A live six-keyboard orchestra (the MIDI Wurlitzer?) goes on tour with Philip Glass's scores for Koyaanisqatsi & Powaqqatsi.
As consumers, we tend to take movie music for granted - it's all part of the soundtrack, mixed in with the dialog and the sounds of screeching tires and exploding heads. Well, youngsters, back in the old silent film days (imagine the sound of creaking bones here), the music was all the sound there was, and it was usually provided live, by a keyboardist holding forth at the mighty Wurlitzer pipe organ - quie often improvising to the moods of the film images.
These days, a movie score is a complex beast, played by a large orchestra of acoustic and electronic musicians. Occasionally there are concert performances of a suite of music form a popular film, but you almost never hear of a recent movie presented with live soundtrack accompaniment.
So, image how Michael Riesman felt last fall when he was given the task of preparing a live performance version of the Philip Glass score to the film Powaqqatsi, a version that could be taken on the road and performed with the film. We're not just talking about your standard complex film score: In Powaqqatsi, Philip Glass expanded his signature churming rhythms nd the keyboard/woodwind based sound of the Philip Glass Ensemble with a veritable army of Third-World musicians and exotic instrument sounds. It was extremely complex to record; it would be even more difficult to perform live. (For more on the original recording of the Powaqqatsi film score, see the World View section in the January '88 issue of Keyboard.)
Riesman was the right person for the job: As musical director and conductor for the Philip Glass Ensemble, he has conducted numerous Glass recordings, including both Koyaanisqatsi and Powaqqatsi. He has also conducted and performed on albums by Paul Simon, Scott Johnson, Mike Oldfield, and Ray Manzarek. He studies at the Mannes College of Music and at Harvard University, where he earned a Ph.D. He has thaught at both Harvard and SUNY-Purchase, has received numerous grants and commissions, and has released a solo album, Formal Abandon [Rizzoli].
And so, without further ado, here is Michael Riesman, telling how he got Powaqqatsi ready to go on the road.
-- Editor
Project description
Powaqqatsi is the name of the second film directed by Godfrey Reggio with music by Philip Glass. Similar in concept to their first film, Koyaanisqatsi, Powaqqatsi is a film without dialog or actors; it is an essay about people and habitat. The soundtrack consists entirely of music, with a few sound effects. Powaqqatsi (a Hopi word that means "life in transition") focuses on the Third World and the impact of technology on traditional societies.
Powaqqatsi Live is a presentation of the film with live musical accompaniment. This accompaniment is an exact replication of the recorded soundtrack, transcribed for a small orchestra (11 players) consisting primarly of keyboards.
Powaqqatsi Live premiered at the Serious Fun festival in Lincoln Center on August 27, 1990, with a repeat performance the following night. A European tour was booked for April-May of 1991. In September, we'll be in London, playing both films at the Festival Hall. A U.S. tour is planned for the 1991-92 season.
On the European tour, we performed both Powaqqatsi and Koyaanisqatsi. Koyaanisqatsi Live was first performed four years ago; it's been in the repertoire since '87. It's not as technologically complicated as Powaqqatsi.
Production notes
The first stage of preparing for the presentation of Powaqatsi Live was to determine the size of the orchestra needed to perform the score. I wanted the group to be as small as possible, to enable the show to be an economically viable touring entity. After studying the score (with which I was already intimately familiar, as the conductor of the original recording), I determined that it could be done with six keyboard players, three woodwinds, and two percussionists, plus myself as conductor. One of the male players would also be required to sing the middle-Eastern vocal solo in reel five of the film. Two female voice were necessary as well, to be drawn from existing players.
The core group was the Philip Glass Ensemble, consisting of three keyboard players, three woodwinds, and soprano voice (doubling on fourth keyboard). Additional musicians were recruited to complete the orchestra with the singing requirements taken into account. Most of the people who joined us had worked with us before.
The transcription of the score for the group was fairly complicated, like a jigsaw puzzle or the computer game Tetris. Initially, I had to make sure that each played could (a) physically cover the notes (especially the percussion section); (b) have sufficient time to change patches and/or disks; and (c) address hardware in a sensible way (splits, layers, percussion pad assignments, and MIDI routing and channeling). At the time of transcription, I addressed the hardware issue only in a general way; the specifics would be worked out during programming. I did, however, create a basic hardware specification that generally was accurate (we ended up using eight samplers for the initial performances, instead of the six units I originally thought would suffice).
I then assembled a complete score from the manuscript, which was written to individual cues, and correlated it with the film soundtrack, which had some additional cuts and repeats. The corrected score was delivered to the copyist with about 30 pages of instructions for the transcription.
Upon the return of the score and finished parts, I began an intense two-week period of programming with three assistants. I created all of the synthesizer patches and fine-tuned the sample disks. The assistants were kept busy with sampling, assigning keyboards for samples, assigning zones and program change maps on master keyboards, assigning percussion pads and adjusting velocity response on the pad-to-MIDI convertors and sample disks, copying disks (backups!), going shopping for additional hardware and software, phoning for technical support, making custom cables, transplanting gear from rack to rack, operating the monitor board, and last but not least, keeping meticulous notes, an absolute necessity in a project of this magnitude.
Considerable effort ws expended merely in laying out the forms for note-taking; we evolved seperate forms for (a) available (unused) patch numbers for each synthesizer (updated regularly); (b) multiple-column master-keyboard information on zones, MIDI channels, units addressed (with patch number or disk number and program); (c) sample disk description (sample name, keygroup, range, program number, memory used); (d) a master list of disks; and (e) a mix outline, created on a computer database (FileMaker II), showing who plays what unit and when, and what sound emerges. These notes, kept together on a clipboard, eventually nearly outgrew the jaws of the clipper.
To allow time to make adjustments in the programming and, if necessary, in the transcription, I scheduled the rehearsal period in two stages: an initial period of sectional rehearsals (over five days) followed by a four-day break for adjustments; and then a final two days on a soundstage with full ensemble and P.A., followed the next day by an afternoon dress rehearsal in the hall and the evening premiere.
As much as possible, balanced in the keyboard section were adjusted at the programming level rather than on the mixing consoles. At the first keyboard rehearsal, all faders on the monitor mixer were set at unity gain. If a patch or sample came up too soft or too loud, the relevant fader was raised or lowered the required amount, and a note was made of the offending patch and the adjustment necessary to bring it into line. At the end of the day, patches and/or disks were rewritten with the logged volume shifts. This process was effective, and at our first full rehearsal, only minimal further tweaking was necessary.
To stay in sync with the picture, I relied entirely on visual cues. I made two sets of colored markings in my score: red ones for musical events (time signatures, tempos, important entrances, repeats) and green ones for cues from the screen (cuts, action, reel changeovers). I logged the visual cues very accurately (to fractions of a beat). Using my green marks, I could tell instantly whether I was in sync or not, and if not, how far out of sync I was. I would glance ahead in the score for an upcoming green event, then watch the screen for its appearance. If it occured ahead of the indicated position, I was behind the picture, and I would (hopefully imperceptibly) pick up the pace a little to catch up. If it was early, I would put the brakes on. I instructed the orchestra never to jump to a new tempo when they noticed me ahead of or behind them, but to adjust slowly. The performance was thus something of a tug-of-war with the players, requiring from me an iron-clad sense of tempo, never going in to any tendency of the orchestra to rush or drag.
Programming notes
For programming, all keyboards and synth racks were arranged in a semicircle. A Machintosh SE, running Opcode editor/librarians for the Yamaha DX/TX modules and the Oberheim Matrix-6, was connected to the MIDI interface (Opcode Studio Plus 2) with extra-long cables, and sat on a rolling road case to permit it to move to the various stations. The Roland Super Jupiter Programmer was used with the MKS-80.
Samples were recorded and/or processed on a Mac II running Digidesign Sound Tools, then transferred to the S900 and 950s via MIDI. Some samples existed from the original recording of the score; others were transferred from an E-mu Emulator II; some were culled fromthe multitrack tapes of the original recording; and some were created specifically for this production.
As a rough estimate, over 400 person-hours were expended on the creation and fine-tuning of over 300 samples or patches.
All brass sounds (except tuba, on the Juno-106) were combinations of samples and Yamaha DX/TX modules. The samples remained the same for the whole show, with keyboard velocity response routed only to control volume. The DX/TX patches changed according to the nature of the music, and provided the velocity response to tone color (brilliance) that gave the player a wide range of expression. For example, with volume pedal full and a low keyboard velocity, a mellow, round forte was produced. With volume pedal at half and a high keyboard velocity, a hard sforzando resulted. Keyboard aftertouch was used as well for expression.
String sounds were created in the same way as brass sounds, but using more synth modules (DX/TX plus Roland Super Jupiter and Oberheim Matrix-6 analog synths).
The balafon, an African marimba, was recreated using 14 samples, almost a seperate sample for each pitch it played. The dousson'gouni, a buzzing harp-like African instrument, was recreated using only four samples. The kora, another African harp, wasn't sampled at all, but imitated with nylon string guitar samples supported by synthesizer. Various drones (tamboura, didgeridoo, human voice) were recreated using long (15-second) loops; the tamboura and didgeridoo were partly second-generation samples, since we had used a combination of samples and live instruments to produce the drones on the original recordings, from which the samples for the live show were taken.
A number of samples were created to imitate exotic instruments but made using Western instruments (as in the case of the kora). Another dousson'gouni, in addition to the "real" one, was made using a paper clip rattling on a bass guitar string. This sound had more bite and bottom than the real one and worked better in an ensemble context. The real one was used only in a solo. For the surdos (Brazilian samba bass drums), we used a tuned orchestral bass drum sample instead, even on the original recording, because it simply sounded better than real sundos with live players. (In the original recording, almost every percussion track was recorded twice, once acoustically and once as a sampled sound. We ended up using about 30% of the sampled sounds in place of the acoustic ones in the final mix, generally because the samples were cleaner and cut through better.)
Sampler switch-over
Between the initial performances and the European tour, we decided to revamp our sampler and use Digidesign SampleCell boards for the Mac. The idea came from Kurt Munkacsi, our producer, and Paul Rice (of Sam Ash Pro). We made a tentative decision around Christmas, and got our first SampleCell card in January, to check it out. We decided that it could be done, and got going. We started with several units: I wasn't exactly sure how many cards I was going to need, since we were replacing eight samplers.
I eventually narrowed it down to using three cards, which were pretty full. They each had 8Mb of RAM, and I was using close to 20Mb of samples stored on the three cards. I think I had one output to spare and one or two instruments here and there; I was just right up to the limit.
At first I was somewhat disturbed that there was no way to use patch change information from a controller to call up different instruments, but I realized that with judicious use of MIDI channel assignments, I could get around it.
The beauty of the system using the SampeCells, in addition to the improved sound quality, is that nobody has to swap disks during the show: It's all in RAM.
We reorganized all the samples. It involved about three weeks of work. We also decided that it was risky business to put all of our eggs in one basket: Being at the leading, bleeding edge of technology - going out with a new system - we decided that we didn't dare to do it without a backup unit. We would like to thank Sam Ash Pro for loaning us a backup Mac IIfx and to Digidesign for loaning us three spare SampleCell cards.
And yes, the backup was needed on the tour. We eventually worked out a way to have both units hooked up and running at showtime, and we had a single video monitor with a switch, so we could look at either one. If we had to change over, it was just a matter of pulling patch cords out of the back and plugging in the other unit.
GlassPages - Philip Glass on the Web
[ 
]