I am working on a project which involves using a Disklavier, unfortunately at the moment I have no possibilities to play around with one myself and to try things out on a Disklavier. I was wondering if someone with already experience with working with Disklaviers could answer me the question of "how fast can a single key/tone be repeated on a Disklavier"? I know it might depend on models, I somehow however need some reference values to work with. Also are there any limits in terms of how long you can let such a machine to play very fast dynamic pieces? For instance is it possible to let a Disklavier play a 4 to 5 minutes of a Black MIDI music, without breaking or damaging something? If you know of some links where I could gather some more information I would also very much appreciate that too.
Why would that be desireable? As a rule of thumb pianos do not sound particularly good when you play them inhumanly fast because the sound needs time to build up. The big advantage you get from a mechanical piano is not being tied to the limitations of two hands and ten fingers. A mechanical piano can of course play passages fast no matter how complex they are, which is also a boon.– LazyMar 20 at 17:39
(By the way, if you're not familiar with Colin Nancarrow's works for player piano, these are part of the same conversation. Using technological means to do what isn't humanly possible. I bet he ran up against some mechanical limitations too...– Andy BonnerMar 20 at 18:06
It can only respond as fast as a piano can. It's electro-mechanically activated, but it's still 'just a piano'.
Sure, it can play faster than almost any pianist, in theory, but you have to wait for the hammers to fall back each time.
I only got about halfway through that torturous annihilation of Greig before I couldn't bear any more, but no, it could not do that.
Also, Midi can't carry data at that rate, so playback would be massively out of time.
(Whatever VST was used to do that piece can't really do it either. You can hear all kinds of note stealing. It's a hideous mess… but that's just my humble opinion ;)
Right ... In the beginning it just misses out the intricated orchestration and mood of Grieg, but later on it just sounds really bad. Probably took lots of effort to make, but from a musical point its worthless.– LazyMar 20 at 17:44
Looks like it was made by a gamer…never mind what it sounds like, look at the pretty patterns.– TetsujinMar 20 at 18:00
That rather looks like controversial substances to me!– LazyMar 20 at 19:07
I definitely find that video hideous too, it was just meant to depict what I was meaning. @Tetsujin Also thanks for pointing out to MIDI. Is MIDI the used format for the Disklavier as input, as I understand, right?– StudentMar 20 at 19:33
Yes, it takes just regular midi files. [I can't remember if it will take type 1 or just type 0, but that's probably easy to google.It's been a long time since I worked with them]. As far as I'm aware , the output engine is the equivalent to a standard 5-pin midi protocol, so subject to its data rate [unlike a VST, which can operate at however fast the computer can pipe data - ie thousands of times the speed.]– TetsujinMar 21 at 7:52
I haven't worked with one and don't have exact data. I have a couple of references:
Impossible piano music on a Disklavier: A Youtuber attempts black MIDI on a Disklavier. Has examples of what's probably the fastest single repeated note. This is a grand piano, an upright can't repeat as fast as a grand. (According to Wikipedia, Yahama currently has 14 Disklavier models from upright to 9-foot concert grand, so check the specifics of yours.)
Towards the end he tries the fastest repeat when playing all the notes at once - that's glacial, about half a second to trigger all the notes and a second between repeats. The guy in the video blames it on MIDI transmission rates, but according to wikipedia and some quick napkin math, MIDI should be able to transmit 88 note on/off pairs in 0.17 seconds. I'll venture a guess that the additional delay stems from some form of current/power limit in the piano hardware (and/or processing limits). He also says that the Disklavier started overheating, and he found it best to quit before he broke something.
Another electro-mechanical piano (not a Disklavier) attempting Rush E, I think it's safe to say that it doesn't do so well in the busiest sections. (Compare to original.)
You may also find these articles interesting (part 1, part 2): A project involving 16 synchronized Disklaviers, and associated issues and workarounds. This was in 2000, so presumably an earlier generation Disklavier, but it illustrates the kind of challenges one may encounter when working with hardware.
The fastest playing is probably by avoiding repeats, play another note or two before you return to the first one. You can no doubt make something interesting, but the exact limits depend on a lot of factors including the specifics of the model you are using. And it won't compete with computer-generated audio for black MIDI.