By Erin Sherbert
By Howard Cole
By Erin Sherbert
By Erin Sherbert
By Leif Haven
By Erin Sherbert
By Chris Roberts
By Kate Conger
As Greg Tomassian, lead singer of Soul Candy, steps onto the slate-black auditorium stage, the din of the crowd sweeps over him like a warm tide. When he snaps his microphone off its stand and draws it to his parting lips, the cheers amplify into a throbbing, meaty roar. And as he rips into the first bars of "The Midnight Hour," Tomassian's brain, his flesh, and his vocal chords ripple with the energy of his thousands of cheering fans.
"You step out with all the eyes in the place upon you," explains Tomassian, 35. "The darkness is just a sea of heads, all bobbing up and down rhythmically to the sounds that you are creating. It's the whole energy-flow of humanity. It's an energy interexchange."
That's the general idea, at least. For now, though, Tomassian sticks with his day job as owner/manager of Cash for Cars in San Jose. He gets together with some pals once or twice a week to play R&B and rock 'n' roll cover tunes, calling the band Soul Candy. But superstardom has so far eluded Tomassian, in part because he can't sing on key. The high notes in the song "The Midnight Hour," for instance, sometimes crack into a flat, grating dissonance. His segues from one pitch to the next occasionally wade into unintended, and unfortunate, territory. The tones of Tomassian and his lead guitarist tend to part ways a bit too often.
"I started taking vocal lessons and working out with this band," Tomassian says. "But whenever I would get around people who knew how to sing really well, I was intimidated, because I would go sharp and go flat where they would stay on key."
Last month, though, Tomassian discovered a miracle that promises to end these woes and, he hopes, clear Soul Candy's path to the big time. It's called AutoTune, a digital sound board containing algorithmic pitch-correction software that forces even the worst singers to warble on key, without otherwise distorting the sound of their voices.
"When I saw this product I went crazy," Tomassian recalls. "I don't have the time to work out as much as it would take to get to that next level. But when I used it for the first time last Thursday night, I was absolutely astonished. I've never had such a good night. When my voice broke up in the higher registers, AutoTune smoothed it so beautifully and naturally that the guitar player was looking at me, giving me the thumbs up. I was motivated to try all sorts of things I never had before."
Tomassian's dream -- to create music talentlessly, yet well -- is about to become reality for the multitudes. A handful of Silicon Valley inventions may allow even the most unimaginative, tone-deaf, arrhythmic dabbler to compose, sing, or otherwise create music virtually indistinguishable from the work of masters.
There's Emmy, a Santa Cruz computer program that creates brand-new symphonic works in the style of the great classical composers. There's Xx, an algorithmic compositional tool that allows untrained slackers to write music by pointing and clicking through an easy-to-use Macintosh interface. There's Harmony, a soon-to-be-released program designed to allow singers to replace their own voices with others of their choosing. These innovations come on the back of a generation of audio engineering technology that makes buying a CD the same sort of what-you-see-is-not-necessarily-what-you-get enterprise as viewing an episode of Baywatch. Digital-audio looping devices turn the art of creating a groovy string of jazz riffs into a series of mouse clicks; rhythmic correction software forces a wayward voice, guitar, drum, or bass back on beat; and tone correction software turns a thin violin or oboe track fat.
And there's AutoTune, a black box about the width and thickness of five SF Weeklys that immediately, and seamlessly, corrects the pitch of miked singers to the next on-key tone.
Using these technologies, Tomassian and his ilk will conceivably be able to belt out a computer-composed song through computer-generated pipes, corrected to a computer-programmed tempo and pitch, and come out sounding not half-bad.
While this thought may delight moonlighting used-car dealers, it dismays our musical elite. When any slob can compose like Rachmaninoff, when every Joe Six-Pack possesses the musical soul of Ol' Blue Eyes, when laying down cool jazz becomes the intellectual equivalent of a video game -- whither our conservatories and back-porch blues jams? Will art and culture become functions of algorithmic complexity? Will the muse disappear from music?
Robert Greenberg, chair of the department of music history at the San Francisco Conservatory of Music, is among those who hope not.
"I don't believe that a true writer or a true composer or a true painter will want to abrogate their role of creator to a machine," Greenberg says. "It's an exercise. It's missing the essential thing that humans add, and that's consciousness. If someone says to me, 'Here's my symphony that was composed by a computer,' I'm simply not going to spend the half-hour it would take to listen to it."
As for performance enhancers such as AutoTune, Greenberg urges his colleagues to man the barricades.
"No real singer worth their salt would ever get near such a thing. This is an advanced karaoke box," Greenberg says. "The Milli Vanilli syndrome is the tip of the iceberg. You have Mick Jagger stuffing a roll of socks in their crotch. Everyone's getting silicone implants. It's part of the same continuum."
While Greenberg scoffs, others are taking the promise of true artificial musical intelligence very seriously.
Members of the Wemberley Winds, an ensemble in England, are right now preparing their pipes for a November recital at Oxford of a brand-new Gustav Mahler opera. Never mind that Mahler died 87 years ago. The score of the new version was composed entirely by a Santa Cruz computer.
And recording engineers -- even those who work with serious singers -- now use AutoTune almost as a matter of routine: "I'd say I've sold about 100 AutoTune boxes," says Ted Bahas, a salesman at Sam's Music in Nashville. "And I've sold another 130 of the AutoTune computer plug-in."
By one measure at least, David Cope, chair of the department of music at UC Santa Cruz, is the greatest composer to have ever lived. While pretenders such as Wolfgang Johannes Mozart composed fewer than 50 symphonies in a lifetime, Cope has penned 1,500. He has also written 1,500 ensemble pieces. In all, his catalog of classical compositions includes 5,000 scores of various types.
"It's really simple," says Cope, explaining his compositional technique in a dimly lit office at UCSC's Porter College. "You pull down the input menu, and what you do is you load the database. This is, in some ways, the most important part of the process. So, you load it -- or you can create a brand-new database. You don't want any mistakes in it. If you have a single mistake in a database, one wrong note, it becomes amplified in the composing process. So you will have 10 or 12 wrong notes possibly in the output, and it sounds miserable, so you have to be sure your data is absolutely correct. That's Step 1."
Cope is describing his work with the program Experiments in Musical Intelligence, "Emmy" for short, which he has used to compose most of his own massive musical oeuvre. Run the program once, and you have a brand-new Cope composition based on the style of the professor's previously composed works. Leave the machine running a few days, and you have thousands.
Like the IBM's Big Blue chess machine, Cope exploits a computer's raw number-crunching power to make a facsimile of human creative prowess. First, he dumps the several dozen compositions he wrote without the help of a computer during the 1960s, '70s, and '80s into a database, which the computer program chops up into identifiable musical segments. The computer then mines these segments for signature patterns. Having thus divined a composer's "style," it reassembles these oft-used flourishes into a new order, like letters inside a flexagon, using as a template organizational benchmarks also gleaned from the composer's earlier work.
The UCSC Orchestra last year performed a 42nd symphony of Mozart's, which, despite lacking a certain, shall we say, life, was nearly indistinguishable from the 41 such works Mozart composed before he began to decompose.
The Emmy program's uncanny ability to mirror the style of great composers earned Cope a cover story in the British magazine New Scientist last year, and dozens of Luddite editorials in British newspapers.
But like a hotel cover band that can't get anyone to listen to its original tunes, Cope has had far less luck getting his own computer oeuvre performed. This troubles Cope, because he designed the precursor to the Emmy programs during the 1980s as a way to circumvent a bad case of composer's block. Now, he says, he considers his computer to be much more than a mere compositional tool. For him, Emmy is the second member of a composing duet, a Hammerstein to his Rodgers.
By his description, one wonders if it's more a monster to his Frankenstein.
"We're in the process of creating new beings: mind-children, if you will. We are in the process of creating a new kind of life," Cope explains. "How should we feel about these beings that we bring into the world? I consider them our children."
Cope has had a difficult time convincing the musical establishment of this.
Of the 12 orchestra conductors he has sent his "Symphony No. 1383" to, none has shown any interest. They're scared, he says: "I believe that down deep inside there is a computer phobia."
Undeterred, Cope revels in the fact that he hasn't had a composer's block since the '80s. Whenever he gets stuck, he just asks Emmy for ideas.
"Before, I would be given a commission and would have to say that a piece was going to take one or two years to finish," Cope says. "Now, I say two weeks."
As accomplished as Cope's program is, its painstaking data-entry procedures and its ancient LISP programming language put it out of reach of all but the most dedicated computer-music buff.
Enter Eric Wenger, the Parisian mathematical genius poised to release the Mozart in Everyman. Wenger was already a Mac-world legend for his landscape-illustration program, Bryce -- which allows computers to paint photographlike depictions of oceans, mountains, and other images -- when he moved from Paris to a Castro District condominium a year ago.
"It's a lot easier to do business here," explains Wenger, who, with his pale, scraggly bearded face, flaccid slouch, and up-to-the-moment leather-pants-and-guayabera leisure ensemble, looks exactly like what would result if a Paris hipster were crossed with a Silicon Valley geek.
A painter, a musician, and an electronic engineer, Wenger has combined his talents into a program that allows even the most musically disabled among us to create soul-stirring pieces. His first musical program, called Metasynth, created a new musical composing language in which composers use the tools of an illustrator -- brush strokes, colors, lines -- to write scores. Xx pulls further away from Wenger's illustrator roots, using an interface that looks like the inside of a grand piano.
Need a sonata? A concerto? A rag? The pianolike mouse interface allows a pretender to sprinkle notes over a musical grid like pepper, and watch them fall into arrangements that provoke an emotional response not unlike the one a concertgoer feels.
If a composer likes, he can turn on a random note generator, whose output is filtered through a series of instructions that force the random notes to comply with a particular musical style. The aspiring composer can also purposely choose notes -- which are automatically kept on key -- then adorn them with classical flourish, using a series of pointer-icons, and come out with something that sounds like a sophisticated piano concerto.
"You give a random note generator a rhythmic pattern, give it a tonality, give it a range of notes -- pitch range as well as a scale -- and give it a dynamic range," says Len Sasso, a Carmel composer. "It takes all those parameters, and it takes another parameter that decides whether to look back and decide whether to use a note it had used before. The notes that were chosen within those limits are chosen at random, but it doesn't really spit out a random sequence of notes."
Using the program to compose at his Castro home office, Wenger's body shivers after he creates a particularly satisfying bar of computer-composed music.
"This is a little bit too random to use, so I'm clipping this piece here, repeating it, and I just click here to remove all the dissonance," says Wenger, using his program to repair a series of notes chosen by a visiting rube. "I like those chords you've created. They're not uninteresting at all. Another interesting thing is to compose using the capture feature, which plays a loop arpeggio, then adding notes on as they sound good."
The completed work, with additional notes airbrushed in, key passages looped on, and with a repeating canon effect added, sounds perhaps a bit dirge-ish, but certainly the work of a journey-man. Elapsed composing time: around 10 minutes.
"You can really create a lot of music very fast," Wenger notes.
Wenger first wrote the program as he began dabbling in computer music several years ago.
"I started doing music at 16 in a pop group. I composed a lot of pop music. My basic formation is in graphic design and illustration, and that's how I got into computer graphics. I did painting software. I went into computer music when I was 29 or 30, and I was using a four-track pre-recorder. It seemed like you had to do a lot of work to do something that wasn't necessarily that good. After I got involved in computers, I realized that there were things you could do in musical composition as well. At this time, my tastes were moving toward classical.
"I have always had a habit of creating my own tools."
While he doesn't quite match Wenger's Euro-geek flair, Andy Hildebrand, a long-tall ex-oilman from Texas, has cooked up an algorithm perhaps as revolutionary as Wenger's. Through a series of ingenious mathematical instructions he's tucked into an audio engineer's black box, Hildebrand has managed to quiet the dissonance of a thousand cracking vocal chords. Hildebrand is the creator of AutoTune, a one-time retirement hobby that he is now turning into a growing business.
After a career of designing seismic oil-exploration equipment, Hildebrand retired a couple of years ago to pursue his dream of becoming a classical composer. During the previous two decades, Hildebrand had used his skills as an electronic engineer to create Landmark Graphics Corp., a $40 million company that makes devices used in the seismic mapping of oil reserves. The technique is used to squeeze the last drop from known oil deposits by creating digital maps using sound waves.
Landmark had gotten too big to be any fun, Hildebrand says, so he quit. He enrolled at the Shepard School of Music at Rice University in Houston.
"I was thinking I was retiring. But it was like a bad inoculation: It didn't take," Hildebrand explains. "I was going to write film scores, but I never made it."
As part of his music-school dabbling, though, Hildebrand developed some algorithms designed to improve the sound of synthesizers. "A friend of mine told me I could make money off that, so I made a program called Infinity, which allowed digital samples to be looped. Repeating complex sounds is very problematic, so I developed algorithms that would do that. The program is now used a lot in synthesizers that create movie sound effects."
Hildebrand was back in business. He quit music school. He hired a distributor. That, too, proved to be a more fortunate move than he had imagined.
"I was with my distributor and his wife and a friend, and his wife said, 'If you could make me anything in the world, why don't you make me a box that would make me sing in tune?' " Hildebrand recalls. "Everybody looked at the floor, so I dropped it."
But the problem began to bother him. In the oil patch, Hildebrand had designed algorithms that fine-tuned the sound waves bouncing off underground rocks to paint a crystal-clear picture of subterranean petroleum deposits. How different would it be to fine-tune the sound waves coming out of somebody's throat?
Troubled, Hildebrand went to bed.
"I went to bed for a month. I do my best engineering in bed," Hildebrand explains. "These are complex things going on technically, and it takes a lot of careful thinking. I do my best thinking with a minimum of distraction. I don't sit near a telephone and don't sit near a computer screen. The best place for that is in bed."
When he finally arose, he had devised a way to alter vocal pitch without changing the flavor of a person's voice, long a studio engineer's holy grail. It was already common practice for engineers to snip tiny portions of out-of-tune vocal track, change the speed, and rerecord at the resulting changed pitch. But, besides being cumbersome, speeding up bits of tape to achieve higher pitch makes people sound like chipmunks; slowing those bits down can sound worse.
With AutoTune, you just sing through a box and come out on key.
Hildebrand has since moved to Applegate, two hours north of San Francisco. He plans later this year to buy out the Los Gatos company that distributes his software. Hildebrand has a hit on his hands, audio engineers say.
"Now, I can't speak for other towns, but I live in Nashville, and let's face it, a lot of country guys can't sing," says Jeff Mac, a recording engineer, explaining the prominent role AutoTune has taken in his hometown's recording studios. "Our buying public expects a whole lot more than people did in the '70s and '80s. People want perfection now."
If programmer Chris Weare gets his way, singers will have access to a sort of perfection that goes beyond correct pitch. They may be able to don the voice of any singer they wish.
"It's like chopping off your head and putting someone else's on it," says Weare, describing his Harmony voice-altering software package.
At this point, the software is designed to be used for creating a chorus effect by playing multiple, altered versions of a single person's voice, says Weare. To demonstrate, he plays a recorded voice singing the words "Sweet Adeline." He then plays a second voice, singing a single, uninflected note.
By analyzing the fleche-shaped spikes and dips of the second voice, then adjusting the original voice according to these characteristics, Weare creates a third. A little tweaking, and he has a fourth. Together the altered voices make a barbershop quartet. The effect might prove useful for musicians who can't afford backup singers, Weare explains.
Weare came up with early versions of this software while working on the movie Meet Joe Black. He had been assigned to morph the voice of Anthony Hopkins into that of Brad Pitt. Weare didn't finish the project in time to make the final cut. But the voice-combining techniques he developed seemed worth salvaging -- especially the possibility of creating a third voice from two disparate ones.
The idea behind the program certainly fires the imagination: Greg Tomassian as Ray Charles? Soul Candy as Queen?
Exciting though the prospect sounds, true head-switching is still a ways off, Weare admits. Now he's limited to slowly, subtly altering voices to create musical sound effects.
Even with this painstaking process, the third voice he has achieved so far is only a minor improvement over those mechanical answering-machine voices that plague corporate telephone switchboards.
A solution might seem far in the future, but for the fact that Weare has some backup. The Illinois company Symbolic Sound is also pursuing the ephemeral third voice. It has released a program that cross-fades sound frequencies in a way similar to the effect Weare attempted to create for Meet Joe Black. Symbolic Sound's equipment is already used to produce an audio-morphing effect employed in some movies, and in radio and television commercials. Kurt Hebel, a partner in the company, demonstrates a British radio spot in which a soccer player's voice morphs into that of a nagging schoolmarm. The fully produced transition is convincing. But when Hebel singles out the middle, morphed voice, it sounds unconvincingly thin, much like Weare's effort.
Hebel admits this. He admits also that he, too, is in the race for the "third throat."
"It's a very, very difficult problem," he says. With current technology, "we're still working on trying to make the in-between sound like a real voice."
By the time that happens, the broadening palette of digital-audio looping interfaces, rhythmic correction software and tone correction programs may have moved closer to the mainstream. By then, Soul Candy may be playing the Fillmore. The San Francisco Symphony may premiere Cope's "Symphony No. 1383."
Will music be dead?
Matt Donner, a salesman for Cutting Edge Audio, a SOMA distributor of AutoTune and other such devices, doesn't think so. "There's really no substitute for getting a slamming performer and hitting it right the first time." Or as the S.F. Conservatory of Music's Greenberg suggests: "It's still shit in, shit out."
Don't tell that to Tomassian. He and Soul Candy are hitting the road.
"I tell you, I feel pretty good, as a vocalist, taking my wares to wherever I want to go," he says. "Even if this band doesn't work out, with this little bit of profes-sionalism, I might be able to take it to the next level.