I've never heard a system reproduce piano in a totally convincing way, such that it sounds like a real piano, even from the next room.
I have been pursuing this for 45 years now, both in my speaker designs in the 70s and for the last 30 years as a recordist.
When I started recording seriously, in around 1987, I was using near state of the art microphones, cables, microphone preamp (chosen, in fact, for its performance with piano sources) and A/D converters—first the built in converter in the Sony PCM F-1, the the first Apogee on the market, the AD500—a $5000 stereo 16/44 converter.
Hearing playback was always a sobering letdown and got me to thinking about what the issue was; how could it be so different from the real thing?
At the time I was using Tannoy PBM 6.5 monitors—very good for the size, price and time (1968or9). I went from those to B&W 610s, I think, then B&W 805s, but didn't trust the bass rendering, then Harbeth HLP3s with Hsu 10" subs, but could never get them to meld. Finally in 1995 got some 1977 vintage Tannoy Ardens to pair with the subs—now I was getting somewhere. Even though they were about 91 dB/W sensitive they sounded distinctly better when I duplicated my 80 W Sugden amplifiers to vertically bi-amp them. Now the sound I was getting had some resemblance to the real thing, whether it was solo piano or orchestra or 200 voice choir.
This experience showed me that, efficient though the Tannoys were, they could have benefited from much more power than I would have guessed was useful.
Incidentally, from the Tannoys I took a series of downward steps, it seems, in the pursuit of a more refined treble sound (the Tannoys were a little coarse through much of the high frequency realm). I went from SP Technologies Timepieces to Klein + Hummel o300s to what I have nowYamaha NS-1000s, driven by a Yamaha B-2 amplifier, used in the midfield; about 6 feet from my listening chair.
At this point, my thought is that generally what is missing in the various attempts to reproduce piano, apart from the obvious (and common) gross errors in tonality is dynamic capability. The 'forte' part of pianoforte should not be taken lightly, nor should the need for about 20 dB more headroom than many texts suggest you need for realistic playback.
So, my conclusion is (and this is a guess, since I cannot afford the equipment this would require) that what is needed if you are to have a hope of fooling anyone with your piano replay is some variation of these criteria:
1. an amplifier capable of an honest 1000 W into 8Ω
2. a speaker capable of absorbing 1000 W without thermal compression of more than 1 dB in the process as well as having a sensitivity of at least 97 dB/W/M
3. said speaker system having accurate tonality and excellent impulse response through the range of 30 Hz—10 kHz, preferably 20 Hz—20 kHz and, if money were no object, 10 Hz—40 kHz.
The object of all this is a playback system with the capability of hitting 125 dB cleanly on peaks at the same time as having the refinement of, say, a Janszen or Sanders electrostat.
Is this too much to ask?
I intend to find out, but it will have to be by DIY in my retirement years. I am thinking the midrange would have to be covered by a horn, like one of the BD variety:http://www.bd-design.nl/contents/en-us/d83.html
or something like the Geddes designs, possibly, but I think I want the horn to cover more of the lower midrange. Adequate bass potential would be needed; I'm thinking something like an 18 and a 15 per side, perhaps dipole, perhaps infinite baffle (built across corners).
As to recordings representing piano accurately— I believe I already have those, and I recorded them.
One example can be found here: https://www.youtube.com/watch?v=8kLBAWutvdw
This is the theme to the movie "The Diving Bell and the Butterfly", played by Paul Cantelon on a nice Steinway in the Recital Hall at the University of Victoria. I recorded that about 10 years ago
It, like the other 2 CDs I recorded for Paul, was recorded in the purest Blumlein way, with the microphone (a stereo ribbon, the Royer SF-12) in exactly the same position every time—I say 'purest' because it is my feeling that Blumlein is best done with ribbon microphones, not circular diaphragm condensers as is the norm. If you take the trouble to listen to this or his other recordings with the speakers positioned as they should for Blumlein recordings—±45º off your frontal axis, not less, and in a configuration which is capable of yielding a very
tight central mono image (I mean tight like coming from a vertical slit in the centre)—you will be treated to possibly your first ever experience of realistic imaging in piano reproduction.
When I was working in a high end home hi-fi shop 25 years ago and played my first recording of Paul (done the same way) on a meticulously set up pair of Martin Logans with YBA amplification, listening to the master tape raw, with no manipulation, I was thrilled to realize I had accomplished my goal. The piano image was completely detached from the speakers, hovering life-size between and behind. The effect was genuinely like the speakers were not even playing—there was no sense of anything
coming from them.
That experience showed me that, since then, whenever I play that recording and it is not convincing, I don't have to wonder if it might be the recording at fault, I know
it is the playback system.
Some other recordings done the same way:https://www.youtube.com/watch?v=NinR6PfVDHIhttp://royerlabs.com/sf-12-3/