduations.
OTOH the mechanism of the key itself will make a significant difference to playability thus tone.

This assumes that not too much resolution has been lost in funky velocity curves that throw away much of the 127 layers.

@dazric

Why do you say only 127  vel levels? I can perhaps get 10-12 levels when I really try… As I see it, isn’t 127 levels more than enough fore realistic playing. Can a professional pianist with a Grand play at 100  200 500 levels, maybe they can? I’m of course an amateur with a bad keyboard so it is not a problem for me, but…

I read that Midi 2 is 64 bit with 65536 possible values….  Do we really need that

@Key Fumbler

Thank you Key Fumbler. I was just going to write the question that  Please can anyone in this forum explain to me the principle Midi piano sensitivity and only 127 velocity values, in a broad context of real instrument.
But now I got some answer from your comment. Thanks.

Best wishes,

Stig

Pianoteqenthusiast, Organteqenthusiast, Harpteqenthusiast, Harpsichordteqenthusiast and experimenter

No, I think you misunderstood what I wanted, perhaps I should have put it more clearly. I'd like to try editing one of my own midi tracks with extended velocity to see if I can hear a difference. Just for curiosity, really.

Sorry, Reaper limited to 127 steps, gotcha!
Perhaps, better ask on their forums, or use  Ask Justin Frankel.

It wouldn't bother me particularly for the reasons described above. I'm still happy to be proven wrong, or fooled for a while ;human hearing system being quite fallible, and certainly open to suggestion.

Yes, all very good and valid points. But bear in mind that, unless you're doing high-level virtuoso stuff, I think most people are unlikely to want to play much more than 115 very often, or much less than 10, so that leaves about 105 velocity steps for general use. Sometimes when I'm editing midi velocities I can't help wondering if some repeated notes would sound more refined if I could build in some values such as '81.25' instead of plain old 80, 83, 81... But maybe it's just me! Will definitely check out Reaper forum as suggested.

Update: I still don't know about Reaper, but it looks as if it might be possible in Studio One. I do have Studio One, but I haven't used it for midi editing because it uses percentages for velocities rather than standard midi values, and that confused me. But it looks as though you can input values such as '46.2%', so it's something I'll have to investigate when I have some time to mess about. I may eventually upload some 'can you hear the difference' recordings!
I just figured out that to convert midi values to percentages 'near enough', multiply by 0.79. Or 0.7874015748031496 if you’re really fussy.

Ah, excellent, thanks Key Fumbler! Just glancing through that thread it looks as if the benefits of hi-res midi may be negligable at best. It's certainly not something I'd want to spend a lot of time on, but I may try it for curiosity all the same. I don't like editing more than I can help anyway. With improv pieces I prefer to keep it raw and will only edit if there's a really bad glitch or to cut out a boring bit. But sometimes with notated pieces I might come across a phrase that makes me think 'ugh, that sounds a bit clunky, I thought I played it better than that'. If I think there's any benefit at all with hi-res midi, I'll save it for those sort of occasions.

Here's another oldie that where one user got a bit heated with me over this:
https://forum.modartt.com/viewtopic.php?id=7673

I tried to play the same note several times in a row as regularly as possible. I rarely get the same velocity twice in a row, it varies by more or less 2 or 3 values, I'm only an amateur but the 127 values are much more precise than the precision of my fingers. I would be curious to know the accuracy of touch of more experienced pianists, do you ?

Firstly want to say 'hats off to k c Paul Li' for showing the technique used!

Indeed, similar can be achieved within Piantoeq by using layers and 'sculpting' what notes/ranges are played by which of the 3 pianos layered.. infinite possibilities either way. Love that people are thinking about these things

About the velocity discussions, here are some thoughts I'm buzzing through after reading this thread, from my perspective..

Pianoteq does practically infinite velocity right now, more viscerally, in real time and without needing extended MIDI to do it.

How's that for a buzzword?.. "Infinite velocity"..

Right clicking many of the main controls in Pianoteq, and you see "humanize".

That allows 'per note' differences to be made "cleaner" therefore with less variation per note - or better..

"wider" - so arguably your piano playing can be more like a real piano, with each note having many and more variations all the time, per note. (not just 'out of tune a bit to give a sense of realism').

I place velocity fuss quite far down the list, at least with Pianoteq.

Variations BTW not just about velocity - things which include esp. new string vibration modelling and so on.

Pianoteq's strings vibrate in a realistic 3D fashion. For each note, or chords, any time your virtual hammer hits any string, this is computed.. the hammer strike will be different for each, depending if it glances off an indirect string vibration, or a direct string vibration upward to the hammer etc.. so when in the middle of playing something, all that - and more things like that in the Pianoteq engine - make far more realistic "in between" or "extra" velocity information. Details important to genuine sonority.

Key Fumbler is IMO on the right path about 127 velocity in standard MIDI mode, not being as limited as people might think (re the way Pianoteq computes notes' real-time tones). - BTW - thanks Key Fumbler - good to see you posting - fairly similar takes on many things it seems.

Future will be extended MIDI - no doubt. Today, most dpainos don't have hardware capability to express things in any more detail.. that will catch up over time.

Don't be fooled though (he said in a topic full of mysteries)..

"Velocity" alone, is just one subset of advantages regarding extended MIDI.. the "benefits" are really not just about extra velocities IMO.

That's said in relation to Pianoteq - I'm sure other VSTis may improve using extended MIDI "velocity range increases" - esp. if just firing 'samples' + 'extra data' .. but what will they do?.. just fire off more samples.. mix up more jumbles of adjacent samples faster.. are sample engineers going to manually 'calibrate' 64000 samples per note?

"Yo Billy Bawb.. boss wants to know when yur gonna be thru making all 64000 middle C samples there, buddy?"

(spits) "Welp.. got a ways tago. Wer at velocity lev'l 1,583.. reck'n a'll be 'nuth'r 120 yrrrrs".

Seriously though - think about "What will you hear different about velocity level 4583, compared to velocity 4584?"

Then think about "Who will decide those exact differences" - or - "What algorithm will describe those differences, while we play?".

Then consider:

It's not like Pianoteq reads in an exact current "65 velocity" key strike and gives you the exact same sound each time (even if absolutely no other notes are playing- there are realistic differences, amounting to 'like further variants to velocity itself'). The differences per note are computed based on the physics of what a real piano does in those ranges.

The 'real world fuzz' is not 'guessed' by Pianoteq, as much as it is calculated (and users can take away or add to that with the 'humanize' quotient mentioned at top of this post).

Everything in Pianoteq is computed in real time - and this is absolutely giving more than even extended MIDI, in range of "outcomes" per note, per velocity measure.

I think the whole thing is fascinating - but I also don't think it's something in too much dispute - just people are not yet really exposed to all the notions around extended MIDI's coming influences.. what Pianoteq does "REALLY".. and people like to predict game changing things.

I'm thinking we have to wait for a looooot of old dpianos to become extinct (patents/pallet loads of cheap keyboards sitting in cargo - probably 5 or maybe even 10 years worth of inventory to shift) before future dpianos begin to really "use it well" (cost of development etc).

There's a load more (techy stuff from the little I know) which is way more interesting than just velocity levels _ I guess I'm repeating some things here though now.

I don't think many have really thought about this without getting caught up in a "It's my first time at NAAM, I'm a musician technologist!" mode (friendly reminder, this is me too).

I know marketing/advertizing suits who go with that wide-eyed and bushy-tailed mindset to big tech/music tech conventions as part of their job or business.. they end up buying into the strangest pitches at the show quite often - they "WANT" to come home with a brilliant "story" to tell the company bosses - - "Within 2 years, we can corner the Lateral Gravity market!".

Maybe there's a tech suit reading, who's thinking something like "With infinite velocity, why, within 2 years we can ... ?". Erm.. no..

Infinite velocity today!

Pianoteq and all its instruments are alive like real recorded specimens, in my use cases.. I often feel, new users miss out on using it well.. often fussing over getting it to sound 'like' something they're accustomed to (or in a badly acoustic bombastic room, tiny speakers, dpiano with non-piano-like actions.. and on).

Putting up with standard MIDI for example in the past decades (with associated primitive hardware to use it - bad keyboards, with un-realistic actions - just 4 velocity levels which, at that very low fence really does matter etc.).. many like me, probably got used to 'using editing tricks' in a DAW, to make "stiff" and boring MIDI performances have more "heart" etc.. but it came down to using studio tricks, more than 'fixing the MIDI'.. and lots of DAWs to this day, allow quantized performances.. god forbid someone plays something with some 'real swing' in it.. it may ruin the production values - where everything needs to be exactly on the 4/4 beat.

When modern pop (etc.) begins breathing more 'human' wrought life (not less) into performances, maybe that could usher in more 'consumer demand' for variability. Currently, more variant velocity is not in hot demand. Listen to most modern music - it is SO incredibly perfected (auto-tune style) in so many ways.. you cannot hear a real player playing anything in much of it.. it's all a perfect "mix".. good on them.. it's an art.. not easy.. but wow, I do imagine good classical pianists hearing nothing but plastic junk. Maybe it takes some thinking 'outside classical' dimensions to understand that putting all those sounds and rhythms together is often as complex as a symphony. I take my hat off to modern performers who get a complex 'hit' together in studio AND manage to perform it 'the same' on stage.. it's as old as the hills "how do we emulate the hit on stage?". It's not easy with the many technical aspects in modern pop.. not that I prefer that to great classical music, or quasi classical music of today. Just a juxtaposition - and use case for extended MIDI, to me, seems more relevant in the quantized world of modern pop (etc.)

But there's the fine argument also that "Well, the musician didn't play those notes - he invented the music and drove the technology toward it's final form. That's just as amazing, as someone learning classical piano and getting to some enjoyable skill level - no lesser than."

In the end, MIDI (standard or otherwise) will make up only a small portion of the entire musical feast. Plenty of new musicians don't think about it.. they 'loop' and swap 'loops' and mix them all together in creative ways, in real time, rather than stooping and zooming in on 'exact velocity' MIDI notes over and over, or those types of things.

In that way (quite well besides Pianoteq), extended MIDI will impact most IMO.. modern creatives.. now, where's my NAAM brochure!

Here's someone who, to me, personifies the type of modern musicians who will just seamlessly begin to utilize extended MIDI as their hardware can do 'things' they like with it. Claudio contends that expression in real time using tech is a joy, important etc. I agree with much she has to say - and quite delight in this performance.

https://youtu.be/q5yxIzs5Wug?t=536

And this by Vangelis in his studio.. demonstrating that, beyond zooming in about fine details of extended MIDI.. we can already impart amazing human emotional impacts - with our own choices in equipment - and experience, coupled with talent may outweigh nuanced issues with non-holistic elements like MIDI velocity alone. 

https://www.youtube.com/watch?v=UOWB7KWS9CA

You provided me an afternoon of thoughtful relaxing - thanks to all in this thread.

Yeah, I'm convinced. Hi-res midi isn't worth the fuss. With Pianoteq it's already possible to get some very finely nuanced effects with the standard midi velocities. The things I do appreciate, though, are note-off velocity and graduated pedalling - they really do make a difference.
This thread has made me rethink my midi editing strategy. It used to bother me if I had several consecutive notes with the same velocity, thinking it might sound too robotic - but now I know that Pianoteq introduces an element of randomness, I'm not going to worry about it!

Yeah, this is one of the technique.
Good points, David.

We don't do such things. We just follow our feelings, our passions.

How about this recording, made by 10 pianos:

https://youtu.be/xvDi-lU5JqI

Me too.

Thank you k c Paul Li, for this video. I have to say I have seen/listened to this long before I had the honor of meeting you here in this forum (I was searching for Beethoven music). Nice to see you and listen again, a bit of Beethoven inside you.

Because, looking at you when playing, the physical movements you do (7,50,  8,45,  at the end, a s o ) you probably translate the music in your thoughts, your mind and emotions. You say you are a concert pianist, yes you are - you know in your head how you want the piano to sound before you touch the keys. You are the instrument and a Grand or Pianoteq are only tools.

Well this is what I think. Be proud of yourself, k c Paul Li, and never stop pushing the best you can be. And thanks for all you have already given us in this forum. I’m impressed.

Best wishes,

Stig

Pianoteqenthusiast, Organteqenthusiast, Harpteqenthusiast, Harpsichordteqenthusiast and experimenter

Thank you, k c Paul Li, for taking the trouble to respond to all of us here! Personally I appreciate what you're trying to do, it's admirable. For my part I'm happy to accept Pianoteq as a simulator, albeit a very good one. But I still want to get the best out of it, and of course I'm delighted when improvements are made with the modelling. By the way, I'm nowhere near being a concert-level pianist - nowhere near! - I'm just a humble music educator doing my best to inspire some joy in music-making.

I would have to listen more carefully on my studio monitors and/or with headphones, but just listening on my laptop I noted some odd changes in stereo image and tone in several places. One of the most obvious/distracting was in the middle of the trill at 4:07-4:09.

If a user is going to blend pianos in this cut and paste fashion to give the impression that there is a single super duper piano then I suggest you need to make sure that the microphone positioning is the same for all the models, and match up the reverb settings.
Broadly all I think what you're going to get at best is a likeable combination where you get the nice bass of one piano, perhaps combined with the the sweetness of a mid-range of another, and the soaring treble of a different model.
This is less likely to be as realistic a model of any real world piano compared to a single accurate model created by Pianoteq, or a tweaked preset thereof - if that is the aim. The interaction of the strings and the body in the physical model will still be unrelated to each other.

Pianoteq allows us to blend pianos. I like that hybrid approach rather than this cut and paste technique. YMMV.

We are all faking it though. So whatever sounds most enjoyable that's the right approach for you! 

Not trying to put you off, hopefully my words aren't unduly negative.
Thanks for sharing. Interesting thread.