The Audio Frontier
Leading computer audio engineers retreat to a Texas ranch to map out the future of their technology... and to admire the emus
Don't poison my well!" warns the cowboy, "or there's going to be hell to pay." It's late one night on a ranch in the central Texas hill country. The cows are heading off to pasture, storm clouds are moving in, and smoke rises from the campfire. As the cowboy admonishes the men gathered before him, the showdown is about to begin.
But this is no ordinary shootout. And these are no ordinary cowboys. They're some 50 software and hardware engineers who work in the Wild West of computer audio. The weekend cowboy addressing the group is Keith Weiner, the handlebar-mustached CEO of DiamondWare Ltd. (Mesa, Ariz.), a company that produces audio software. As Weiner puts it metaphorically, they're here to spend the next three days cleaning up their poisoned well: the faulty and undependable pit of audio wares. These are the strings of bits that translate into bongs, shrieks, beeps, honks, and even the occasional musical interlude when you boot up your computer, blast a monster in a computer game, or crank up Radiohead. If success crowns their efforts, they'll not only improve computer audio, but also make consumers care more about computer audio in the first place.
The occasion is the seventh annual Project Bar-B-Q, an offbeat and influential three-day retreat for leaders in the field of audio engineering. Every October, engineers from computer software and hardware companies such as Microsoft, Texas Instruments, and Dolby Laboratories gather at the Guadalupe River Ranch near Boerne, Texas. For them, this long weekend is an opportunity for competitors to work together on divining the challenges that face them collectively.
"As engineers, reinventing the wheel isn't the highest of our priorities," says Rob Rampley, lead software engineer for Line 6 Inc. (Agoura Hills, Calif.), a leading developer of digital music composition software. Rampley, with two-toned hair and rock star jewelry, is one of many here who look like musicians. "This isn't about solving all the problems," he adds. "It's about finding the problems that need to be solved. This is our chance to collaborate as an industry."
"We have a consistent presence at Bar-B-Q," says Devon Worrell, senior engineering manager and architect for chipset and software engineering at Intel Corp. (San Jose, Calif.). "It's a good forum for working on the longer-term directions of where our industry is headed." Worrell has dressed for the occasion: a plaid shirt, blue jeans, and a green bandanna tied around his neck.
In the past, the Bar-B-Q has spawned such special interest groups as the Interactive Audio Special Interest Group, a coalition of hardware and software engineers who work to improve interactive audio, and the Game Audio Network Guild (San Juan Capistrano, Calif.), which works to advance video game audio. The project has also resulted in the establishment of standards for audio formats like the Musical Instrument Digital Interface (MIDI), which allows electronic instruments and devices to interact with each other, and Microsoft's DirectMusic, a composition tool for interactive music.
The goal is to improve hardware and software for music on computers over a period of five years. Just how the Bar-B-Q participants go about doing that provides insights into the world of audio engineering. It also holds some lessons about how engineers in any discipline can not only think outside the box, but leave the old carton in flames.
Singin' on the range
"I can't take a breath/I can't make a move/I can't think a thought/I can't sing a song with a minimal groove."
The world's most prolific computer game musician—he's written the music for hundreds of games and is known as The Fat Man—sits near a tie-dyed amp, plucking a double neck guitar as he sings these words to an original tune. He's dressed like a modern day shaman: a jet fighter's flight suit strung with a necklace of bear teeth, a couple of green glow sticks poking out of his chest pockets ("to light the way," he explains), and a pair of red scissors dangling from his shirt ("to cut through the BS").
The number that The Fat Man—or as he's known outside the computer audio community, George Sanger—is playing is indeed minimal, just a few stark notes that alternate behind voice. It's a song called "Dr. Death," an instrumental track that he wrote for a ghoulish computer game, The 7th Guest, Part 2—The 11th Hour. Sanger's many tunes have also been used in the background of such other classic games as Wing Commander. The Fat Man is playing The 7th Guest song for me to illustrate a point: behind every computer game is a songwriter; inside every computer game is a song. It's the audio technology that brings that game and its music all to life. The problem is, he says, that computer audio wares haven't been living up to their end of the bargain.
Despite the success of MP3 and the hype of 5.1 surround sound—the Dolby standard that utilizes a six-speaker system—there's one inescapable fact about computer music, Sanger says. The quality of computer audio tends to be poor, although many listeners may not realize it. But the engineers who come to Project Bar-B-Q feel the pain—poor speakers, audio downloads that crash the system, and game soundtracks whose lame audio quality makes them want to strangle themselves with their mouse cords. This is the metaphorical poisoned well they spoke of earlier, and they admit that the poison is in part due to the lack of communication among the audio system designers—like these cowboys at the ranch.
To help these friendly rivals interact freely, Sang insists that for their stay at the ranch, Bar-B-Q attendees drop their corporate personas and, as he puts it, "turn back into human beings." Shirts with company logos are banned outright. To boost the down-home spirit, attendees adopt cowboy names like Tex and Marlboro and spend the evenings in a live music jam session in the ranch's lounge. Without this human connection, Sanger says, computer audio, the soul in the machine, is dead. As he concludes the song, the refrain bears out the lyrics: "I can't get away from the singular fact/That I can't take a breath/Because I'm seeing Dr. Death."
It's early the next morning. Mist rolls over the hills. The smell of sausage and eggs wafts up from the kitchen. Around the ranch, the creatures stir: the cows, the towering emus— ostrich-like birds also raised on the ranch—and the ranch's other unlikely residents, the engineers. The Bar-B-Q's "work hard, play hard" ethos has left the conference's attendees with throbbing heads and ringing ears after the all-night jam session in the lodge.
Inside a conference room, Sanger begins the first morning's powwow. "Okay, everyone," he booms, "start clumping!" Behind him is a whiteboard scrawled with a long list of Big Think topics for the attendees to clump together into discussion topics for the weekend. The subjects range from the interoperability of digital music-playing devices to principles of interactive sound and music design.
Scattered around the room are a few dozen Bar-B-Q'ers who resemble a mix of guitar shop slackers and Silicon Valley commuters. They represent all parts of the digital music spectrum, the makers of the music that plays on the machines and the makers of the machines that play the music.
To many outside the industry, these are the unsung heroes of the digital age. After all, while computer audio is everywhere, it's seemingly nowhere, audible, but below the radar. When did you last attend to the sound of a dial tone? Or the blast of a rocket fire in an Xbox game like Halo? It's as if these sounds just exist. But, in fact, legions of people, just like those at the Bar-B-Q, dreamed them up, designed them, coded them, and engineered the hardware that plays them. It's computer users' widespread lack of what the Bar-B-Q veterans call audio awareness that has led to the incessantly second-rate audio experience that most computer users endure.
For a case history of computer audio, the attendees cite computer game music, the runt technology of the US $10.8 billion game business. Back in the 1980s, the music was the domain of a programming language known as FM synth, which produced blips that sounded like synthetic backwash. But once programs moved from the tiny confines of the 3.5-inch floppy to the far more capacious CD-ROM, computer game music could finally sound musical.
The music needed help from hardware, though. Thus, when Sanger was commissioned in the early 1990s to work on the first game CD-ROM, The 7th Guest, he and other computer audio engineers seized the opportunity to dump FM synth for the more orchestrally general MIDI, the venerable hardware and software standard that produces sounds resembling actual musical instruments.
Since then, despite the leaps to digital audio, newfangled tricks like 5.1 surround sound, and Microsoft's DirectMusic (which lets game music be cued, as in film), computer audio specialists say it has been a fight for respect and, more crucially, disk space. Graphics get far more room on CD-ROMs than sound; computer audio has been an afterthought. Software publishers put minimal effort into computer audio. "Game music is a huge success if it doesn't drive you nuts," Sanger says ruefully.
Once the brainstorming session heats up this morning at the Bar-B-Q, some key technical issues emerge. One is digital rights management, the watermarking technology that allows content creators to manage music files distributed over the Internet while ensuring that the creators are compensated for their work. Another issue is how to make a Dolby 5.1 surround sound system wireless. There's also the question of how, in general, the burgeoning legions of digital audio devices—MP3 players, personal computers, video game consoles, and television sets—will converge.
The convergence is happening in two ways. First, the devices themselves are becoming hybrids—cellphones that play MP3 audio, MP3 players that act like PDAs. Second, there's a coming together of lifestyles: consumers listen to music as much at their desks and on the treadmill at the gym as in front of their stereos. Listeners are expecting their music anytime, any place. This demand, says Ron Kuper, vice president of engineering for Cakewalk (Boston), a developer of digital music composition tools used by musicians, means that "we don't know what kind of technology is going to come out of [this convergence]."
By lunchtime, having determined the weekend's discussion topics, the attendees have split up into four groups. As part of the Bar-B-Q's imperative to produce specific action items that participants can pursue when they leave the ranch, each group must return to this room in two days to present just what their new technologies are going to be.
Riders in the storm
Storm clouds unleash a downpour on the ranch. Inside, the Plumbers huddle together hard at work. The nickname denotes one of the Bar-B-Q think-tank groups. It consists of Intel's Devon Worrell, Cakewalk's Kuper, and eight other engineers who have decided to tackle the technical challenges down in the infrastructure, or plumbing, of their wares.
And just what is the main challenge? Look no farther than the room next door, where another group sits around a large, oak table strewn with greasy onion rings and an arsenal of toys: Power Prop Flying Gliders, rubber band guns, and a flatulent, noise-making goo called Flarp.
But what really matters are the instruments left over from last night's jam session: guitars, amps, banjos, synthesizers, and an intestinal snarl of cables. While it's easy enough for these guys to plug in and jam with real-life musical instruments, such simple connectivity eludes the millions of creators of digital music. The challenge, the Plumbers conclude, is to devise similar ways for the myriad digital wares to connect.
At the brainstorming session the day before, David Zicarelli, CEO of Cycling '74 (San Francisco), an audio composition software development company, expanded on how the industry has evolved from the "age of recording to the age of composition." Songs are no longer static. Once digitized onto the desktop, they can be easily modified and sampled, then swapped over peer-to-peer networks. In Zicarelli's view, whereas once music fans could do little but slap a piece of vinyl on a turntable and listen passively, "People are starting to think in terms of programming their own worlds."
The Plumbers agree that while audio composition is migrating to the computer, there's no way to collaborate as easily on a computer as one might in the jam room next door. The software add-ons that bring supplementary features to existing programs are just too varied. So someone who uses music composition software—sometimes called an authoring tool—to compose a track might lose an effect such as reverberation when he goes to a professional studio using another of the many authoring tools.
"There's no interoperability between different authoring tools," Kuper says to the group. He proposes creating a standard that would make such interoperability possible. Worrell, however, thinks standards aren't worth so much of the group's attention. "We're on a quest to create something simple that can be used across platforms," says Worrell, "It reduces work, but the person playing a computer game or listening to a CD, doesn't see anything new." The Plumbers decide to take a break, heading outside to stroll among the emus.
In another conference area, the self-anointed Process and Principles group is wrestling with a matter that every engineer struggles with: getting the respect—and financing—from superiors that's needed to adequately pursue the many challenges at hand. "We're making decisions with every line of code," says Rampley of Line 6.
As the discussion flows, they identify the problem: they cannot define for their superiors what "quality audio" really means. These engineers feel like they're often fighting for something that their bosses don't truly understand. If there were a detailed definition—a document—articulating the features of good audio, maybe, for once, they'd all be on the same page.
They decide to label the document, "Principles of Good Audio." There's a real sense that the Bar-B-Q has given them the soapbox they need, the opportunity to come together and write their mission statement: the commandments that have eluded the industry. Then, hesitantly, one of the guys raises his hand to say what's on everyone's mind. "I'm concerned that we're going to come up with a great document, and audio people will think it's cool, but [our bosses] won't care." Everyone nods in understanding. One of the charms of Bar-B-Q is commiseration.
The last act
By the time the final presentations are due, the weekend cowboys are pretty well spent. Between the brainstorming sessions all day, and the musical jam sessions all night, the pace has been unrelenting. The lodge is cluttered with guitar picks and toy robots. Beer bottles line the hot tub deck. And the whiteboards are smeared with layers of magic marker screeds. But the "hard work, hard play" rallying cry has certainly achieved its goal: stripping these engineers of the proprietary feelings that accompany most conferences and getting them to collaborate as a community of like-minded technologists.
The troops proceed to the presentation hall, which is filled with multicolored balloons and, for tension relief, several boxes of Lone Wolf Bang Snaps, little wads of paper that explode on impact. As Ron Kuper, Devon Worrell, and the rest of the Plumbers step up to the microphone, it transpires that the change they had concentrated on making is not quite what they had expected.
In the course of their discussions this weekend, it became clear that Microsoft Corp. (Redmond, Wash.), in fact, was in the process of developing what it called a Universal Audio Architecture, which would enable better interoperability between audio wares and would be released with its next operating system. Given this situation, the Plumbers had prepared a report that addressed, as Kuper told the rest of the attendees, "the synchronization and latency management issues that arise in the converged media environment." That mouthful means engineering a solution that, for example, shortens the interval between the moment a gamer pulls a trigger on a Playstation 2 controller and the instant the gun blast comes out of the speakers.
As for the Process and Principles group, their document on what they termed "Audio Values" met with great appreciation. Besides defining just what constitutes quality interactive sound and how much fidelity should be sacrificed in the name of bandwidth, the group proposed an experiment to determine how consumers perceive differences in three sample rate formats: from the high end of 44.1 kHz, which is CD-quality, to the lower range 22 kHz, or 11 kHz. They want to prove that even the least discerning listener would hear the difference.
After the applause, Rampley reflects on the value of the Bar-B-Q. "As engineers," he says, "we tend to get caught up in the details of our work. But it's a good exercise to get together as a group and identify our problems. It really allows us to just slow down and look at the big picture." And keep their well from getting poisoned at all.
To Probe Further
For more information on Project Bar-B-Q, see http://www.projectbarbq.com/. This site includes reports from conferences dating back to 1996.
The Interactive Audio Special Interest Group, a group founded through the Bar-B-Q project, is a collective of software and hardware engineers who work to improve interactive audio applications. See http://www.iasig.org.
An IEEE standards proposal, Project Authorization Request 1599, is exploring the possibility of a standard that would use XML to better integrate data present in various layers of musical recordings. See http://grouper.ieee.org/board/nescom/1599.pdf [PDF].
The homepage for George ("The Fat Man") Sanger, founder of Project Bar-B-Q and a legend in computer game audio development, is at http://www.fatman.com.