So: Do We Have a Network Yet?
By Chris Doerring
Dynamic Market Systems

Do we have a networking standard yet? The short answer, once again, is “no.” As in year past, this year’s AES generated rumors of progress along with signs that convergence on an industry standard for transfer and routing of digital audio and device control data is going to be a long and winding road.

Seems kind of silly, doesn’t it? If we had such a network, digital consoles, signal-and-loudspeaker management “system brains,” and amplifier/loudspeaker systems (whether joined umbilically by speaker cables and sense lines or integrated into the speaker enclosure) would instantly become more valuable. If these key sound system elements could talk to each other and to system operation and maintenance techs, pro audio systems would deliver a higher quality of service to its various client universes, and could therefore become more profitable.

Manufacturers would see a huge increase in demand for new products, because the value of a networked digital mixer, for instance, is so much higher than that of a standalone digital console. Ignore the fact that multiple sets of A/D and D/A converters impose unnecessary costs and performance penalties on the system.

Just imagine the benefits of a console that presented objective, quantified information that the operator could compare with what he or she is hearing. Similar benefits in terms of shifting the effort from grunt work and basic problem solving to intelligent and creative use of the system on behalf of the performers and the audience are waiting at every link in the signal chain.

So let’s see: if we had a network standard, everyone would sell tons of gear and there would be lots more work for installers, rental houses, designers... the entire industry would benefit. To get an idea of how important standards are, try to imagine the modern office without Ethernet: how many desktop and laptop computers, printers, servers, etc. do you think you’d see?

If a whole industry can grow up around a networking standard, why can’t pro audio agree on one? Apparently, most audio manufacturers are owned by Bill Gates wannabes. In other words, the majority of business people who have funded R&D projects in the network space have brought them to market as a feature set across a product range (IQ, QSControl, NexSys, for instance) instead of as candidates for industry-wide acceptance. Third party efforts, whether privately funded (Lone Wolf) or association driven (AES 24) have failed.

The legislative route to a standard seems like a dead end at this point. Pro audio’s best hope lies in the growth of a de facto standard. One network needs to acquire enough existing nodes and perceived possible nodes to reach critical mass [for a longer explanation of critical mass and the underlying concepts, click here].

Candidates for the role of digital audio network standard race towards critical mass on two legs: the technical debate as to which is “better,” and the market debate as to which is “bigger.” Better is always a hot topic at AES. After all, it’s the largest muddle of audio engineers assembled on the planet each year. As a marketing guy, I am fortunate to be disqualified from participating in the “better” disputes. In my experience, bigger is better anyway: I liked BetaMax and MiniDisc just fine, thanks, but the network economics were too powerful. I watch VHS cassettes and use portable CD players just like every other red white and blue blooded American consumer.

So here’s the state of play in the race towards critical mass, the process that has the best chance of producing a de facto industry standard that can benefit all players. I declare CobraNet the front runner: it’s got strong backing from Cirrus Logic, 18 licensees including all the major amp names, most of the “digital system brain” players and some speaker manufacturers. CobraNet looks very strong, but as long as it’s below critical mass, there’s room for other contenders to exploit possible weaknesses. Number one is per-node cost. It may not be a problem for console or signal processor manufacturers: the CobraNet port adds a significant cost to the unit, but not to the entire system, especially when compared to the value added to all nodes (aka products) by the network. Loudspeakers are a different matter, however. In a typical system design there might be one or two consoles, one or two “brains” per room, but twenty to five hundred speakers. With that many loudspeaker nodes, you need a cheaper node to get an acceptable cost/benefit ratio.

Digigram brought their solution to the show. It’s called EtherSpeaker, a one-way 32 channel speaker wire replacement that uses the same “physical layer” (Cat 5 twisted pair copper and RJ45 connectors) as CobraNet (and SoundWeb and Ethernet, for that matter). Per-node cost is significantly lower, because EtherSpeaker just does basic signal routing chores: no control, no bidirectional communication, just multi-channel digital audio.

BSS SoundWeb has about the same number of installed nodes as CobraNet, but it’s a proprietary closed system. Although BSS is the latest CobraNet licensee, the company representative I spoke with claimed that SoundWeb is too far ahead of CobraNet in some key areas to allow a bridge between the two most widely used digital audio networks (which would create critical mass by joining two islands to form a continent). I’d have to guess that BSS wants CobraNet ports on next-generation OmniDrive product, at least as an option, and acquired the license as insurance against CobraNet’s reaching critical mass and becoming the de facto standard. At this point, SoundWeb is out of the network standard race, whatever its technical merits may be: BSS has no licensing programfor the technology.

Yamaha’s mLAN is slow out of the gate, but could accelerate quickly. mLAN is an “audio layer” on IEEE1394, aka FireWire. FireWire is only 6 years old, was invented by Apple Computer specifically for multimedia, and won a Technical Emmy last year. It’s already revolutionized the video production industry by hot-plugging digital video cameras into desktop computers running video editing software. To date, only four companies have announced mLAN licenses: Yamaha, Otari, Korg and Swissonics. To those who cite FireWire’s distance limitations and clock jitter problems, I say “go jump in the Pacific.” Try leaping off the pier at Redondo Beach, CA, where the performing arts center is shipping 88 audio channels around the building on an mLAN network. mLAN is available as a royalty-free license, and a 32-channel 400 MB/s chip is due next year from Otari and Yamaha. Four of those could be connected to a single FireWire “link layer” chip, for up to 128 channels of audio, plus MIDI, video, computer control protocols like RS232, etc.

Close behind mLAN is Gibson’s MaGIC (Media Accelerated Global Information Carrier). MaGIC is a high-bandwidth, low-latency protocol that exists on the IEEE802.3b physical layer (Cat 5 cable and RJ45 connectors) just like SoundWeb, CobraNet (or at least QSC’s CobraNet-based RAVE network hardware) and EtherSpeaker. Like FireWire, MaGIC is ready to handle a range of digital media plus control signals (although MIDI is currently the only defined control protocol). The network does not have a built-in IP layer, but does utilize the hardware-specific MAC address protocol, and reserves space for TCP/IP headers if desired. MaGIC plans to exhibit at CES and NSCA, but will not be sharing booth space with Gibson Les Pauls, L-5s and J-200s at NAMM. Like mLAN, MaGIC is a royalty-free license.

Gibson plans to have a $20 chip available sometime in 2002.

At the moment, none of the visible network standards are close to critical mass. CobraNet’s early lead is substantial, but could evaporate quickly under the right circumstances. Meanwhile, engineers continue to debate the merits of one standard or another, their employers continue to dream of becoming digital audio Macrosofts, and the industry puts along without the rocket boost a digital audio networking standard could provide.