The idea of connecting together various types of computers and data sources is becoming a topic of great interest to the computer world, and it's begun to make its mark on the world of music and MIDI as well. There may be a LAN in your future.

IF THE REMOVAL of any single hi-tech appliance could bring the world to its knees, it would have to be the telephone. As much as they can be a pain in the arm, telephones are the most important means of communication in modern society.

Lest you think this is a British Telecom industry profile, let me assure you that it's actually a look at music and communication, specifically communication over data networks. Telephones happen to be the oldest and best example of how such networks work. Someone wants to exchange information with someone else; telephones fit the bill admirably.

Computer Networks

LETS LOOK AT computers. In any organisation currently using computer information, or information stored on computers, one of the biggest problems is making sure that such information is available to every individual who has need of it. But people won't always be using the same type of computer equipment as each other. One of the problems here is that computer manufacturers (like most manufacturers of hi-tech products) would rather you used their equipment to the exclusion of anyone else's. So when designing their computers, they use different operating systems, different architectures based on different microprocessors - in short, anything to make your life miserable if you use somebody else's machines. This is called incompatibility - sort of a hi-tech equivalent to Madonna and Sean Penn.

Now that we know that some machines don't work well with others, and that one of the objectives of computing is the transfer of information, let's look at a music scenario. You and I work in the same office, recording studio, film editing facility or whatever. I've been working on my Macintosh II for the last six months, and you've been using your IBM PC AT for the past year. You've compiled a long list of people doing research in acoustics in your database; I've been storing up waveform samples of different experiments by those same researchers. This morning, you and I realised we're actually working on very similar problems, and decided that we should exchange databases and then merge them to make the best information available to both of us. We're all smiles until you see my Macintosh and I see your AT. Time stands still as we realise the futility of trying to put our months of research together by a simple data transfer. ATs and Macs don't talk to each other without a certain amount of hairpulling.

Dilemma defined. But you really don't care about our fictitious problem because you're really using a synthesiser not a computer, and information interchange and networking don't really matter to you. But do you use MIDI to connect to any other data information source, such as an effects rack or another synth? I thought so. That, in simple terms, is also networking. So you do care about networking after all, don't you?

Networking using MIDI is more convenient because all the manufacturers have agreed on a standard that is implemented on each and every MIDI instrument (excluding system exclusives, of course).

Before this agreement, we were stuck with the problem that exists in personal computing. If you were using synths made by the same manufacturer before the introduction of MIDI, the chances are that you could link them up in some nominal way to make them work together. But if you had a Roland keyboard, say, and an Oberheim drum machine, you could forget it. It just couldn't be done. It was a case of apples and oranges - or Apples and IBMs.

Networking has become the buzzword of the late 1980s. Applying it to MIDI instruments, let's say you and I want to work on different parts of the same piece of music - such as different scenes from a film score. Here's where we get into networking, or more precisely, local area networks, or LANs as they are called.

LANs

LANS ARE EXACTLY what you'd expect: networks for closely located machines that don't really have a great need for accessing information from the outside world. Office intercoms are a rudimentary example of a LAN, but since they are part of the phone world, they have an entirely different set of weird acronyms, usually having to do with PBXs (Private Branch Exchanges), but you get the idea.

"Networking with MIDI is convenient because all the manufacturers have agreed a standard that's implemented on every MIDI instrument."

A LAN allows you to connect your computer to mine, the office gossip down the hall and the boss upstairs. All of us can get at the same info without leaving our desks - even if we have different computers, because the network handles the exchange of information. So even if I can't plug my PC directly into yours, both our machines will be capable of accessing the same network and exchanging information over it. You can think of it as being like watching videotapes in VHS or BETA format; while we can't swap tapes between my BETA and your VHS machine, we can plug them into the same TV set and copy each other's videos. This however, is a passive form of data exchange, because the original material remains untouched.

A LAN allows active interaction, so that I can correct or update information contained in the master system, as can any other user in the system. I'm ignoring any discussion of data security and freedom of access to information. Although these are related topics, it's not really important here.

Adventures in MIDILAN

TAKING OUR BASIC LAN idea back into the realm of music, my IBM PC XT clone with a Roland MPU401 MIDI interface, and my Alesis Midiverb, DX7II and TR707 drum machine form a small LAN. By adding a friend's ESQ1 and Korg DRM1, we now have a multi-user network where more than one individual has access to the data (samples, drum sounds, patches, controllers and so on). The problem is that in this environment, limitations such as the number of voices, the linking of MIDI In/Out/Thru, and the general direction of the information flow determine who has access to what. If linked up in a particular way, my friend becomes redundant in the network and I exercise complete artistic and technical control over all the instruments. So much for data exchange.

Unfortunately this rather defeats the purpose of having a network, in much the same way a phone line running only from your bedroom to your kitchen would defeat the purpose of a telephone system. In order for all users to take advantage of all instruments and their assorted facilities, it becomes necessary to think about networking them.

Let's look at the benefits of putting our existing MIDI-based machines onto a LAN. First of all, we could do away with the concept of "master-slave" connections. Just as in the networking of computers, everyone has access to everything, and no user (player, performer, composer or whatever) has control over any other. A system of this sort would allow for multi-directional passing of MIDI information, which means that my friend in the previous example would be able to access the data in my instruments.

Another benefit of networking MIDI instruments would be the ability of all the machines to continually send out MIDI info to specific addresses or stations, without worrying about where that address was physically located in relation to the sender. If I'm Instrument One on the network, I might be sending out patch changes to Instrument Six, which in turn might be instructing Instruments Three and Seven on controller settings at the same time. The limitations now become how many voices each machine has, and whether they can accept certain kinds of change and edit information, but we have escaped the restriction of the cabling setup as dictated by MIDI In/Out/Thru ports.

Unless you're spellbound by the theory, you're probably beginning to wonder what the practical applications are for such an idea. Well, take live performance as an example. If a number of players are using MIDI instruments on stage, it is usually in their best interests for each to have their own setup - because each performer needs complete control over their sounds and equipment. The current MIDI standard specs don't allow for efficient multiple-user stage systems, simply because of the master-slave relationship described above. But what if you could take all those wonderful new instruments and racks and put them on a Local Area Network, so that each performer had access to all the data on stage? It opens up a lot of interesting possibilities.

"As the importance of networks grows in traditional computing, it will also become vastly more important to .the ways that we transfer musical data."

A real and very efficient use of a MIDI-based LAN would be in film scoring. Imagine you have a database of sampled sound effects in a number of modules, and you have a couple of technicians working to apply sound effects to an unscored movie. Instead of having only a single user who can reach into the database, with networking you have the capability of multiple access, so that work can be done simultaneously if necessary. It would certainly speed up the process considerably if a group of individuals could take separate parts of the same movie yet work on scoring those parts at the same time.

Unfortunately, we run into a problem here. With all of these wonderful interconnections and multi-directional information transfers going on, you start to slow down the network with the sheer quantity of information being passed over the lines. An alternative method of creating an efficient network has been espoused by Professor William Buxton of the University of Toronto. His idea is for an actual MIDILAN, which would contain a number of "nodes" of MIDI machines.

Each node would be controlled by a "server" - a computerised device acting as a complete MIDI merge and MIDI filter system. This server would then be linked to other servers controlling other MIDI devices. The servers would be linked via a LAN and would be the "postmen" of the system. They would pick up and deliver any messages to and from the individual devices within their nodes and send them to the appropriate "address". The beauty of this is that the LAN where the servers reside could send information faster than the baud rate (a communications speed of 31,250 bits per second) currently defined by the MIDI 1.0 spec. So there could be an increase in transmission speed between servers, before the information was slowed back down to the MIDI-specified speed.

This latter aspect, as outlined in a paper by Chris Meyer, could be used to overcome some of the problems of transmission speed inherent in the MIDI 1.0 spec. As we try and force more information down the throat of each MIDI cable, we increase the burden that the line must bear. Consequently, much of the demand for a MIDI 2.0 (or Super-MIDI) spec is from people requiring increased transmission rates. However, this could make all the current MIDI-equipped instruments obsolete. Such an occurrence would certainly cause unrest amongst the troops and would probably lead to a full-scale rebellion. Anyone who already had many readies invested in equipment built to the existing MIDI spec would be faced with the "old" problem of incompatibility obsolescence. It could even be the end of civilisation as we know it.

The MIDILAN does allow for an increase in speed, and thus overcomes some of the objections having to do with transmission problems. A number of manufacturers are currently interested in the concept, and it is anticipated that by the end of this year the idea of MIDILANs will be a very hot topic indeed.

AudioFrame

FOR THOSE OF you not willing to wait and see how the concept of a Local Area Networking of MIDI devices turns out, there is a glamorous, though somewhat expensive, alternative available today. An audio workstation developed by the American WaveFrame Corporation allows you to do all the things that I've been discussing here, but builds its network specifically on IBM's proprietary Token Ring network using PCs. The workstation is called the AudioFrame, and is composed of modules incorporating such things as analogue-to-digital/digital-to-analogue converters, sampling synthesis, 14 Megabytes of memory, a variety of MIDI and SMPTE options, and a link into the Token Ring network. The user interface is an IBM PC, either of the AT class or the new PS/2 systems. The software for AudioFrame is the basic MS-DOS running Microsoft's Windows, so there aren't any surprises in having to learn all kinds of new operating systems. Oh yes, and the system has the potential to interface with about 250 devices at one time.

Not only can you have multiple users on the AudioFrame simultaneously, but you can also have multitasking, which is having those users work on separate and independent applications at the same time. There are a lot of features and benefits to the AudioFrame, too many to go into here. Let's think of it as a Synclavier with a built-in LAN.

As the importance of networks grows in traditional computing, it will also become more important to the ways that we transfer musical data. The potential for use in recording and film studios, within bands, in sequencing and performing, and in allowing more personal freedom within an electronic music environment is enormous. One quick look at how fast things change in this industry will tell you that the commercial implementation of networks in music is not too far off.

The advent of a MIDI-based LAN may not be as earth-shaking as the installation of the first telephone network but then again, you can't perform your latest musical masterpiece on your telephone either. At least, not yet.