Artificial orchestras

This weekend's Wall Street Journal (no link, 'cause I read it in hard copy) included an article Fugue for Man & Machine [Weekend WSJ, May 5-6, 2007] that's blatantly designed to stir up the purists:

Amid all the troubles facing the classical music world in recent years..., none has mobilized musicians more than the emergence of computers that can stand in for performers... Now a new alliance of conductors, musicians, and engineers is taking a counterintuitive stance:  that embracing the science is actually the best hope for keeping the art form viable and relevant.

Keep in mind that arguments over "canned music" are nothing new.  My local ballet company uses recorded music for most performances, and has done so for years.  In Las Vegas, several of the show rooms use recorded tracks to augment the live orchestra, and they've been doing it since the 1970s.  And as the millennium opens, many touring show companies, and even some in London's West End, have added synthesizers and other electronic "enhancements," with an eye on cutting the body count in the orchestra pit.
So it's already being done.  And it can be done.  So as usual, the questions comes down to "just because we can do something, should we do it?"
As usual, the situation is a trade-off, and the trade-off is being negotiated at the margins: Orchestras are expensive beasts. The article cites the valid problem of "aspiring composers who [can't] afford to have their creations performed... can now commission a high-quality computer-generated recording for a fraction of the price." But it goes off the rails in the very next sentence: "...ommunities facing the loss of their orchestra... could keep performances in town-- even if it means a computer stands in for half of the players."

The technology has gotten better: The article's example is the Vienna Symphonic Library, an enormous collection of multiple samples (usually a "few dozen") of individual notes played by real musicians. These samples can be assembled jigsaw-puzzle style to form a complete performance. There's even software to help select which sample to use:

"...embody intelligent Performance Algorithms into an easy-to-use VST/AU plug-in to orchestrate an inconceivably vast array of samples. For the first time hundreds of inspiring articulations can be combined in one Preset and on a single MIDI track" -- VSL site

The result sounds rather authentic. The Journal authors confronted a couple of music professors with four (short) samples of Beethoven's Symphony 7-- three real orchestras, one artificial. Although neither professor specializes in music performance (one chairs the Composition Department at Eastman, the other is the dean of Music Technology at Berklee) the results are still embarrassing for the purists, as each professor initially identified one of the "real" performances as being the "artificial" one.

But all recordings are "artificial," even the "real" ones. Ever since recording moved from instantaneous-cut acetate discs, producers, even of recordings of  "real instruments," have used multi-tracking, overdubbing, and time-shifting (don't bring in the string section until you've figured out which performance you want to release). While this saves time and money (and indulges artists who can't figure out what they want to do until they get into the studio), it risks losing the immediacy of performance created in real-time. Even classical recordings frequently include an amazing amount of edits (nobody likes to talk about how many, but I know some engineers and producers) mostly because complete retakes, especially with a full orchestra, are expensive.

Live performances are another matter. First thing, recorded instruments still don't sound the same as instruments played live.  (Don't write letters.  Instead get your ears checked, and then go out and listen to some live music-- and I don't mean live music played through a public address system.)  Second, audiences are going to resist the idea that half of the "musicians" they paid to hear are really sound files.  If the "live" performance is really to be half-artificial, then what's the point?

And even if the notes and the sound are close to authentic, the subconscious interaction of musicians playing together still can't be duplicated by assembling individual performances. (Listen to the precision playing by Billy May's band, as the entire sax line bends the pitch as if played by one player, or the tightness of execution by Count Basie if you don't believe me.)  While good musicians can immitate this interactivity, even they will say it's not as good as the real thing.   The resulting lack of "soul" is a problem for many of today's recordings (pop and classical alike), and may partially explain the state of the music business:  Even though the average listener isn't "expert" enough to identify what's missing, she realizes something isn't there, which makes it lack excitement.  Result:  Falling sales.

But, as I said above, the use of artificial orchestras is primarily a cost consideration, and cost decisions come at the margins first of all.  The question will always be whether artificial is "good enough," not whether it is "as good."  For the audience, as artificial supplants real, the question is whether we will realize what we've lost in the process.  Or care.

Posted by: Old Grouch in In Passing at 02:24:54 GMT | No Comments | Add Comment
Post contains 846 words, total size 6 kb.