Piano World Home Page Forums Our Most Popular Forums Piano Tuner-Technicians Forum Electronic piano tuning

As an aside, Chris Leslie and I are working on a tuning game. You can enter your sequence, or play someone else's. The game cycles through the checks and you can tweak each note as you go. Then you get a score.

One problem we have is the tone generation. Currently using 8 partials modelling different iH. (B values)

I would eventually like to use the tuning produced by a user, to play some midi files so people can compare the sound of one tuning to another, or ET to an HT, for example.

Does anyone know if that is possible? To use a access a sample within a game and tweak it's tuning? We are using Java.

Hi all, I can vouch for tuning accuracy in Pianoteq. Since it is synthesized mathematically, they can plug in any curve, stretch, octave size, temperament, you name it. Their stock ET that comes pre-loaded is right on the money (using aural tuner's interval checks).

I can't say the same for sampled piano software such as Synthogy's American Concert D and Galaxy's Vintage D, because I do not own them or have access to them.

I'm with Mark on finding sub-standard (sometimes bad) tuning on stock DP's. The best to my knowledge is Yamaha's flagship line. The thirds are pretty darn even (taking into account that they 'stretch samples' over a couple of chromatic neighboring keys on some models). I think somewhere Yamaha said that they hired a piano technician to oversee the sampling of the original Yamaha CFIII, and to be ready to touch it up if things go astray during the pounding of all individual 88 notes, times however many velocity layers (ppp to fff) that they advertise per note. That's a lot of recording!

Coming in behind Yamaha is Roland, Casio, and the rest. Mixed results from those, especially Roland (who supposedly samples different Steinways for each register of your DP!). That's an exercise in futility trying to get that DP in tune with itself!

I would bet that the software sampled pianos that I mentioned above fall into the same category as Roland, Casio, etc., unless they hired a technician to be on site during the sampling like Yamaha did.

The best of both worlds, and possibly the future is hybrid modelling/sampling so they can dial in a note accurately like Pianoteq's synth engine can, AND at the same time have a realistic piano tone when you press a key.

Yeah, unfortunately Pianoteq's reproduction of piano tone is not quite there yet. However, their sounds between the notes (sympathetic resonance, damper resonance, soundboard resonance, etc.) are awesome. This is where sampling falls short, because no one has sampled every possible combination of notes, velocity, and pedal position from 88 possible key presses - which is practically infinite. This must be handled with a mathematical model IMO.

But the reproduction of a single piano note is where sampling shines. This is where the infinite possibilities of steel/copper wire vibrating over a wooden soundboard work against Pianoteq - no one has yet modeled mathematically ALL the possible imperfections and interactions of atoms within an acoustic piano system. It's much easier to playback a recording than to try to build the piano's complex tone with sine waves on a computer.

Hence the need IMO for a true hybrid - Pianoteq's math for resonance, and sample library's recordings for accurate tone reproduction. Who knows when these two will truly meet?

I wonder if it might be possible to sample and in some way "morph" the samples back into tune, probably by post processing the sample. ???
Something that SEEMS sample recorders should be doing is looking at each sample as it is made and calling for a correction right then and there, maybe tweak a unison here or there ?

I am out of my depth here, but I question the idea that the actual movement of the strings and wood have to be modeled.

"Black box" representation, this input causes that(/those) output(s).

Hi prout and RB,
Thanks prout for the reference to that virtual piano modeling paper. I will definitely check that out (I've always been interested in what it would take to completely model the tone of the piano, with all its parts, idiosyncrasies, etc.).

RB, to truly model the complete piano system mathematically, it seems that researchers have to go down to a very fine resolution (possibly molecular), which, as prout pointed out, is not in our current tech's grasp as of 2013-2014. Not to do it in real time anyway.

But computers are getting exponentially more powerful every 2 years so who knows when we will be able to simulate physical systems in real time? There are a lot of other use cases for accurate physical simulation besides piano and sound reproduction of course. Think 'the Matrix'

http://hal.inria.fr/docs/00/78/75/51/PDF/RR-8181.pdf

Well, that is "a" way and probably a very good way if it is (partly) for one's doctorate thesis.
At least one of those folk are on the Pianoteq team.

BTW, it finishes at the soundboard, i.e. the case is not considered (case reson_on_ances, pianos have those, right ?)
IMO this is where electronic and physical instruments REALLY diverge, piano case (or guitar body) vs speaker system.

Horowitz could have make better MUSIC on my keyboard and rack gear than I would ever make on his physical piano.

This is an interesting concept and may be true. However, would Horowitz have been satisfied with not having the response he was expecting?

I made a Tunelab Pro tuning curve for my Yamaha P-60 digital. The curve was absolutely flat. No stretch at all. On a few notes I can hear a small wavering in the pitch, like a slightly out of tune unison. This digital has a very lifeless sound which grows boring very quickly. I don't use it anymore since I bought an acoustic upright.

I don't know exactly how it's done, but I do know that for standard midi instrument sets, the sampled piano is exactly in tune all the way up and down with all the other instruments (e.g. pipe organ, flute, trumpet etc etc) so the manufacturers (or programmers?) have overcome the inharmonicity problem somehow.

I don't have a dedicated high spec. electronic piano so I can't comment on those.