Article from Music Technology, December 1989
Steinberg Research are responsible for some of the most significant developments in computer music systems in recent years. Nigel Lord talks tech with Charlie Steinberg on a rare visit to Britain.
WITH ALL THE ATTENTION STEINBERG'S CUBASE SEQUENCING SOFTWARE HAS ATTRACTED IT SEEMS AN OPPORTUNE MOMENT TO PUT THE MAN BEHIND THE NAME ON THE SPOT.
IN THE FIELD of music technology, the pressure to absorb increasingly high levels of information is coming to be seen by a growing number of people as a law of diminishing returns. In recent years, I myself have adopted what can only be described as a "need-to-know" approach to accumulating knowledge - especially in areas such as MIDI. If, for example, I come up against a problem which an understanding of one of the more esoteric MIDI functions would help overcome, I pick up one of several excellent reference books on the subject and look it up. If it seems particularly useful and likely to crop up again in the future, the chances are I'll remember it, if not, then I probably won't.
The problem, working on a magazine like Music Technology, is that it's all too easy to convince yourself that you're surrounded on all sides by people for whom the complete and unabridged MIDI specification has become a second language and who have the instruction manuals for every major instrument released over the last five years neatly stored away in memory ready for instant recall.
And so it was, when confronted with the prospect of an interview with one Karl "Charlie" Steinberg (yes, that Steinberg), I felt a sudden and uncontrollable inferiority complex descend upon me which was to last until the moment I found myself in the bar of the Kensington Hilton with a glass of Clausthaler in my hand (to keep a clear head), face to face with the man himself. And then it lifted...
Charlie Steinberg is clearly a very clever man, but he wouldn't want to burden the world with the weight of evidence. Far from the manic, machine-code babbling Teuton of my imagination, here was a quiet unassuming individual for whom the design of MIDI software obviously still holds a strong fascination, but who suffers in no degree from the rather oblique perspective of those who live and breathe technology. He is quite happy to point out where his areas of expertise lie, and, just as importantly, where they don't. Thus, on a number of occasions throughout the interview, he re-directs my questions to his partner - Werner Kracht - one of the principal designers of Cubase, the new flagship of the Steinberg range, and a name which users of a number of Steinberg products will possibly be familiar with.
AS A LONG-time user of a couple of Steinberg packages (and having been severely impressed by Cubase), the quelling of any initial trepidation as to the sort of person Herr Steinberg would prove to be, left the field open to ask a lot of those questions which few users get the chance to ply software designers with. For example, how do bugs manage to find their way into programs which presumably go through rigorous testing before release; and why can't MIDI standard actually mean standard? But first, let's get the more predictable questions out of the way... Like, how did he come to be writing music software for a living?
"It started when I was working as an engineer in a studio, and I was into electronics and computers. My partner at the time was a studio keyboard player who was also working in a music shop, and he showed me the first MIDI spec. So I started writing a MIDI program for a Commodore 64 which we had - just as a hobby. But it turned out that people were really interested in it, so we built a small interface, and it just grew from there".
The program of which he speaks was, of course, Pro16 which emerged in 1986 to the acclaim of all those who used it; music software had come of age. But what prompted the move over to the Atari as a host computer?
"A lot of musicians in Germany were really interested in the Atari when it came out -just because of the MIDI sockets. At that time there was no real software for the machine, but they bought it because they thought it had potential. So we started to look at it. And we got a lot of support from Atari in Germany who were interested in what we were doing. "
On the subject of Atari, there has been much speculation over the years about just how MIDI sockets ever came to appear on the ST, especially since so little mention is made of it in the computer's manual. Did their contact with the company ever reveal the reasoning behind it? "Well, the story goes", offers Kracht, "that it was the son of Jack Tramiel (the head of Atari) who was a musician, and he pressed his father to put them on..."
Sounds plausible enough, I suppose - this is an American company we're talking about, after all. But how did Pro24 itself come about? Was it started from a clean sheet of paper, or did it prove possible to upgrade Pro16?
Kracht: "No. It was started from scratch. I had to learn a new assembly language; the two computers really can't be compared - the Atari is a much more complex machine. On the C64, for example, you could write a program in less than two months, but the ST takes much, much longer".
So it wasn't possible to upgrade between the two computers, but how about upgrades on the same machine? I've often wondered if each version of a program like Pro24 is considered to be the "last word" at the time it is written, or is it always assumed that another version will follow?
"It is a continuous process", explains Kracht. 'We'll develop the new features for the latest version, then there's a break, a month's testing, and we start on the next version."
Was Cubase, perhaps, originally planned as Version 4.0 of Pro24, or was it always conceived as a separate program?
Steinberg: "It was always separate. We learned a lot from the Pro24 and we got a lot of user input which has been put into Cubase as far as it is possible. But the programs were always different. The basic structure of Cubase allows for expansion: we put a lot of work into creating data structures which would make it expandable. And that wasn't the case with Pro24. We had upgrades, as all software should have, but we couldn't always foresee what would happen in that field. Now we have much more experience and have created a program which can be easily expanded."
"FAR FROM THE MANIC, MACHINE-CODE BABBLING TEUTON OF MY IMAGINATION, HERE WAS AN UNASSUMING INDIVIDUAL FOR WHOM MIDI SOFTWARE STILL HOLDS A STRONG FASCINATION."
Encouraging words for those who have already taken the plunge with Cubase, and certainly very much in line with the company's assiduous approach to updating which has always made their software such a worthwhile investment.
When was the last time you heard of someone offering practically the full purchase price back on your old equipment when traded in on this year's model?
Those who have made the decision to upgrade will no doubt be aware that though there are considerable areas of overlap between the two programs, a fairly substantial conceptual leap has had to be made to arrive at the kind of advanced data manipulation possible with Cubase. Having said that, one of the most impressive aspects of Cubase is its inherent simplicity: though very sophisticated in music software terms, there is nevertheless a logic to its structure which makes it accessible even to those who prefer to leave the manual on the shelf...
"That was something we were very much aware of when we were designing the program", explains Kracht. "I know a lot of guys who can work with this program who have never used computer software before."
"We've also been very careful in the layout of the program", Steinberg continues, "by thinking about what were the most important things for a musician and making them as easy as possible to access. Really, it's all a question of appearance and making it possible for people to work with data in the way which suits them best".
But would he not agree that by providing so many different ways of achieving the same result, you effectively make the program more complex than it needs to be - especially for those struggling to understand its operation for the first time?
"Yes and no. In some cases there's an obvious way of accessing something which we should decide when the software is being written. But in other ways this would limit your creativity because people have so many different approaches and different ways of working. I don't think a program should push you in a certain direction".
And it's always better to have too many choices than too few. It has often occurred to me, however, that by designing software which appeals to both studios and to individual users, you're in danger of having to compromise a little too much. Surely there's an argument for writing software specifically for the needs of studios and software more appropriate for home or individual use? Steinberg doesn't agree:
"I think one of the most important things about software packages is that the low-level user can buy a program that is also used by professionals - it gives him a very good feeling. Of course, with Cubase it's likely to be a bit more divided - it is really a high level product - but it's still available to anybody who wants to use it.
"Maybe for beginners there should be smaller programs to lead them into the field. But I think that when you start working with something like this you soon feel the need for something better to expand your creativity. And the thing about these programs is that you can sit at home and write a piece of music and then take the disk into the studio if you want to record it."
WHETHER USED BY studios or individuals, the most tempting aspect of Cubase's design is almost certainly the multitasking facilities which it offers - a product of the M.ROS operating system under which the program runs. But was the adoption of a new operating system necessitated by decisions as to what the program was to be capable of, or was it, perhaps, an option taken at an early stage which only revealed its full potential as the program developed?
Steinberg: "No, we realised two years ago we'd need to have a system like that. The whole multitasking thing came from work we were doing with the SMP24 which as you probably know was designed to serve the Pro24. After the routines for handling different devices were written, we had a pool of routines which we could use for connecting devices for timing and suchlike, and we didn't have to put much more into it to make it multitasking. So that's where it started; and looking to the future, we knew we'd have to have a system like M.ROS."
One of the criticisms which has been most often levelled at Cubase is the sheer quantity of RAM required to run it in true mulititasking mode - that is to say, with two or more programs resident in memory at the same time. A few words of clarification from Steinberg and Kracht, however, and I have to confess I began to see things in a rather different light. Instead of looking at it as a program which was made so complex it had to have a Mega ST to run it, it becomes clear it was always intended for such a high level of onboard memory but thanks to the skill and ingenuity of the software writers it has been tailored down to enable it to run an a standard 1040 as well...
"Actually, many people aren't aware that you can run Cubase on a 1040 pretty well", Steinberg explains, "RAM is always a problem, but we put a lot of work into reducing the amount of RAM that the program needs just so that it can be used on the 1040. In Germany, musicians seem to be more aware when they buy a new machine that they're going to need more RAM these days. I could never work with less than four megabytes - not just when developing the software but also when making a serious piece of music."
"ONE OF THE MOST IMPORTANT THINGS ABOUT SOFTWARE PACKAGES IS THAT THE LOW-LEVEL USER CAN BUY A PROGRAM THAT IS ALSO USED BY PROFESSIONALS."
And Kracht points out: "Most professional programs like Desktop publishing and graphics packages need a Mega ST to drive the laser print".
Clearly, awareness is growing that if you want enough processing power to indulge in multitasking, for example, there's no alternative to RAM - and large quantities of it. But to some extent, it's a reciprocal arrangement: as software increases in sophistication, the demand for bigger computers and hard disks will increase, and as people have more computing power at their disposal, so the programs will become more complex.
All of which begs the question, how far can you go with a machine like the ST, now into its fourth year of production? After all, no matter how much RAM you load into it it's still just an ST. Does it remain a good enough host computer to warrant that kind of expansion? Steinberg thinks it does.
"Even now there's plenty we can do with it. Memory is a problem, but it's affordable - at least for the semi-pro. Of course, there's more to come: Atari will bring out the TT soon, and though it's more expensive it will be the way to go if you need a faster machine and you want to push the limits even further".
On the subject of pushing things to the limit, the advent of such facilities as realtime MIDI sound processing on Cubase (as well as a number of other sequencing packages around at the moment) seems to make a mockery of the criticism levelled at MIDI when it first appeared: that it was inherently limited. Steinberg agrees: "From the technical point of view, it's true to say MIDI is pretty slow and not very accurate - but most people don't actually hear any of that. There's so much information reaching your ear that it's satisfying enough. It might be different for professional musicians who know just how important it is for timing and so on - they have a more refined ear and can hear if the snare is a bit behind, or whatever, because of MIDI delays.
"But these problems can usually be solved because the range of the delay always has to be related to how it is perceived by the ear - and as long as it sounds good it's OK. In fact, you can squeeze anything out of the Atari's MIDI ports and to most people it will sound OK. It's different for professionals: they tend to have more MIDI outputs - using SMP24s for example."
Presumably Steinberg doesn't feel the much-mooted MIDI version 2 is ever likely to see the light of day?
"I don't think so. I think there will have to be a faster interface for sampling - its pretty boring waiting for a half-minute sample to be transferred to the computer via MIDI - but that will depend on there being a proper sample standard. Also, in the future I think there will be networks which have MIDI links - in fact that's already happening. You have units connected in the normal way through MIDI, but the buss is driven with a very highspeed interface so that you can control many units with normal MIDI data."
For more information on one such system, see the Media Link feature last month. But as software designers, have they ever felt restricted in what they could achieve by the limitations of the MIDI spec?
"I suppose I'd have to say yes", admits Steinberg, "but that's just a matter of quantity. MIDI is fast enough for sending, say, four or eight voices if these can be split up into several paths. And as I say, I think that will be the next step - having separate MIDI links. But you have to remember the computer itself has its limitations: MIDI is pretty fast compared to the calculating speed of the Atari... "
Quite. And as if to prove just how groundless the claims were of those who doubted the viability of MIDI just a few years ago, my attention is directed toward the Mega ST which had been set up in the corner of the room ready to reveal the delights of Cubase version 1.5 (which is being prepared for release even as I write). Selecting the Options menu reveals a new addition to the arsenal of facilities already on offer in the form of a dynamic MIDI Manager, which provides real-time control of equipment parameters via MIDI, making it possible, for example, to edit reverb levels or synth parameters from Cubase itself.
You can create your own on-screen graphics for the controllers you wish to access, you can size them, determine their range and what kind of data they put out. Not only that, but all changes in controller data are recorded in real time along with the track, and "snapshots" of any setup may be taken and stored for future use. Obvious applications in MIDI controlled mixing systems suggest themselves, but given the open-endedness of the system, a potentially huge array of control parameters may be brought within reach of the operator without ever leaving the main program.
Looking at Version 1.5 of Cubase, there can be little doubt that this is the kind of music software which will take us through to the early 1990s - and therefore, to the tenth anniversary of MIDI itself. Which reminds me, weren't there a couple of questions about the more annoying aspects of MIDI software I was going to ask? Like, how do those bugs still manage to creep in? Charlie Steinberg has his answer ready.
"It's simply human error. You have to remember there are perhaps 50,000 lines of machine code, and the computer goes through one command after the other. If you have done something wrong, sometimes it will crash immediately but very often it's like... You do this, then you do that and if you then do this - it happens. And of course you cannot test for all the possible combinations. But on Cubase, for example, we had about 20 or 30 people testing the program over a long period and they wrote down all the bugs as they occurred and we were able to sort them out. It's not a perfect system but it's the best that we can do."
OK, then perhaps he can throw some light on why MIDI standard still doesn't actually seem to mean standard? But the look of resignation and the shrug of the shoulders tells me he can't. Clearly there are some aspects of high technology which mystify even Charlie Steinberg.
Interview by Nigel Lord
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!