This past Monday ( (Sept 21, 2009)) there was a landmark concert event in Manhattan featuring a stunning piano, retrofitted with a reproducing system created by a unique and exciting musical company.

The Piano: Steinway D #133291 (circa 1909) in American Walnut (restored by Faust Harrison Pianos)
The Company: Zenph Studios
The Place: Steinway Hall
The CD: Rachmaninoff plays Rachmaninoff. Sony Masterworks. Released Sept 22, 2009. Reviews etc. are on the Zenph website.


I think most here are pretty well versed in player piano technology, and have experienced at least one of the major systems out there; be it PianoDisc, Disklavier, Stahnke LX, or Pianomation. Perhaps, you've been privileged enough to experience the original Stahnke retrofit , or the Ceus system.

This was different.

For those of you who are not aware, the Zenph approach is to take previously recorded masterworks by the world's most renown artists and bring them back to life with modern re-performance technology. Instead of re-mastering an original recording, they recreate the performance: on a superior instrument, with superior player technology.

Without getting too technical, Zenph figured out how to extract the original live performance from recordings (no matter how old or noisy or distorted those recordings might be), and install it to a high resolution MIDI piano playback file. It's a complicated process that takes an incredible amount of time, but the ultimate result is a "live" performance that is played exactly as if the artist was playing right there at the piano. Every phrase, every nuance, every minute detail is just as if you were there with the great pianist himself (or herself).


Monday's concert featured Rachmaninoff. There we stood/sat listening to and watching one of the greatest composers of all time playing his own works. We experienced live and first-hand his agility, musicality and technical prowess. The only thing missing was the flesh and bones (or at least a hologram) of the man himself. And if this wasn't enough, the amazing violin virtuoso Joshua Bell stopped by to jam with Sergei as we all watched in total awe.


In a word: Incredible


Sara Faust had this to say:



For those of you who might be interested in experiencing some of this wonder yourself, we've arranged for the piano to be displayed at Faust Harrison Pianos throughout the month of October. Perfomances at Faust Harrison are being scheduled as I write this. If there's enough interest, I might be able to arrange a performance for the PW community - let me know. If you stop by (any time), you also might be able to make your own recording and then immediately hear yourself re-played live by the world's greatest reproducer on the one of the world's greatest pianos. On November 6th, the piano will be moved one block south so that Rachmaninoff can play the piano at Carnegie Hall.

There are recordings available on CD of Rachmaninoff playing various pieces (including some of his own) on a new Bosendorfer. It has been a while since I read about this, and I may be muddling some of it, but to the best of my recollection the pieces had been recorded by Rachmaninoff on a reproducing Aeolian piano, and some brilliant computer types figured out how to take the wax recordings and reproduce them on a computerized Bosendorfer.

Anyway, the CDs are fabulous.

This is great news! I have already ordered both CD's. Thanks for posting this Jeff! GP

I had, at one time, in my posession a floppy disk called "Rachmaninoff plays Rachmaninoff" or something titled to that effect. I used to use Flight of the Bumble Bee to illustrate the difference between the regular Disklavier and the Disklavier Pro. I recall it being distributed under the company name "Live Performance".

I also remember being told that the MIDI information was interpreted from a piano roll recording of Sergei's actual playing. I was also told that the piano roll technology at the time could capture basic dynamics and timing. Something having to do with limiting the amount of air that passed through the hole punchings (I'm no expert in analogue player piano technology).

I have only heard those files played on a Yamaha Disklavier though, never tried them on a PianoDisc or alternate system.

It should be mentioned that the player system in this restored Steinway D is a genuine SE system (designed by Wayne Stahnke), capable of full high-resolution recording and playback to the standards of the original SE systems factory-installed in Boesendorfer pianos during the 1980s. Through a gentleman in the UK by the name of Richard Shepherd, a modernized version of the SE system is available for commissioned installations into certain models of pianos, as was done for this Steinway D. This would be prohibitively expensive for most of us, but this level of performance is essential for high-end applications like Zenph's restoration projects.

Thanks for the clarification Mark!

Will any new recordings be made with this?

I believe 2 recordings have been made! www.zenph.com

The direct link is http://zenph.com/store/index.html

Another interpretation of Toman's question is "Will Zenph be using the recording feature of this SE?" (that is, recording the data of live pianists' performances) I sure hope so! Zenph has already made their Glenn Gould and Art Tatum re-performances available for the Live Performance LX system (details at live-performance.com). Any new high-resolution recordings made on this Steinway SE would be easily adaptable to the LX and could be converted for use on other player systems as well (Disklavier, PianoDisc, Pianomation). The potential is there for some GREAT new material!

Thanks for sharing, Jeff! Very cool indeed. I had heard about Zenph's work with Glenn Gould's Goldberg Variations, and had heard about their Rachmaninoff. I'm definitely going to get them sometime...

As an aside, what a lovely Steinway D! Don't believe I've ever seen one in mahogany before.

@ Toman - I am not affiliated with, nor qualified to be, a Zenph spokesperson. That being said, I think the purpose of the Zenph technology was to restore and revitalize famous recordings of the past.

The piano records, and records amazingly well. They could certainly make modern recordings and use the technology to their advantage in terms of multiple takes with various mic placements, or fixing mistakes or reinterpreting a performance.

So it makes sense to me that Zenph will probably use it a few ways.

Well, you are more than welcome to stop by the showroom and hear the Goldberg Variations first hand, as played by Gould. It certainly will beat any audio system I can think of .

I grab on the above quote by Jeff, but not actually talking to him (exclusively)!

Gould has been quoted on mentioning that once one has made the 'perfect' recording, there's no reason to perform the piece again. Recordings were the death of live performance. (Books are in my studio, so don't have access right now for bibliography or for the exact quote).

Of course he was wrong since we still get live performances!

Or was he?

Now we don't need live performances! We're getting dead ones as well!

....

I have a huge grudge with recordings themselves (and I've mentioned it a few times in this forum already). There is a failure to realise that a recording is actually a different medium than a live performance, and that it could actually even be perceived as a different art altogether!

You get all those pianists trying to sound like the CDs! Yet they shouldn't!

Live performance is about interactivity between the composer and the performer, the performer and the audience and the audience themselves! And if you take one of the three out you get neil! The composer is always there in classical music, whether dead or live, since (s)he left their instructions to the score. The audience is there in a live performance (otherwise we are not talking about a live performance, but a studio one, or a recording), and the performer... Well we can be missing that now!



That said and to be clear I think that it's completely fascinating to be able to do that and I think that for research purposes as well as for furthering the idea of music onto new idioms this is simply STUNNING! (So I'm not only nagging).



The idea of fixing mistakes, etc, is already there for quite some time, but this is the first year (2009) that one can actually take out notes, or repair notes, or clear blue notes, etc in a recording. The one way is being described here. The other way is Melodynnes DNA technology (Celemony). You take a recording, it analyses the notes and you move them around! Want the Pathetique by Kempff in a C maj instead of min? No prob!

Scary things! Scary...



Finally. I'm just trying to wonder if 128 steps of midi technology are enough to capture the full spirit of Rachmaninoff (or anyone at that). If so we are looking at a whole new world!



PS. Sorry for the philosophical tendencies of this post!

I'd love to...but I'm on the other side of the country.

Jeff,

Thank you for posting this. It is newsworthy, no doubt.

I'm sorry you could not come down to Philly for the CEUS live distance performance that we had last Fall, but I may find an excuse to get to Manhattan this month.

Cheers,

Interesting to see and read all of this.
John Q. Walker first appeared at Yamaha Artist Services sometime around 2001. He even purchased a DCFIIISA Pro (Disklavier), right after it was used at the first-ever Piano-e-Competition in 2002.

Yamaha was quite closely involved with his work with Disklavier and high-definition midi (which the Pro utilizes for its recording and playback). The Disklavier, then, was recorded for the first two SACD SONY CD projects - Gould and Tatum. I remember the first time I actually heard some of the Gould and Tatum played back on one of our instruments. It seemed to fulfill a lifetime wish of being in the room when these magna artists played.

It's interesting, then, to see a different approach, using an LX playback system and an antique piano. I hope this works out well for Zenph, as it continues to make the way for reproducing pianos as serious instruments in the arts.

James

Hey man, is that the same dude I saw playing in the subway one day? He got a few quarters and he seemed pretty good. Wish he had played Turkey in the Straw though instead of that elevator music.

Hi James,

Thanks for the clarification regarding Yamaha's involvement. I recall, seeing this debuted at NAMM some years ago hearing about Yamaha's involvement and wondering if any new software for the Disklavier library was going to come out of it.

When I learned that Zenph's primary focus was to be making audio recordings, instead of hi rez MIDI files for the Disklavier, I have to admit I was a bit disappointed.

Be that as it may, I think they have a far better chance at success producing these recordings for the most convenient medium for the highest possible number of customers, which at the current time seems to be Audio recordings. In regards to his choice of pianos; I know that John Walker reads these forums, but am uncertain if he posts. I don't want to speak in place of him, as I am not qualified nor authorized to do so.


Regarding the FHP Rebuilt Steinway D; I don't know if the word you used earlier, "Antique", adequately describes this instrument. Is it still considered an antique if it has been fully restored to new condition?

Yeah Rich, I know I've been a bad neighbor. Leaving Manhattan feels a little like escaping Alcatraz at times. I would promise to visit you soon, but I think we both know that it won't mean a thing until I actually arrive, lol.

Keep the Facebook updates coming and I'm sure I will make time in my schedule to get down there as see you.

Call me picky or whatever you want but the re-recording in this sample doesn't sound to me like it has adequate fidelity to the original:

http://zenph.com/listen.html

I listened to it several times and the delicate energy and lightness dynamics of the 'hop' feeling of the 1-2 rhythm is superior on the original. In the original, the notes that come between the 2 beat and the next downbeat trail off to almost nothing and leave you with a gap waiting for that next beat. The dynamics are flatter on the re-recording.

I also think the new piano is voiced too softly and there's too much reverb. This leaves the definition of Cortot's quick and beautiful left hand work lacking.

Also the scratchy noises in the original are deceptive and gratuitous. It would have been more honest to at least touch up the scratches as would be done in any recording that one could actually buy.

What Zenph has been able to do is nothing short of amazing. To be able to re-create a perfect duplication of a artist's playing from a "sound" recording is simply incredible.

When comparing that listening test Charles, you must take into consideration several factors, one being the acoustic environment of the original recording, that cannot be duplicated by Zenph. Also, the piano/s that these artists used cannot be duplicated exactly.

What can, and is duplicated with extreme precision by Zenph, are the exact notes, rhythms and expression of the original performance. I once heard a demo disc produced by Zenph that recorded this original Cortort piece in question...his original performance was on the left track, and Zenph's re-performance was on the right track. Both played simultaneously, and it was exactly the same on both the left and right tracks, speed, timing, rhythms, expression pedaling etc. You could not tell the difference between the two. This was done as a demonstration by Zenph, that what they are doing is a real re-performance of the original, keeping in mind that you cannot reproduce the original acoustic environment, the effects sound-wise in that acoustic environment of the original piano that was used, nor can you reproduce the exact timbre etc of the piano that was used.

I have no doubt that the rhythm is precisely the same. Presumably this would be the easiest thing to copy from an audio recording and I'm sure it can be done with great precision.

The dynamics are harder, not least because a performer adjusts his or her performance according to the piano and to a lesser degree the acoustical conditions of the venue. I'm sure that to a degree they have done this with precision as well, but it's much more subtle than rhythm and there are many factors involved.

After our recent hammer discussions I have to wonder what kind of hammers are on that piano. On Cortot's original, the trailing pp notes have a different tone than the rest of the notes. Surely this has to do not only with velocity of the hammers but voicing and composition of the hammers (and everything else, but to a lesser degree). On the re-recording, the tone sounds largely the same across a certain range of dynamics.

Furthermore it seems dubious to me that one would want to do the re-recording in a venue that is obviously different from the original. Since performers do adjust their performances to their venues, presumably the most historically accurate location wouldn't have all that reverberation. Maybe there was reverberation in the original location but the recording equipment at that time couldn't pick it up, but again, listen to just the last few bars of both versions. To me it sounds like there is enough fidelity in the original that you can tell that there is less reverb.

At the risk of sounding like an acortot-style curmudgeon, the re-recording doesn't really impress me. It is a Cortot-style recording perhaps suited to some segment of present-day consumers and it's fine for that. (I don't work in the industry so I don't have to have *too* much respect for the market, but I think it's reasonable to have some).

It seems more about making a statement about technology and yet another effort to breath life into a genre (and an instrument) that is past its heyday. Kind of like Lang Lang's histrionics. I'd rather they step back and just respect the work -- by which I don't mean they shouldn't do the re-recordings at all, but rather that they should do them more faithfully.

Come on who are we kidding. I just listened to it again and the voicing is completely different on those pianos. You can hardly hear the attack on the re-recording.

Actually, I think Zenph is respecting their work, by trying to re-create what we today would have heard had we been alive at the time they recorded. What I wouldn't give to have been able to hear Caruso live! It's the same for these great pianists, but at least we have the ability now to hear them live...and in some cases, in our own living rooms...if you have an LX that is.

The Cortort example I believe, was Zenph's first example of their technology. They have not yet released a full CD of Cortor's playing. The subsequent releases, the Gould and Tatum recordings are fantastic. I have both of them for my LX system as well as the 2 Sony CD's.

I don't think it's possible to re-create all the exact parameters of the original recording venue, the instrument etc of those original recordings.

It is possible however, to re-create the exact performance, as Zenph has done so beautifully.

One can argue that one or more of the original recording parameters were not exactly the same etc etc, and that they were not faithful to the original, but in the end it really does not matter. What matters is that we can hear what it would have been like to have Rachmaninoff play live, in this case, on a beautifully restored high-end piano. This to me, is what it's all about.

These strike me as incompatible statements. Not only is it not a re-creationg of the exact performance, it's not what it would be like to hear Rachmaninoff play on the restored piano.

An exact re-creation would involve a piano that sounds like the original; on the other hand hearing what Rachmaninoff would have played on the restored piano is impossible because every pianist adjusts their playing to the instrument they're playing on, and it's impossible to know how Rachmaninoff would have played on the piano we have today.

Since the former is at least within grasp if we voice the piano like the one in the recording, I'm just saying we should try that instead of changing things to show off technology or some technician's even round voicing of choice.

No matter what we should keep things in perspective. However well this is done, it's a *copy* just as a copy of the Mona Lisa is still just a copy. I think the word 'exact' is too strong.

Again I'm not saying it's not worth doing; I'm just saying how I view the project.

I understand your point, and as you say, it's best to keep things in perspective.

Perhaps someone from Zenph will chime in hear to give us their perspective.

I applaud Zenph for their efforts, and I hope they will continue with their work.

Jeff, thanks for your perspective on this.

I only know what we did with Zenph. I for one was very proud to be working on such vaunted restorations, through our own instruments and John Q's various shades of genius.

I was of the impression the 1909 concert grand was picked because of its historical nature and sound. How much was brought up to the present day, I do not know. But 1909 is a hundred years ago. Definitely antique. It's even pre-1917 - the year the D became the design we know. Am looking forward to hearing it - much like the even-older Bechstein which I played next door. James

Just to be clear, the system in this Steinway is not an LX but rather an SE. The LX costs around $8K installed and is a consumer-oriented, high-resolution playback-only system. In contrast, the SE is intended for professional applications, includes a full recording subsystem, is based on completely different electronics, can play with a slightly wider dynamic range than the LX, and costs well over $100K.

Both systems share the same specifications for music data though, including temporal accuracy to 1/800th of a second, 1020 hammer velocities and 256 pedal positions (surpassing the capabilities of standard MIDI). Recordings originating on the SE and Disklavier Pro can easily be converted to play on the LX with very impressive results.

Regarding correcting notes, etc, one of the most fascinating parts of the Zenph presentation was a performance by Rachmaninoff when he was "dog tired" with fluffed notes and awkward phrasing. Zenph is really dedicated to recreating a performance, not necessarily correcting it.

Also, the system used was not LX, but the custom SE system referred to earlier.

As a non-expert but just an interested person in technology, the Zenph process presents some very interesting questions. I happen to have purchased the Glenn Gould re-performance and enjoy it (much more than previous releases). I don't believe the question is whether the re-performance is a quintessential representation of the artist's work. IMHO, I think the question concerns a comparison of the original recording (very flawed technology) compared to the pristine re-performance and which version is closer to the actual intent of the artist.

Any recording is a "recreation" of a performance. I think we all can agree on that. The original recording technology was very limited and I believe we all can agree that the limitations of the recording compromises what the artist was hearing in producing his work. I read (although I can't find the original review) that one person originally felt that the re-performance was so very different in terms of dynamics that despite the precision in rhythm, sustain, etc. But he then did an interesting thing. He compressed the Zenph version and then converted to the file to mono to be comparable in format to the original mono recordings. Suddenly, in headphone comparisons (left vs right), the dynamics sounded equivalent. That reviewer then concluded that the Zenph re-creation was indeed closer to what the artist must have actually produced compared to the inferior original recording.

Perhaps, some of these questions would be reduced if Zenph were to do the following:

1. Take a modern pianist to play on a modern piano (outfitted with the player components) and record a high fidelity stereo recording.

2. Take the recording and compress it down, turn it into mono, add some noise and clicks.

3. Take the inferior, compressed, mono, noisy recorded version and do the Zenph "magic." Oh, and the people who do the "magic" must not have access to the unprocessed original recording. They must only have the doctored up, inferior version to work with.

4. Now play the Zenph re-creation using the SAME piano that was originally played with the same microphones in the same locations.

5. Now you have the original recording and a re-performance to directly compare on the same piano.

It would be an interesting comparison.

Regardless, I like that Zenph makes a strong effort not to "correct" flubs. They do seem to be committed to provide an accurate representation of the artist's playing (just on modern instruments with superior recording technology).

Just wondering.

Mr. Peabody

I just learned from John Walker, and would like to make this correction, that the system Zenph has created is not recording MIDI data . And making a further correction on James' post above, the system is not a Stahnke LX. It's an SE system.

I posted this on bealf of John just to ward off any technical discussions that arise regarding MIDI, since this system does not use that interface.

However, any technical detail beyond what I just mentioed will have to be addressed by the Zenph team, as I am not knowledgeable enough to discuss it.