KERA Arts Story Search



Looking for events? Click here for the Go See DFW events calendar.

'Rhapsody in Blue' to Get 'World Premiere' – This Time, in Dallas.


by Jerome Weeks 12 Feb 2012 10:28 PM

Tuesday at the Meyerson, George Gershwin won’t be living and breathing – still. But he will be playing his Rhapsody in Blue with the Dallas Wind Symphony. Yes, it’ll take a couple of rehearsals. And some remarkable digital technology.

CTA TBD

Conducting the Dallas Wind Symphony in rehearsal, Jeff Hellmer consults with ‘George Gershwin,’ the computerized read-out of the Rhapsody in Blue score.

The Dallas Wind Symphony will present an all-Gershwin concert tomorrow for Valentine’s Day. The composer can’t be there. But KERA’s Jerome Weeks reports, George Gershwin will certainly play — thanks to some advanced digital help.

  • KERA radio story:

  • Expanded online story:

[Sounds of orchestra tuning up continue under.]

The Dallas Wind Symphony is rehearsing George Gershwin’s Rhapsody in Blue in a way no one has ever rehearsed it. Or ever played it.

Conductor Jeff Hellmer: “All right. Everybody ready?”

[Rhapsody starts and continues under].

The Wind Symphony is trying to recreate the performance of Rhapsody that took place on June 10, 1924 when it was first recorded. Paul Whiteman conducted, and George Gershwin was at the piano – just as they were when Rhapsody made its famous debut earlier that year in Whiteman’s “Experiment in Modern Music” concert in New York.  To fit on both sides of the 12-inch record, Gershwin’s first, ambitious orchestral work was shortened from 15 minutes to nine and half. Nonetheless, the 1924 recording is as close as we can get to that first Rhapsody, and the Wind Symphony has been configured just like Whiteman’s original, 28-instrument band, right down to a banjo.

Yet there’s no pianist.

That’s because George Gershwin’s part will be played by a Yamaha Disklavier PRO – a kind of super-computer player piano mechanism inside a classic, nine-foot concert grand. It’s essentially a “playback” instrument; what’s truly special here is what will guide the Disklavier: the Zenph Sound Innovations software that meticulously replicates a musician’s performance as computer data.

Jeff Hellmer, director of jazz studies at UT-Austin, will conduct the Valentine’s Day concert.

Hellmer: “The people at Zenph have painstakingly studied Gershwin’s piano performance and have attempted to codify every aspect of the 10,000 notes that Gershwin played. And then they have the piano recreate each of these notes as closely as possible – and in fact, eerily closely.”

A recorded performance is digitally broken down along a complex set of factors including not just pitch and volume but also a pianist’s entire ‘attack’ — his use of pedal, tempo, etc. In effect, Zenph ‘reverse engineers’ the performance, taking into account the player’s style, the particular piano he used, the recording environment — all the factors that led this performer to create this set of sounds at that moment. In 2006, the company released its first CD of such a recreated performance: Glenn Gould’s landmark 1955 album of Bach’s Goldberg Variations. It has since released  CDs of performances by such pianists as Sergei Rachmaninoff, Oscar Peterson and Art Tatum. The original source recordings are frequently in mono, and some, as with the 1924 Rhapsody, were recorded before the advent of electric microphones. At the time, all the musicians would have crowded around and played into a gramophone horn with the sound waves being etched into a master. Such an analog method was not kind to high or low frequencies.

These CDs have been received with great excitement in audiophile and music circles. The concerts – Zenph calls them “world premiere re-performances” – have been described as “ghostly” and “supernatural,” even “science fiction.” They allow people to hear these giants perform –“live” — in optimum audio conditions.

But as fascinating as all this is, it’s not the company’s ultimate goal. Zenph is not really in the recording and concert-touring business. It’s in the digital musical interface business, creating rehearsal tools and musical tutors for the computer. John Q. Walker is the founder and chairman of Zenph.

Walker: “So, interestingly for us, it’s not about the recording. It’s about the data.”

The performances and the subsequent recordings, he says, are really just beta tests proving they got the performance data right. Having created these highly nuanced maps of great musicians at work, Zenph is looking forward to ways to put them to use. What can we learn from them? Precisely how did this master play? And can that be broken down in such a way that it can be taught?

[music clip starts under]

This is from the iconic 1959 recording by Leonard Bernstein. It’s the more richly orchestrated arrangement of Rhapsody by composer Ferde Grofe in 1945 that most of us know.

But this is how George Gerswhin played it in 1924.
[clip]

Disregarding the tinny, scratchy sound quality, there are a number of striking audio features. The first is that there’s no percussion, none of the booming kettledrums in the Bernstein version that give the passage a swaggering, military air. The early gramophone technology actually could record some kinds of percussion, but quite often, they weren’t used — as with the Rhapsody. So the Dallas Wind Symphony has had to work out how to handle the percussion.

But even more striking is how fast Gershwin plays. This is not the lush or dreamy Rhapsody we often hear. Gershwin is brash, almost frenetic. He’s 25 years old, the world is coming to hear him play something more than one of his pop tunes — and he’s showing off.

Walker: “We have these expectations from pianists interpreting the score these many years. And George just goes for it [laughs].”

Hellmer: “You have to remember that Gershwin was what they called a ‘plugger’ back on Tin Pan Alley – where he would play popular tunes in the window of a publishing house to attract sheet-music buyers. And the way that he played Rhapsody in Blue in 1924 seems very similar to how he might have plugged a song.”

Right now, that’s a headache for the Dallas musicians. The Whiteman band had played Rhapsody half-a-dozen times before they recorded it, so they knew what to expect — showing off like he did, Gershwin didn’t keep tempo. He sped up and slowed down all over the place. The Wind Symphony is trying to shift with Gershwin on the fly; it’s a first rehearsal for them and it’s stop-and-go. Hellmer conducts from a Zenph-designed computer display of the score, but he’s also listening to an audio ‘click track’ in an earphone, a track that registers Gershwin’s accelerations and hairpin turns.

But capturing and accompanying that brashness and energy — that’s the spirit of the brilliant, young pianist the Dallas Wind Symphony hopes to conjure onstage Tuesday night.

SHARE
  • What an awesome story! Thanks.

  • What an awesome story! Thanks.

  • nmlhats

    Cliburn at the Modern is having a Zenph event on April 14th. Pianist Jose Feghali will play and his performance will be captured by the technology.

  • Wow cool I have a Disklavier, I wonder if I can get the recording for my piano?

  • Wow cool I have a Disklavier, I wonder if I can get the recording for my piano?

  • Anonymous

    The Dallas Wind Symphony WILL record their ‘Rhapsody in Blue’ performance, but as a regular old (digital) recording. I don’t know whether that will ‘play’ on a Disklavier. I don’t know enough about Disklaviers — I do know, however, that the Disklavier PRO is a significant upgrade from the original Disklavier. You might be able to find out more info through the Zenph website.

  • JeromeWeeks

    The Dallas Wind Symphony WILL record their ‘Rhapsody in Blue’ performance, but as a regular old (digital) recording. I believe you should be able to play that on your Disklavier — I don’t know enough about Disklaviers to say for certain. I do know, however, that the Disklavier PRO is a significant upgrade from the original Disklavier.

    UPDATE: See my response to Andre Speyer, below.

  • Andre Speyer

    Maybe not a world premiere. Performed by the Cape Cod Symphony three years ago.

  • Andre Speyer

    Maybe not a world premiere. Performed by the Cape Cod Symphony three years ago.

  • Anonymous

    In one way, you’re right: The Cape Cod Symphony live concert would have had, essentially, some of the same effect that the DWS should: the ‘Rhapsody’ being performed with the piano seemingly being played by the ghost of Gershwin and an orchestra accompanying it (him).

    But otherwise, I’m sorry, you’re wrong. What the Cape Cod Symphony did — according to the Cape Code Times (http://tinyurl.com/yh7g5yp) — was a recreated ‘Rhapsody’ using the original ‘Rhapsody’ piano roll, the template that powers a player piano — with the piano roll converted to a CD and played on a Disklavier (which answers bluen’s question above — when the DWS recording pf ‘Rhapsody’ is available, he should be able to run it through his Disklavier). Actually, the piano roll of ‘Rhapsody’ has been used quite often for recreated performances like this, some are available on CD. But it is NOT the same thing as what Zenph is doing.

    Zenph developed their data from the 1924 gramophone recording and created something much more detailed than a piano roll. In fact, I asked Walker if they could recreate a performance from a piano roll rather than a recording, given the low audio quality of some early wax-cylinder and shellac-disc recordings. He said, more or less, that they could, but it really wouldn’t be worth it because a piano roll offers such little data. Basically, it’ just gives you the right notes in the right order; there’s no nuance, very little of the ‘performer’ in it.

    What’s more, it’s been rumored that after ‘recording’ the piano roll, Gershwin went back and poked holes in it — to add more notes. So it’s not even ‘authentic,’ reflecting what Gershwin actually played.

  • JeromeWeeks

    In one way, you’re right: The Cape Cod Symphony live concert would have had, essentially, some of the same effect that the DWS should: the ‘Rhapsody’ being performed with the piano seemingly being played by the ghost of Gershwin and an orchestra accompanying it (him).

    But otherwise, I’m sorry, you’re wrong. What the Cape Cod Symphony did — according to the Cape Code Times (http://tinyurl.com/yh7g5yp) — was a recreated ‘Rhapsody’ using the original ‘Rhapsody’ piano roll, the template that powers a player piano — with the piano roll converted to a CD and played on a Disklavier (which answers bluen’s question above — when the DWS recording pf ‘Rhapsody’ is available, he should be able to run it through his Disklavier). Actually, the piano roll of ‘Rhapsody’ has been used quite often for recreated performances like this, some are available on CD. But it is NOT the same thing as what Zenph is doing.

    Zenph developed their data from the 1924 gramophone recording and created something much more detailed than a piano roll. In fact, I asked Walker if they could recreate a performance from a piano roll rather than a recording, given the low audio quality of some early wax-cylinder and shellac-disc recordings. He said, more or less, that they could, but it really wouldn’t be worth it because a piano roll offers such little data. Basically, it’ just gives you the right notes in the right order; there’s no nuance, very little of the ‘performer’ in it.

    What’s more, it’s been rumored that after ‘recording’ the piano roll, Gershwin went back and poked holes in it — to add more notes. So it’s not even ‘authentic,’ reflecting what Gershwin actually played.

  • Kiwiflite

    Back in the late nineties, we played a version (with the now dissolved Syracuse Symphony) of Rhapsody in Blue that sounds remarkably similar to that described in the story. The technicians had started with a version Gershwin recorded when he played both the “orchestra parts” and the solo piano parts. The technicians then removed all the orchestra’s notes from the digital files (since the piece is largely in “call-and-response form – just like a real “Blues,” this was not destructive of the original solo piano line. A click track was then added; I, as conductor, wore headphones, and we more-or-less pulled it off in several live performances. We all remember, I am sure, how (recklessly) fast Gershwin played: in modern parlance, one would say that he “had a tendency to, ahem, rush.”
    It was a curiosity, really. Hard to say that there was nearly the artistic/creative tension that exists when a living person is playing it, live. But, also important to point out that, contrary to the strong implication of this Art Seek article, this is not exactly the first time this has been attempted. I am sure there are many other examples other than Cape Cod and Syracuse…

  • Kiwiflite

    Back in the late nineties, we played a version (with the now dissolved Syracuse Symphony) of Rhapsody in Blue that sounds remarkably similar to that described in the story. The technicians had started with a version Gershwin recorded when he played both the “orchestra parts” and the solo piano parts. The technicians then removed all the orchestra’s notes from the digital files (since the piece is largely in “call-and-response form – just like a real “Blues,” this was not destructive of the original solo piano line. A click track was then added; I, as conductor, wore headphones, and we more-or-less pulled it off in several live performances. We all remember, I am sure, how (recklessly) fast Gershwin played: in modern parlance, one would say that he “had a tendency to, ahem, rush.”
    It was a curiosity, really. Hard to say that there was nearly the artistic/creative tension that exists when a living person is playing it, live. But, also important to point out that, contrary to the strong implication of this Art Seek article, this is not exactly the first time this has been attempted. I am sure there are many other examples other than Cape Cod and Syracuse…

  • Anonymous

    As I indicated in my earlier response, there have been numerous ‘recreations’ like this attempted over the years, and some are even available on CD. But the Zenph version entails a different level of digital sophistication. The process is similar: The orchestral parts are stripped out of the recording, leaving Gershwin’s piano playing. But you don’t indicate what happened next — that is, what was done with the remaining audio and how it was performed. I’m assuming that the piano part was simply transplanted to CD and played by the Disklavier?

    Please remember that the Disklavier PRO (introduced into the US in 1999) has much greater precision, a much finer “capture” of a performance. To give you some idea: Yamaha upped the velocity levels of MIDI (musical digital interface data) of the then-standard 127 to 1023 — almost a factor of ten.There are now 256 increments of measurement for just the use of the pedal.

    But that’s simply the playback instrument. The heart of the process is the digital analysis and detail that Zenph extracts: The 10,000 (or so) notes that Gershwin plays on the piano were transformed into more than one million bits of data — for a musical piece that, when it’s boiled down to just the piano, isn’t half the length of the nine-minute version of the ‘Rhapsody.’

    Yes, the basic idea of ‘recreating Gershwin at the piano’ has been around even before the Syracuse performance.But the execution has not been done at this fineness, this “aliveness.” Or so I’ve been told.

    I have the same reservations that you do about the symphonic accompaniment with what is essentially a pre-recorded track. As you note, there’s no real give-and-take, only ‘take,’ as it were. But Zemph has also produced CDs of ‘interactive re-performances’ (Joshua Bell with Rachmaninoff, Zuill Bailey with de Falla and Granados) that have gotten raves. So I await the DWS recording with curiosity and an open mind.

    • Kiwiflite

      I love your characterization of this as all “take.” Spot on! And, I have an open mind, also. What you are saying is that the PRO version has far greater nuance when compared with earlier versions. This is where it gets really interesting, because our experience in Syracuse and, it seems, the experience of the musicians in Dallas, is the LACK of nuance that “Gershwin’s” playing demonstrates.

      Don’t get me wrong. I am not trying to diss Gershwin’s playing. Far from it. But, those of us in the business of performing for a living in front of a live audience know one thing for certain: the space you are playing matters.

      Gershwin recorded whatever particular performance has been digitized in a given room or space. I doubt it seated one or two or three thousand people. Any attempt to overlay his original performance with a live accompanying group in a big (reverberant) space is going to involve a massive disconnect.

      All that aside, my reading of the original article leads me to conclude that the reader ids supposed to think that this whole process is being attempted for the first time. It is not. You may be correct that this particular version is new; that this particular version is more nuanced. But it is not, as the article attempts to demonstrate, a fundamentally new idea.

      I wish them well.

  • JeromeWeeks

    As I indicated in my earlier response, there have been numerous ‘recreations’ like this attempted over the years, and some are even available on CD. But the Zenph version entails a different level of digital sophistication. The process is similar: The orchestral parts are stripped out of the recording, leaving Gershwin’s piano playing. But you don’t indicate what happened next — that is, what was done with the remaining audio and how it was performed. I’m assuming that the piano part was simply transplanted to CD and played by the Disklavier?

    Please remember that the Disklavier PRO (introduced into the US in 1999) has much greater precision, a much finer “capture” of a performance. To give you some idea: Yamaha upped the velocity levels of MIDI (musical digital interface data) of the then-standard 127 to 1023 — almost a factor of ten.There are now 256 increments of measurement for just the use of the pedal.

    But that’s simply the playback instrument. The heart of the process is the digital analysis and detail that Zenph extracts: The 10,000 (or so) notes that Gershwin plays on the piano were transformed into more than one million bits of data — for a musical piece that, when it’s boiled down to just the piano, isn’t half the length of the nine-minute version of the ‘Rhapsody.’

    Yes, the basic idea of ‘recreating Gershwin at the piano’ has been around even before the Syracuse performance.But the execution has not been done at this fineness, this “aliveness.” Or so I’ve been told.

    I have the same reservations that you do about the symphonic accompaniment with what is essentially a pre-recorded track. As you note, there’s no real give-and-take, only ‘take,’ as it were. But Zemph has also produced CDs of ‘interactive re-performances’ (Joshua Bell with Rachmaninoff, Zuill Bailey with de Falla and Granados) that have gotten raves. So I await the DWS recording with curiosity and an open mind.

    • Kiwiflite

      I love your characterization of this as all “take.” Spot on! And, I have an open mind, also. What you are saying is that the PRO version has far greater nuance when compared with earlier versions. This is where it gets really interesting, because our experience in Syracuse and, it seems, the experience of the musicians in Dallas, is the LACK of nuance that “Gershwin’s” playing demonstrates.

      Don’t get me wrong. I am not trying to diss Gershwin’s playing. Far from it. But, those of us in the business of performing for a living in front of a live audience know one thing for certain: the space you are playing matters.

      Gershwin recorded whatever particular performance has been digitized in a given room or space. I doubt it seated one or two or three thousand people. Any attempt to overlay his original performance with a live accompanying group in a big (reverberant) space is going to involve a massive disconnect.

      All that aside, my reading of the original article leads me to conclude that the reader ids supposed to think that this whole process is being attempted for the first time. It is not. You may be correct that this particular version is new; that this particular version is more nuanced. But it is not, as the article attempts to demonstrate, a fundamentally new idea.

      I wish them well.

  • Terry Paul

    In the late 1970’s, Reno Philharmonic performed with the “Ghost of Gershwin’.. the grand piano was elaborately hooked up with wires etc, to a digitization of one of George Gershwin’s Rhapsody in Blue Performances… all the key-touch, pedal action, dynamics, tempi, were clearly performed! He definitely performed it slower compared to current performance practices. It was such a privilege to experience this as a violist in the orchestra!