In Part Two, Nigel Jopson concludes his talk with Peter Zinovieff, the creator of Electronic Music Studios.

In the same rack as the Analytical Engine is the Synthetical Engine, or DOB (Digital Oscillator Bank). This was designed by Peter Eastty and David Cockerell for EMS, in order to give the studio's computer, which was already capable of handling sophisticated music programs, an equally sophisticated instrument on which to play the music. When you read the specification for the DOB, you'll realise just how relatively puny conventional analogue VC instruments are in comparison to the facilities at the EMS studio.

Three D-to-A converters feed the outputs; there are three banks of 64 digital oscillators each, all oscillators are tuneable from 0 to 16 KHz in 0-125 Hz steps, and each can be at any amplitude on 256 steps (ie each has its own envelope generator), and each oscillator can be assigned to any waveform (generated previously by computation). The data exchange-rate for the computations involved in setting parameters has been kept to a minimum, in order to allow realtime playback of sounds. All variables, like the distribution of frequencies, the waveforms and envelopes, are programmed by the computer as a set of what are called 'function tables'. One of the programs, (DOB uses the same MUSYS programs, like 'VOCAB', as the filter banks), will give the composer access to these function tables, which can then be worked out empirically, drawn on a TV screen with a light pen, 'or you can try random numbers if you think you're in luck.' Switches on the front of DOB allow the user to examine the state of each oscillator for FFT (Frequency Function Table - curve of oscillator tuning), AFT (Amplitude Function Table - linearity of envelope curves, log or lin envelopes), and WFT (Waveform Function Table).

The main sound-producing devices (apart from Synthi 100 and Vocoder) connected and controlled by the computer are as follows: the 500 filters of the Fast Fourier Transform Analyser, the 128 filters of the Analytical Engine, the 64 analogue filters (which can also be configured as oscillators), and the 192 oscillators of the Synthetical Engine. This gives a complement of three analysis systems and two synthesis systems at the sound processing side of the computer-controlled equipment. Next in line for examination were the various devices at the programming side of the set-up.

'Squeeze Me', so called because of a layer of special foam rubber beneath the keys, is a seven-octave keyboard for which the computer can analyse three functions: which notes are being pressed, how hard and how fast they were pressed. Thus, as well as being a completely polyphonic device, each note can be 'read' by the computer for its own envelope. Zinovieff explains: 'It isn't really an instrument, because the only thing it says to the computer is which keys are being pressed. In order to demonstrate that I often have a program running where the keys are actually a typewriter — in other words it doesn't really matter what you press — it's what the program does with what you press. You could arrange it so that the scale was an octave between the top five notes, and then very microtonal for the rest — it's all a question of program.' A similar 'computer logic' applies to the 32 sliders and 32 switches above the keys — the computer scans these and changes various parameters according to their positions. When a computer program is called up there are 'default values' — for instance, slider 11 normally controls the overall amplitude multiplier - but any slider can be assigned to any one of 500 different functions by initiating a computer sub-routine called 'QSU' (Q for Squeeze Me, SU for Slider Update).

Zinovieff's dedication to producing inspirational input devices is demonstrated by another, stranger, typewriter. It's just a little box with keys on it, which the computer interrogates and interprets as typing, but it's radio controlled, so that you can address the computer studio from the seclusion of your favourite field. 'It also has a radio receiver and miniature loudspeaker on it so that you can have the sound transmitted back to you — you should be able to go up to half a mile away.'

In the 'quiet area' of the computer studio is another piece of 'special' EMS equipment, 'Show Me'. This device has a TV screen and a light pen plus associated controls, and can do three things: it can display typing, like the VDU; it can be used to answer questions and give commands by simulating things like the 'Squeeze Me' sliders (in the form of bars of light) which can be 'moved' by propelling the light pen along them; or it can be used as a plotter by drawing curves (say for the DOB waveforms) on the screen, or you can just draw a few points and get the computer to join them up, or the computer can be made to draw on its own (as a test program, the computer sketches a spider's web, and finishes off with a little spider in the corner!) 'Show Me's TV screen can also be used in conjunction with a video camera.

'If you pointed the video camera at a score, then there's no reason why that score shouldn't be interpreted. Say the score is something very simple, like the curve of a waveform, then you could feed that directly to the computer.' Peter described in his Diary (1976) how the video camera interface was tried out; the camera was pointed at flowers in the garden, and certain functions were allocated to parts of the picture, for instance, density of image to harmonic content.

'...I run VOCAB and I assign the video parameters to control DOB. Suddenly the most marvellously rich, varied and amazing sounds come pouring out of the speakers. They ripple and change. They are the first absolutely fantastic sounds that the oscillator bank has made. Everyone is moved. At first no-one believes that they are not pre-composed but as I turn the camera the sounds change. They are convinced. In some ways we are frightened. It's like the story of tuning in to the death cries of roses as they are cut.' They rang Germany and Paris and held the telephone to the speakers; the children came down to listen. Perhaps these are the sort of moments which make the computer studio exciting and rewarding in a way that no other music could be.

Computer programming, good programs, is clearly the key to whether a studio like this lives or dies: without these, the computer would be like an analogue synthesiser with its potentiometers disconnected and no patch cords or pins. I think this is the department in which Peter Zinovieff and his studio will reach truly valuable and productive ground, while other computer em experimenters remain cloistered or concentrate on their own esoteric projects. The new power of machinery like that which I have been describing relies, for its relevance to music as a whole, on the controlling programs' accessibility to 'ordinary' musicians and composers.

'I hope that in a few weeks, when I've finished this update, all the instruments will be easier to use. One of the problems is that all of the programs are so big that nobody can remember them — it's a real Kafka situation. That's why I always warn people, when they're building a studio, and they've got some brilliant engineer who says "Don't worry about the programming, we can get that together" — they're absolutely wrong — the engineering counts almost for zero in the end, because you can always replace hardware, the programming is what really takes the time.

'MUSYS was written by Peter Grogono in 1972, and the other maths programs by Alan Sutcliffe a bit later than that; my first program, which has got lots of similarities with the present VOCAB, was written in 1969 or '70, and so every new program uses all of this experience, and uses a lot of the actual sub-routines. It would be very difficult to come to a complex studio like this and start programming again - it would take at least three years. It takes a very long time to complete a good program. It's very simple to operate one piece of machinery, but that's not what one's interested in, it's blending it together with all sorts of controls. Even with a very simple program, for perhaps every hundred things there's a mistake — whether it's hardware or software.'

Peter reckons that there is about a year's 'transitional period' when people come to work permanently in a studio such as his '...and then when they go, bits of the thing will never work again. Peter Eastty has gone now, and I think when the digital filter bank breaks down it will never work again — so that will be an investment of £40000 down the drain. Jim Lawson, who was our programmer, left without finishing the very good work he was doing, and that was like a year's effort completely gone. I think this problem occurs with any advanced technology, and this is as advanced in this field as you can get.'

So far we've had a run-down of the sophisticated equipment complement of the studio, plus a bit about the philosophy behind it and the task of running it. To tie it all together, I asked Peter to give a (rather simplistic) example of how the hardware and programs might be used in practice. The piece of music starts as a sound (mudsplashes, for instance), which has been recorded on tape, and will now be used as the basic 'input' material.

Peter: 'I'd use VOCAB, which is our system program, I would decide whether to use the digital or analogue filter bank to analyse it; then I'd start the tape recorder up and the computer would gobble up all this information and store it on disc. Then I could replay this from the disc, giving the information to the DOB, and out would come something that certainly wasn't as good as the mudsplash, but had a lot to do with it, with the right sort of frequencies and the right sort of changes. The simplest thing would be to play it directly back, with one filter corresponding to one oscillator, and well-tempered pitch, or whatever pitch you were analysing — this would be a 'default value' of VOCAB, it would naturally assume that you wanted the oscillators tuned to well-tempered pitch, and 1:1 with the filters. This is where the fun would start, because there's lots of interference that you could impose; you could process it only on the peaks, or only play every 50th event, or merge it together with wind noise, and all of these things you do as numbers. You could break up the tempo by adding a random element to the clock rate, or you could add an element that was derived from a set of numbers that you'd previously put in... and so on and on. When we were working on the piece with Birtwistle recently, we mixed the Bach chorale and church choir with one another digitally: the result was extraordinary, there were elements of both, with this mad organ coming out almost as if it were saying words.'

When the studio is working with a composer, Peter programs and operates the computer and Robin Wood operates the tape recorders and mixer, although EMS have proved in the past that the MUSYS computer language can be taught to composers with no previous knowledge of programming. Perhaps understandably, most of the people who've worked at the studio seem to have left the operating and deciding to Peter. 'It's important to get people working here who accept the limitations, because so often people think "Aha! a computer, anything is possible" — but it's only possible if it's interesting enough to work at it. The people I work fast with are the people who give you a lot of responsibility. Birtwistle is like that, and so was Henze when he worked here. They probably have in their minds some idea of the sort of sounds that they want to work with, but very little idea of what the final sound is going to be, and then you have to please them.

Robin: 'Harry might say, "I'd like an aura — very thin ..."'

Peter: '... perhaps high — or low — I don't know!'

Robin: 'So Peter taps away at something, sets a different function table, and says: "Is this anywhere near it?" Harry might say "Yes", but he's just as likely to say "I don't know."'

Peter: 'Then the real problem is to decide when you're going to shave off bits using fine control: you mustn't use it until you're sure that it's the sound they want.'

One day, one day, the sound that you want will be there, as simply as pointing a camera lens at a garden of flowers. But it will not be there by magic, because we don't know precisely what the sound is until we hear it; our descriptive powers are limited by the possible ways which we think there could be of interpreting our ideas. But if we can enlarge those possibilities to the greatest extent, then we have given ourselves a framework from within which we may be able to make inspired choices intuitively — like deciding which flower to point the camera at — and if the framework (read computer program) is good enough, then those choices will be creatively fruitful.

Peter: 'One is always struggling for the ideal way of interactive sequencing, you want to be able to sequence any type of sound and interact with that sequence.' I'm sure that the struggle would take a slightly downhill turn if only more people would realise that, for better modern music, there must be better, more versatile instruments to play with, and there must be support for the very, very few who are researching and developing the means of making them (make music).