A new interfacing standard is just around the corner which can carry digital audio, MIDI, timecode and hard drive communications down a single cable, in addition to the digital video and multimedia signals which the likes of Sony, Apple and others are already sending down it. Paul Wiffen, long an advocate of high-speed digital interfaces, investigates and finds that he may just have a new religion to evangelise. This is the first article in a four-part series. Read Part 2, Part 3 and Part 4.
Over 10 years ago, I experienced one of these full-on road-to-Damascus-style conversions when I saw Marcus Ryle of Alesis plug a single fibre-optic cable resembling a strand of angelhair pasta between two of the then brand-new ADAT tape machines and transfer eight channels of audio data between them in a single pass. Leaving that demonstration, everyone agreed that we had just seen the future -- but everyone else seemed to have seen a different future to me. While everyone else raved about the sound quality of these machines (I wonder how they feel about those statements now?), I dismissed the ADAT itself as "just one more tape machine to chew up your work the night before you mix it down," and instead started to babble incoherently that one day all digital transfers would be made that way. Ever since, I have been crying 'ADAT Optical!' like John the Baptist in the wilderness, hoping that someone could see past SVHS video recorders doing multitrack audio to the greater truth that lay hidden behind. I have raved to others about it till their eyes glazed over, I have taken hideous jobs just so I could work with it, and berated manufacturers to put it on their equipment. And, like my predecessor John, I have had some success in spreading the word. Today many people use ADAT lightpipe as their method for moving eight, 16 or even 24 channels of digital audio between their computer and their mixer.
However, I stand here today to tell you that there comes one after ADAT Optical, whose shoes it is not fit to clean (to paraphrase John, Moses and Martin Luther King). It shall take you places that ADAT Optical never could, to the Promised Land where MIDI and Audio shall flow entwined in a single strand. I may not get there with you but I tell that I have looked over Jordan and have seen the... Sorry, got a bit carried away there.
The Potential Of FireWire
This more recent conversion took place in 1998, not on the road to Damascus but in Cork in Eire, at the Apple plant there. I had been invited, as a then-employee of Korg, by Neil Murphy, who was in charge of testing third-party music hardware and software, so that non-Apple developers could be informed that changes to their software and hardware would be necessary when innovations in the Apple Mac OS, CPUs or motherboards were in the offing. Frequently, Neil had to do this even when the Apple innovations themselves couldn't be revealed to anyone, let alone shown or provided!
On that first visit, spotting that I happened to have a newly purchased Sony D100 digital camcorder with me, Neil asked if he could test it with a new interface on an Apple computer that I wasn't even supposed to see, claiming that there was now a way of getting the digital camcorder to interface with the computer directly in the digital domain -- FireWire. Previously, I had been using a Miro Motion DC20 card to digitise video for my QuickTime demos for Korg, and I was pretty impressed to discover that an Apple computer could take video in directly in DV digital format from a socket already supplied on my camera. I had seen the sticker marked 'IEEE 1394 FireWire' on the camera when I had bought it, but just dismissed it as marketing hype. Conversationally, I remarked to Neil that it was a shame that the same was not possible with digital audio.
"Oh, sure it is," he replied casually, "and through the same port. Someone just needs to write the protocol and put the connector on mixers and synths and things!" I was dumbfounded. How could the same interface be used for digital audio and video? He explained that it ran so fast (between 100 and 400Mbps) that it could do several streams of the more demanding digital video (with associated 44.1K stereo audio) or hundreds of channels of digital audio on its own. With a bit of care, it could probably do digital video and audio at the same time. "Incredible," I cried, "so all you would need is a MIDI interface on the computer and you would have all the multimedia bases covered!"
"Oh, it'll do MIDI as well without even working up a sweat! Again, all that's needed is the protocol and the connectors on the musical instruments."
I looked up at the overcast Irish skies, the clouds parted, and like my biblical namesake in a sentimental Victorian painting, I caught a glimpse of Paradise! ADAT Optical had reduced the spaghetti in my studio to angelhair pasta -- but FireWire could eradicate it forever, and the constant replugging of cables, miswiring and crawling on the floor that went with it. I dedicated my life on the spot to do whatever it took to get an audio protocol for FireWire out there and adopted.
Apparently, Apple had developed this interface under the trademark of FireWire earlier in the decade, but had placed it in the public domain to encourage takeup by third-party manufacturers. There, it had been catchily rechristened IEEE 1394 and snapped up by the likes of Sony (who dubbed it iLink) and JVC as the perfect way to interface DV cameras to computers and video components. Since the codecs in the cameras had already compressed the DV data by roughly 5:1, IEEE 1394 (let's call it FireWire from now on, I have always preferred names to numbers) could transmit this much information plus stereo linear 44.1kHz audio without even breathing hard. But the lion's share of the available bandwidth on FireWire (which supports data transmission rates switchable between 100, 200 and 400Mbps) was still untapped. The 1394 TA (Trade Association), formed mainly by Apple and video companies like Sony and JVC, had quickly established the DV protocol, and started shipping products that included the connector and firmware support.
In January of 1999, Steve Jobs drew a standing ovation at his MacWorld keynote speech by hot-plugging a DV camera into a Mac (ie. plugging it in with the power on) and having it show up on the desktop and video pictures appear on the Mac screen within seconds, without rebooting. The first G3 with FireWire shipped a few weeks later and the rest is history. Apple's conquering of the DV editing market with the iMac DVSE and its included iMovie editing software led inexorably from this.
Very shortly afterwards, the first FireWire drives began to appear and immediately showed data transfer rate advantages over SCSI and/or IDE, even though many of them had mechanisms which were designed with those older interfaces in mind. Indeed some even have a sort of FireWire-to-IDE interface inside them which doesn't support the transfer rates that are theoretically possible with FireWire. Nevertheless, they still outperform SCSI and IDE several times over, and this can only improve drastically as purpose-built FireWire drive mechanisms come onto the market. Other types of drive mechanism are also now becoming available with FireWire interfacing, including CD-R and even DVD-R, and this should prove a huge growth area with mass-market potential, in particular for applications like real-time video streaming between devices and real-time backup to mass storage. A large potential market always drives a new technology faster than if only a select few can get behind it.
However, multi-channel audio or multi-buss MIDI is by comparison a much smaller market. Few companies in the 1394 TA were that bothered about multi-channel audio or MIDI, so they resolved to pass the responsibility for these protocols to more interested bodies. It therefore fell to the AES (Audio Engineering Society) to oversee the creation of a multi-channel audio and timecode transmission protocol for FireWire, and to the MMA (MIDI Manufacturers Association) to do the same for MIDI.
Of course, the other issue which is relevant is what the takeup will be on mLAN interfacing amongst Yamaha's rival manufacturers of mixers, synths and effects. Clearly, smaller companies can only benefit from the increased compatibility which mLAN/FireWire will give them. I know of one product which will be mLAN-compatible by the September AES in LA, as the head of Swissonic's design team is currently in Japan with Yamaha's principal engineer to fast-track the development of a FireWire option for their eight-channel 24/96 A-D converter/mic preamp, the AD8. Sources at Apple tell me that Swissonic are not the only ones making peripherals mLAN-compatible, although they are too discreet to mention any names. The big question, of course, is will the Rolands, Korgs, Akais and Emus follow the mLAN path? Of course, Yamaha probably make the broadest range of music and audio equipment, so you could equip an entire mLAN studio with their products, but do you really want to be forced to use a Yamaha mixer, Yamaha synths, a Yamaha sampler, Yamaha effects and Yamaha I/O devices? My guess is that with Apple so firmly in the mLAN camp and the PC community easily adaptable to follow suit, it won't be long before we start to see mLAN as an option at least for other major manufacturer's keyboards, mixers and effects. Korg took the ADAT interface to their heart for a while (although sadly they never quite got around to supporting all eight channels on their keyboards). Mackie and Soundcraft offer various differing third-party interface support for their digital mixers, as do newcomers Roland and Fostex. Perhaps it is on synths, where 5-pin DIN MIDI sockets have been a de facto standard for so long, that the greatest danger of non-proliferation of mLAN lies. It may be some time before manufacturers of synths and samplers recognise the advantage in being able have a single direct connection to a computer, which can carry the MIDI for triggering and librarian/editor packages, the audio for sample input and sample editing, and the multiple channel outputs for mixing. For too many years, hardware manufacturers have claimed that the cost of separate analogue outputs with their associated A-D and D-A converters means that they can only be supplied as options or not available at all. Separate outputs in digital via FireWire will end up saving a fortune. In the meantime, there is always the Yamaha mLAN8P (and its descendants) to bring any stragglers into the fold!
mLAN On Other Platforms & Products
Of course, the other issue which is relevant is what the takeup will be on mLAN interfacing amongst Yamaha's rival manufacturers of mixers, synths and effects. Clearly, smaller companies can only benefit from the increased compatibility which mLAN/FireWire will give them. I know of one product which will be mLAN-compatible by the September AES in LA, as the head of Swissonic's design team is currently in Japan with Yamaha's principal engineer to fast-track the development of a FireWire option for their eight-channel 24/96 A-D converter/mic preamp, the AD8. Sources at Apple tell me that Swissonic are not the only ones making peripherals mLAN-compatible, although they are too discreet to mention any names.
The big question, of course, is will the Rolands, Korgs, Akais and Emus follow the mLAN path? Of course, Yamaha probably make the broadest range of music and audio equipment, so you could equip an entire mLAN studio with their products, but do you really want to be forced to use a Yamaha mixer, Yamaha synths, a Yamaha sampler, Yamaha effects and Yamaha I/O devices? My guess is that with Apple so firmly in the mLAN camp and the PC community easily adaptable to follow suit, it won't be long before we start to see mLAN as an option at least for other major manufacturer's keyboards, mixers and effects. Korg took the ADAT interface to their heart for a while (although sadly they never quite got around to supporting all eight channels on their keyboards). Mackie and Soundcraft offer various differing third-party interface support for their digital mixers, as do newcomers Roland and Fostex. Perhaps it is on synths, where 5-pin DIN MIDI sockets have been a de facto standard for so long, that the greatest danger of non-proliferation of mLAN lies. It may be some time before manufacturers of synths and samplers recognise the advantage in being able have a single direct connection to a computer, which can carry the MIDI for triggering and librarian/editor packages, the audio for sample input and sample editing, and the multiple channel outputs for mixing. For too many years, hardware manufacturers have claimed that the cost of separate analogue outputs with their associated A-D and D-A converters means that they can only be supplied as options or not available at all. Separate outputs in digital via FireWire will end up saving a fortune. In the meantime, there is always the Yamaha mLAN8P (and its descendants) to bring any stragglers into the fold!
Cut to the AES trade show in Amsterdam later in 1999. My determination to have some effect on the adoption of FireWire led me to fly in a day before the show opened for the first meeting of the newly-formed IEEE 1394 Standards Committee. It was this committee which had originally been set up to define the multi-channel and timecode transmission protocols for FireWire. I joined with a great sense of destiny being fulfilled, and discovered that things were already underway! Yamaha had already submitted a so-called 'Audio and Music Data Transmission Protocol Version 1' to the 1394 TA committee and had it approved, and proposals on timecode transmission and other enhancements (such as the transmission of copyright data along with audio) followed, all ratified by the AES Standards Committee.
Sadly, it proved harder for me to get to see how the MIDI side of the protocol was developing (if at all), as the MMA was more difficult to penetrate. Like the 1394 TA, you had to be the employee of a member organisation to participate in the process. I was no longer working for Korg by this time, so I made overtures to James Grunke, the MMA President, who I had worked with in his Atari days on the Falcon launch, and he began to look into the idea of my acting as an observer/liaison with the AES standards committee.
However, events overtook us. At the MacWorld show this January, and again at the NAMM trade show in February and the Frankfurt Musikmesse in April, Yamaha not only went public with their mLAN protocol, but also demonstrated it running with 16 channels of audio and MIDI being sent out of a G3's built-in FireWire socket to an O2R mixer and an S80 synth using ASIO and OMS drivers respectively. In addition there was a dedicated unit, the mLAN8P, designed to act as a breakout box for eight channels of audio I/O and two of MIDI, complete with a 12-channel, 4-buss digital audio mixer, built-in DSP effects and software mixing/patchbay applications for the Mac. Having done a lot of the work necessary to send both digital audio and MIDI over FireWire, Yamaha were opening up the protocol they had come up with as a general standard for transmission over FireWire, instead of keeping it as a proprietary system (with all the potential interfacing problems that might cause for owners of equipment from other manufacturers). This, in my opinion, showed the same kind of foresight Apple had exhibited when they put the 1394 spec into the public domain in the first place.
In actual fact, mLAN had been talked about in the aisles of trade shows for a while. A special web site had been set up which talked in general terms about a chip set Yamaha had been developing for about a year, and there had also been discussion of the types of data which the putative protocol would be capable of transmitting, and the transmission speed. Essentially, what Yamaha had proposed was a musical version of a Local Area Network (hence mLAN). If you're not familiar with LANs, they're simply small networks in which computers and peripherals are linked together to comprise a single system, with resources shared between multiple users. What had become clear in January of this year was that Yamaha planned to use FireWire hardware connectors as the means for mLAN to communicate between computers and other musical devices such as keyboards, samplers and effects units. They had produced a YGDAI option card (CD8-mLAN) to allow mLAN to interface with their O-series digital mixers and an option board (mLAN8E) for interfacing their MIDI synths, samplers and other modules to mLAN. The mLAN8P acted as mLAN breakout box, allowing audio, MIDI and effects-assisted mixing via its MIDI sockets and its audio ins and outs. Not only that, but they had already written the necessary drivers for the Macintosh to talk to all these interfaces directly through its FireWire port. There was also a four-page handout which, instead of the hard sell normally associated with a product brochure, opted instead for a softer explanation of the protocol and the benefits it would bring. Since then, Yamaha have been actively recruiting other manufacturers of musical and audio equipment to develop mLAN interfaces and options, but their biggest triumph must be to bring Apple themselves into the program.
mLAN In The Mac OS
At their worldwide developer conference in Cupertino in May, one of the features of Mac OS X which Apple was planning to preview for their development partners was a multi-channel upgrade to Sound Manager which uses mLAN to send audio out over FireWire. Another piece of the puzzle which has recently come to light is that having failed to persuade Gibson, current owners of OMS, to sell it to them, Apple have instead hired one of the principal programmers of OMS to develop MIDI communications side-by-side with the audio channels. In other words, Apple appear to be planning full-scale support of mLAN, including MIDI, within the Mac OS. This could be a major advance for the support of music on the Mac, the like of which has not been seen since Apple's well-intentioned efforts with MIDI Manager were stymied by the legal shenanigans of the Beatles' corporation of the same name (sadly for all Mac-based musicians, this organisation had prior claim on the name Apple, and chose to use this to block the development of a built-in Mac musical interface, although it is difficult to see just how it could have conflicted with the interests of The Beatles).
With support for mLAN in the Mac OS including MIDI, its future seems not only secure but positively rosy! It means that Yamaha have secured the support of the computer platform which reigns supreme at the more professional end of the music industry -- the place where cutting-edge developments normally take place. For Apple, it is a bold stroke which answers a real problem that has been emerging recently with third-party MIDI operating systems. The most widely supported of them all, OMS, already looks pretty long in the tooth, with large chunks still written in 68k code, and was almost buried recently by crashing problems with USB MIDI interfacing. MOTU's FreeMIDI, which seems to be much more reliable, is sadly not supported by the major European sequencing packages. This means there is a gap into which mLAN can slot, unifying multi-channel digital audio and MIDI within the Mac OS.
Increased Software Support
So if there are all these advantages to mLAN, you might ask where the catch is (in my experience, there often is one with any major new development). The main one I can identify so far is that whilst there will be a cost saving on separate outputs, A-D and especially D-A converters for separate outputs, and perhaps no further need for half a dozen different connectors for all the different signals which may be coming in and out of a musical instrument, there is a corresponding increase in the amount of software that needs to be written. Now, whilst the cost of that may be deferred over enough sales for it to be far less than some of the hardware savings mentioned above, I can see a situation where the unfinished software syndrome on new products could get worse. It won't be so bad if this means that products are released without separate audio outs implemented via FireWire, and these follow later via a free update. What will worry me is if manufacturers try to offer all the published facilities, but without the extra code necessary having been properly debugged. This could make new products (even) more unreliable, and make the already familiar feeling of acting as an unpaid beta-tester even more common.
But it is not fair to prejudge things; the advantages which mLAN will bring will not necessarily trawl a thousand bugs in their wake. At least the programmers will be concentrating on one protocol, mLAN, so the potential for conflicts is perhaps lessened.
The other caveat, which history has taught me well, is that no matter how much bandwidth you have, one day you will run out. Remember how 16 channels was going to be all we ever needed for MIDI? And surely, no-one was ever going to use better-spec digital audio than 16-bit, 44.1kHz? SCSI was also the fastest data interface we would ever need -- then it was SCSI-2, then Super-Mega-Wide SCSI 17... and so on. The famous last words of interfacing are "well, you'll never need more than that, now will you?"
Having sounded that cautionary note, I do think that if you just use audio, or just MIDI or just video, then you will probably be all right for a good while. The problems will come if you are trying to send four streams of video for DVD with a 5:1 surround mix of 128 incoming audio channels, whilst sending MIDI timecode, the data from 258 MIDI busses and Uncle Tom Cobbley and all. We must learn the lesson of USB (to think how many different things you are trying to do over a single interface) and make sure that the different types of signal do not tread on each others' toes.
Next month, in Part 2, I will look at the actual nuts and bolts of how the IEEE 1394 hardware allows mLAN to do what it does, and exactly what kind of performance you can expect from a FireWire connection running simultaneous video, audio and MIDI.