In the second and final part of our interview, Fairlight co-founder Kim Ryrie discusses linking his company's CMI to smaller systems, the future of storage media, and why DSP could be the musician's next big set of initials.

AS MY CONVERSATION with Kim Ryrie continued, we got around to one area that unites his company's ultra-sophisticated, ultra-expensive machines with those of less ambitious configurations - MIDI. The great five-pin standard has become an integral part of Fairlight's latest CMI, in as much as the Series III has three MIDI input and four MIDI output ports, and MIDI control is a central feature of the new CAPS sequencing software.

"CAPS can record multiple MIDI channels simultaneously", says Ryrie, "the original reason for this being that I suspected one day people might get interested in playing like they did in the good old days, in bands and so forth. I think that has to be the next step with MIDI-based recording.

"Another advantage of this approach is that if people are doing compositions using PC-based composer software, they can, if they want to, subsequently download to the Series III without having to transfer tracks one at a time."

So presumably, a MIDI sequence dump standard would find favour with the Fairlight team?

"Absolutely. Some of our users are already asking for that kind of facility. Some Fairlight owners are producers working with fairly small bands, and these bands are putting together their compositions on an Atari setup or whatever. They want to be able to load their stuff into the Fairlight and carry on working with that, which is perfectly valid."

Generally speaking, Ryrie is far from dismissive of the low-priced music technology that is emerging from the R&D labs of Japan and America nowadays.

"It's amazing what has come out. Of course we're aware of the Akais and the new Casio 16-bit sampler. They certainly keep us on our toes.

"I know that if I was putting a studio together at home I would probably tend to run out and buy some of those things too. But - and obviously I'm biased - I would still want them to be built around a Fairlight system. There's an inherent architecture and expandability about the Fairlight, and an inherent sound quality standard, that you really can't achieve with those kinds of products.

"Our basic policy is that we assume someone who is going to buy a Fairlight will be using that as the basic machine in their production environment. But we will be adding to the Series III whatever software might make it easier for people to use those machines with the CMI, if that's what they want.

"For instance, if you don't like multitracking you could instead buy an eight-channel Series III CMI together with other MIDI instruments, and have a fully-sequenced production which can be recorded direct to Sony F1 or whatever. You'll use the Series III for the sounds that you'll need it for, and you'll use the other machines for the sounds that they're good at. That'll allow you to get a compromise system which will be better than a £2000 setup, but which won't be the fully-blown Series III."

One area in which the Series III inevitably scores is its massive amount of memory, and the possibilities it offers for digitally-based recording. But as it turns out, Fairlight aren't about to rush into any one digital recording medium just yet.

"There are about six options that we're looking at", Ryrie reveals, "which include optical disks, R-DAT, high-density floppies with data compression, and hard disk - and also perhaps a combination of two of those. They all have different price levels, and what we've been doing is talking a lot with users to try and ascertain what they really want.

"What we've found is that not many of them want hard disk. Most of them have multitrack and they're not particularly keen to outlay the amount of money that a hard disk system would cost. They're also very concerned about the backup situation - the time it would take and the cost of it. R-DAT or optical would solve that problem, and the technology is just around the corner, though synchronised R-DAT drives are an expensive business at the moment.

"Erasable optical disks are working in the labs at the moment, but aren't yet commercially available. There is another problem, in that the transfer rate to optical disk is slow; you could only record two tracks at a time to disk, and even doing that would be hard because of error detection and correction requirements.

"There are high-bandwidth optical drives starting to appear, but at the moment optical disk transfer rates are running at about a quarter of the rate for hard disks. But the good news is that with optical disks, you're talking about half a gigabyte to a gigabyte of storage.

"If you're using WORM (Write Once, Read Many) optical disks, then the costs are just astronomical when you consider that you're only getting two tracks down. If you wanted to record eight tracks then you'd need four drives; each disk is costing you £100, and you can't erase it. That's obviously too expensive.

"On the other hand, if you were recording onto hard disk and using optical disks for backup, that's more efficient. But because optical can only record two tracks at a time, if you had an hour of recorded material it'd take almost that time to back it up before you could use your hard disk again.

"We don't like to leap onto a particular type of drive when we feel certain that the technology is not yet adequate. Otherwise you've got a situation where some people are spending maybe £3000 on an optional drive only to discover six months later that something really useful has come out at a similar sort of price. We have to hold back until all of this settles down."

ON A MORE general level, Fairlight are busy keeping up with new technological developments that may have significant implications for the music industry - though Ryrie is understandably cagey about giving away too many secrets about his company's investigations.

"We're in contact with a couple of different projects which we're hoping to get involved in, or which we're hoping will be suitable for us to get involved in. Both of these projects use completely new techniques for creating complex sound, and we're hoping that these will lead to a new generation. They'll be DSP-based, of course.

"We have a DSP team now who are working on determining what will be the most effective technology to use that will give the required flexibility and complexity of sounds, and control over those sounds in a meaningful way. Obviously control over sampled sounds is reasonably limited, but then it's also pretty limited on FM sounds with the technology available today.

"There's a new technology that allows just a couple of variable parameters to dramatically and predictably affect the whole nature of a sound.

"Both of these projects are happening at different universities in different parts of the world: one in Italy and one in England. The locations are a secret, though. We're very impressed by what both projects can do, and what we're trying to do is come up with a hardware solution which will be able to handle both of these technologies, because it's a bit hard to know which will be the more popular until you've done it and got them out in the field."

However, Ryrie feels that today's commercially-available DSP chips aren't up to handling the processing power required by these new technologies.

"I think DSP is the direction that everyone will be going in. But there are problems with DSP today. When you look closely at the current low-cost versions of DSP products, you'll notice that it's not the numbers which we have been used to associating with 16-bit digital. The reason for that is that when you start doing complex things (signal processing such as reverb and equalisation, for instance) there's an amount of noise generated by the mathematics that has to go on - through rounding of values.

"What it really boils down to is that the more powerful the processor, the less of that problem you get. So in a few years I think DSP will be the way to go, but right now you can use DSP but it's very expensive to get what you really need. We've avoided jumping into the early DSP processors because we feel it's a first-generation thing. We'd rather stick with the technology that we know, and use DSP when it's really firing on all cylinders.

"We're using Transputers for the video side of what we do, but we don't think they're particularly suited for audio applications - certainly not for what we want to do. A Transputer roughly has the power of a 68020, but the beauty of Transputers is that you can just pile them up on top of one another and do all your work in parallel.

"But digital audio doesn't work like that. Although the Transputer can do DSP processing, it's often more appropriate to use a chip which has been more specifically designed for DSP applications."

And for anyone out there who's having difficulty keeping up with the jargon here, Ryrie has a more easily-assimilated explanation of current trends...

"What we're talking about here is replacing electronic components with just a processor that will do the functions that you would otherwise get using logic gates and medium-scale integration chips. If you look at most systems these days, the hardware functions are done by integrated circuits that are basically doing logical functions. DSP means that you just have this incredibly powerful processor. You have an A-to-D converter in front of it and a D-to-A converter after it, and connected in there somewhere is a bit of high-speed memory. In theory that box can do anything - from synthesising sounds to playing back sampled sounds to real-time synthesis to signal processing."

DOES KIM RYRIE see everything ending up in a single box?

"I think that probably everyone is assuming there may be more of a workstation approach to the whole thing, whereby all the functions can be done from a very user-friendly control panel, a series of panels, a keyboard combined with a control panel, or whatever.

"Obviously there's lots of scope within that. I don't think that anyone is going to come up with the system that's going to suit everyone. Some people on stage like to have 10 keyboards, even though today you only need one to control everything. It's the same sort of thing in a studio: some people like to be surrounded by mountains of effects equipment with knobs, flashing lights and buttons; they might not be so happy to have this one rather quiet-looking control which is capable of doing everything they can think of. It may obstruct the creative process of some people to have to work like that. But in any case, that's the direction we're heading in."

At the end of the year, Fairlight will be releasing a Computer Video Instrument Series II, intended as a fairly low-cost broadcast-standard system. But in the longer term, the company is looking to integrate the audio and video sides of their operation.

"I guess our long-term goal is the production of a system which will be the next-generation audio system - and to some extent the present generation - with video production. Powerful video effects generation, high-resolution graphics, computer art... all that sort of stuff, in a package which is integrated. And affordable, because I think that's the key.

"It'll be two separate boxes and two separate products, but they'll be designed to work together from the user interface point of view, which is closely tied in with the creative process. Integration means the user interface. We need to work out ways of doing things which are specific to working on music and video simultaneously, which we haven't thought too much about yet.

"At the moment a composer is given a video and he adds the music to the video, which is a fairly simple situation. If it's a music video then of course the music is already done, and the video producer has to add the vision. First of all we have to cover both of these processes, but beyond that I think there's a requirement for creating both the music and the video simultaneously. There's no reason why a system can't be made friendlier in that direction."

Well, when you've got a tradition of research and innovation that's as strong as Fairlight's, there seems no reason why anything can't be done. Given time, of course.