John Walker re-creates great performances

It's a tad selling but still worth while. If this has been posted before, my apologies.

Thanks! I had not seen it. Fascinating!

I found it disturbing. He is lauding the fact that we can digitally remove the human element from music. He claimed his digital piano was like having Glenn Gould in the room. He said it would be possible to hear Horowitz "play" something he never recorded by digitalizing his style.

Performances are not just about the artist producing sound. There is a synergy between the artist, the audience, the instrument, the hall and the times that makes a performance magical or not. If we remove the virtuosity, the human emotions, the risk of a live performance, don't we devalue live music?

I also cannot accept that the sound of a digital instrument can match the sound produced by a resonating box of wood and strings. Every piano is different. Every piano sounds different in different venues and with different artists. Isn't digitalizing a performance similar to the compression that takes place when music is recorded? You lose so much in terms of dynamic range and tones.

No thank you; I'll keep my Steinway and buy tickets to live concerts.

Perhaps this thread should be combined with the Life experiences, personality and emotion in music thread.

gooddog, times are changing ... and they are changing very fast whether anyone likes it or not.


While this isn't directly related it might help put some of this in perspective. There are also many other talks given by Kurzweil and other individuals that address the exponential change in technology.

They aren't using a digital piano. It's basically a very high resolution player piano - still a resonating box of wood and strings, but the individual's performance is recreated on it.

What's interesting about the technology is that it will enable people to fix wrong notes in a single take. A recording could be captured on high resolution MIDI, edited later, then re-recorded with everything fixed. It's an interesting technology, but one that I doubt will be mainstream for quite some time. The hardware is rather expensive!

For more info, here's the guy's company (which just bought TimeWarp Technologies, makers of some software piano teachers will recognize: Home Concert Extreme, Classroom Maestro and Internet MIDI.)

http://www.zenph.com/

If only this particular TED talk was around a couple of years ago... I could've used something like that in my research!:D

I fully agree and I'll only add that a concert/recital is primarily a social event!

Then I'll just report my obsession (apparently) with recorded music, which should NOT be called music (exactly for the above quoted reasons)...

Thanks Kreisler. I already sent gooddog a PM explaining this somewhat.

When I was in Vienna about four years ago I played Bösendorfer's CEUS piano. I also heard (with keys and pedals moving as well) Oscar Peterson on the same piano. He had made a recording for Bösendorfer. It was like sitting next to him and watching him play. I never saw him live but that was pretty close to the same thing. http://www.boesendorfer.com/en/ceus-reproducing-system.html

We do live in interesting times.

Here's a short demo ...

This is awesome! Glenn Gould most certainly would have applauded and embraced this technology.

Yea, but liszt85 would comment that Gould's humming was missing and therefore an integral part of the performance was missing as well.

Yes, but just imagine, you are sitting in your family room and you think to yourself, I'd fancy listening to a bit of the Goldberg's. And out of your personal piano comes the sound of Glenn Gould, with every artistic nuance he laboured so hard to perfect during the recording process.

Furthermore, Gould disliked the fact that he sung with the piano, took pains to make certain it wasn't picked up in the recording, and even said that he would be put off by any artist he was listening to that indulged themselves like that.

What's interesting about the technology is that it will enable people to fix wrong notes in a single take.

I bought the following CD (Vol 1 and 2) many years ago though it's still in a box from our move five years ago.

Amazon - Entremont

The original recording was made in Europe and the information was then sent to New York where the sound guys could take their time miking the piano. Also, any wrong notes could be corrected by the performer or anyone else. All of this was explained in the CD liner notes.

And Dave Horne would argue that once this technology becomes mainstream, he has to only perform once in his entire life. He can then send out a piano (that's recorded his performance) to all the different venues that want him to gig while he watches television and drinks beer (that he seriously needs right now) in his apartment.

Oh and yea.. I'm sure there are robotics groups working on creating robots that can move like humans. So get one of those too and make it look like you. Then have it mimic your playing on the piano while the piano plays what you've recorded. Technology is amazing, isn't it?

Oh and yea.. I'm sure there are robotics groups working on creating robots that can move like humans. So get one of those too and make it look like you. Then have it mimic your playing on the piano while the piano plays what you've recorded. Technology is amazing, isn't it?

Haven't you ever done any modeling work on any aspect of the human brain as part of your research? Where do you think that information will wind up?

That particular TED talk was from 2007 I think, and Zenph has been making waves in this area for several years now. You just gotta visit conferences in the US and Canada more often.

That's exactly what I do. That information will go directly into helping us understand our brain better, not to replace us with robots. That may happen a century from now (nobody knows).. but that certainly is not the intention of the field as of now. Neural network models are used in automatic navigation and handwriting identification, etc. If you want to say that one day you can stay at home while your robot goes to work, then that's going into into how you can misuse technology (there's even a movie about this.. can't recall the name now).

So as an academic working directly on modeling the human brain, I'm aware of the potential misuse of all the information we gain from these methodologies. Have you thought about the adverse effects of technology at any point in time? Scientists have been known to wish they never found out some of the things they did (including the atom bomb). You don't mean to say that the atom bomb was good for us do you? I'm sure Nuclear Physics is important to us.. for energy, etc., but the bomb is an example of a misuse of that technology. You can similarly find parallels in this case too. All you need is some foresight.

The piano was real.

I, for one, would love to have my favorite historic recordings 'restored' with stereo (or more). Imagine the '32 recording of Liszt's Sonata by Horowitz without the hiss and crackle, in full stereo and in full sonic range.

The web site of these technicians/artists/whatever they are includes words to the effect of "contact us and tell us what the really worthy projects would be for us to consider". You should skip over there and tell them what a great performance Horowitz gave in 1932 and how bad the noise is on that recording.

Hmm, maybe I'm the only one that likes noise with some of my recordings. Richter's 1965 performance of the White mass sonata??

You would probably love some recordings I have of Paderewski and Hoffman.

If one could remove the noise without altering the recording, I think you would love it. I hope you like that recording because it's good, not because it's noisy.

Nah, the haziness of the recording adds to the mysterious atmosphere! Same with Richter and Sofronitsky's rcordings of the 6th and 9th.

If I may bring into the discussion the idea of the 'uncanny valley' (which is a computer graphics and robotics term, but I think applies here). The more realistic (but not perfect) the images are, the more disturbing they get for us humans, until they get it perfect.

For more information:

http://en.wikipedia.org/wiki/Uncanny_valley

and on a more humorous side: http://www.youtube.com/watch?v=FKTAJBQSm10

with that in mind, I think we're are looking at a more "sociological" issue rather than anything else...

Most really excellent pianists relate to the particular piano they are playing and to the space in which it is being played. None of which is conveyed by this technology.

It's now obvious to me that I've been wrong about all of this.

Unless we use the same exact piano in the same exact condition in the same exact location, replicating the same temperature and humidity conditions, we're doomed to failure. What was I thinking when I stared this thread ... and we're not even discussing the emotional state of the performer or the audience.

Oh, the humanity.

In terms of what the post to which I was responding was suggesting, yes, failure is guaranteed, because the "exact nuance" they were talking about ain't gonna happen.

Ah yes, the nuance factor. I'll have to dig up my nuance meter and have it calibrated just to be accurately nuanced.

Replaying the nuances of dead performers is just a little creepy to me - it seems more of a parlor trick than something useful.

But I do see one application for this that might actually be useful. There are hundreds of thousands of student instrumentalists worldwide in need of good accompanists. Take, for instance, the Brahms clarinet sonatas. I really wanted to play these (on clarinet) when I was a student, but the piano parts turned out to be too difficult for anyone I could coerce into trying them.

Can this software respond in real time to another performer? Can it match changes in tempo and dynamics? Why not hire collaborative pianists to record all the standard repertoire, incorporate that with a computer that can provide the feedback from live audio, and sell a digital accompanist to schools. The software equivalent to a music-minus-one recording, but with the advantage of being able to respond in real time to the soloist, play at their tempo, and so forth.

Now that's something that would be useful instead of just a parlor trick...

Sam

Once you do that, try the video again. The Gould bit he played did not, in fact, sound like Gould playing. It sounded like a MIDI-driven player piano doing an approximation of Gould's playing.

You could match every conceivable nuance, but as long as somebody knows a robot is playing, they will say it sounds like a robot.

The same arguments can be rallied against all types of recordings (especially the old ones with all the static noise in the background). Even the most modern recordings have this shortcoming. The type of speaker or headphone you listen through, the type of format it is on (CD, mp3, vinyl, etc), the year in which it was recorded, the type of equipment used to record the music, other extraneous noises in the room that contain the stereo speakers, and the acoustics of the room itself in which the speakers are located all effect how the recording is heard and ultimately dramatically alters what the actual performer would have heard on the day of the recording. But also, most recordings are not just full single takes (most live recordings are included in this as well), they are bits spliced together to make the most satisfying artistic rendition that is almost impossible to do on a single full recorded take, which means that the pianist never actually played exactly what you are hearing through your modest sound system.

Incidentally, you can make similar arguments for a concert hall as well. A mezzo piano heard in the front row is going to sound different than a mezzo piano heard in the upper balcony. The direction that sound waves bounce off the walls can greatly alter how a piece is heard (this is why microphone placement is practically an art in and of itself). Et cetera.

The reality of the situation is that if you want an exact 1:1 correspondence between what the performer hears and what you hear you simply will never get it, but we can certainly try to get as close as possible and that is what this new technology is trying to do, which is why it should be embraced rather than shunned.

As to the argument that it lacks "humanity," whatever that means. I would like to give this analogy. I have a print of (the overly romanticised) Jacques-Louis David's "Death of Socrates" on my wall. This is not the actual painting, it has been copied digitally and then re-produced by a super high-res printer. Does the fact that I know that Jacques-Louis David did not paint this actual copy make it any less of a work of art? Is it any less enjoyable because of it? Is it anymore lacking in humanity? The answer is simply no.

No CD 318 no Gould!

You know that his really late recordings were done on a Yamaha and that his really early recordings (e.g. the first Goldberg's) were not on CD 318 right?

When YOU make a recording, do you ask the technicians to add a lot of hiss and turn down the sound of the piano?

I know of several cases where there's been an argument between the artist and the sound engineer. The artist would want to be able to hear the hall, recreating what one hears when they're at a recital - the piano at a slight distance and resonance around. But sound engineers like a more "pure" sound - piano only, preferring to add reverb in later so they can control the sound of the "hall."

I've never heard of a sound engineer preferring digitised reverb to recorded reverb. Most large spaces are recorded in such a way that reverb is picked up by extra mics in the distant portions of a space. These mics can then, after the recording process, be selectively faded in or out according to taste. But that is beside the point, because the argument you are describing is "what sound should we produce," not "how do we produce a certain sound." In the latter case, the sound engineer knows best.

Until someone actually does that, we won't know.

BTW, some of pieces from the Rachmaninoff "Window in Time" project don't sound like they are being made by a robot to me, and I wouldn't claim to be able to tell them from a human playing. I don't know if I think they sound like Rachmaninoff himself exactly, but they do sound human enough to me to pass.

The thing that seemed strange to me was the recording from the pianist's viewpoint. We all know that that is not the best position from which to hear the piano. We have to "suffer" through that when playing ourselves - why would anyone want to do that for a recording?!

I have this argument with every sound engineer every time I record. Without exception.