What is it like to sit down at the Synclavier - the world's most expensive instrument - and create a piece of music? David Mellor finds out and tells the whole story.

FIRST STEPS

Since I wanted to explore the Synclavier, rather than merely use it as an alternative to my own MIDI system, I went into NED's demonstration studio with a head almost completely clear of any musical ideas. I wanted the ideas to come from the options available. What I had decided was that I would start with a mellow echoey piano and bring in an orchestra to support it. I also wanted to use plenty of sound effects in the piece. I sat down at the Synclavier keyboard with some trepidation...

An optional extra to the Synclavier is a WORM (Write Once, Read Many) optical disk drive. I described this in more detail in Part 1 (SOS June 89), but suffice to say that it is like a mega-mega floppy disk which can hold around five and a half hour's worth of samples! Storing that many samples is one thing, finding the sample you want could be quite another were it not for the Synclavier's librarian software. All samples (Sound Files) are listed under Categories, the names and organisation of which is up to you. For example, if you had an oboe Sound File you might categorise it under 'Orchestral', or under 'Woodwind', or under 'Long thin black instruments' - or all three. It is entirely up to you, but however you categorise, it is a very quick process to arrive at exactly the right sound.

As Nick Williams was scrolling through the Categories on the Synclavier's Macintosh II workstation, I noticed a Category named 'Water'. This seemed interesting, so I asked Nick to open the Category so that we could see the Sound Files it contained. There is a quick audition facility which we used to examine a number of items named 'Paddles1', 'Paddles2' etc. These turned out to be different samples of a canoe paddle striking water, or something similar. We loaded four of them into the Synclavier's memory and Nick made up a patch so that they could be played at their original pitches on consecutive keys. Since the Synclavier is an integrated sampler and sequencer - or 'digital memory recorder' as NED call it - track allocation is straightforward. Each new Timbre is automatically allocated to the next vacant track of the sequencer, whether it is loaded in fresh from disk or simply a modified version of a Timbre already in memory.

Recording was started by - would you believe? - hitting Record. I recorded a short series of paddle strokes and Nick made a loop out of them (each track can be looped independently) to last as long as I thought I would need them, plus a bit to spare.

Obtaining the right piano sound was a little more involved. Finding the basic sound from disk was no problem, but I wanted it to have a couple of repeat echoes. Obviously, this could have been achieved using a digital delay, but (a) we didn't have one, and (b) I wanted to keep the entire piece within the Synclavier as much as possible.

Each Timbre on the Synclavier can be made up from as many as four Partial Timbres, or Partials for short. Think of them as similar to four MIDI instruments playing on the same channel, each with their own sounds and modulation characteristics. We loaded the piano Timbre onto Partial 1 and copied it to Partials 2 and 3. Creating the echo effect meant reducing the level of Partials 2 and 3, and delaying the attack of each Partial by an appropriate amount. Nick also suggested that we panned Partials 2 and 3 left and right, which sounded good.

On top of the paddles and piano, I laid a couple of tracks of strings and a long looped sample of woodland atmosphere - bird song and rustling leaves.

REALISM

So far, the arrangement was quite simple - that's the way I like things - but I was impressed at how natural it all sounded. Thanks to 16-bit working no doubt, and the high recorded quality of the samples. The mood of the piece seemed to suit a clarinet, but the one we found on the optical disk seemed a bit 'dead' and unresponsive when we first tried it. The art of good sampling is to use the performance capabilities of the equipment to make sampled sounds realistic (or at least expressive to some degree, if absolute realism is not what you want). With the clarinet, we used some delayed vibrato - a very small amount. The vibrato modulation could have been sine, square, triangle, ramp, or any one of several choices on offer, but sine proved best on this occasion.

The keyboard velocity was set to control volume and attack. I found that if the initial reed 'rasp' on the sample was only played on loud notes, it sounded very much like a real clarinetist playing. Polyphonic aftertouch was assigned to volume so that I could swell a note after it was played. The response of the keyboard can, by the way, be altered for each new Timbre to suit either feather-touch or ham-fisted keyboard players.

Although the notes I played in the first take were OK, a couple were too low in level. We went to the event editing page to increase the volume. Unlike MIDI, volume level is expressed as a percentage. 50% volume corresponds audibly to a level reduction of 6dB, 25% to a reduction of 12dB.

The method of operating the Synclavier is interesting in that it combines basic operation from the keyboard's button panel with the possibility of more detailed editing from the Mac II workstation. In normal recording, most of the things that you would want to do can be controlled from the keyboard. These include manipulation of the Timbres, control over modulation, sequencing, and storing and retrieving the sequence. But when it comes to editing individual notes, there is no substitute for a detailed, informative screen display.

The event editing screen is particularly good. Figure 1 shows a listing of a different sequence. Four tracks are shown here, but the number can be varied. Obviously, the more tracks on screen, the less screen width there is available to show all the information. As you can see, in this case the Pitch names, MIDI Note Number and Duration of each event are shown. Note that the timings are resolved to thousandths of a beat! We also used the sequence editing page, shown in Figure 2, a lot for looping, duration editing, etc.

To extend the first section I continued with some more strings. Normally, the Synclavier would continue recording on the same track that the strings had been allocated to previously. Nick advised that I should alter the parameters very slightly. This would not be audible, but it would count as a new Timbre and the Synclavier would start a new track. This removes the risk of accidentally over-recording the start of a track. Although you can set a Mark and this will become the new start point as long as it is active, button pusher's finger remains a problem with all systems. We made sure to save the work after each new addition, as an extra safeguard.

Some celli, more clarinet and a stereo sound effect of children playing completed the musical content of this section. A non-musical 'percussion' effect was added to provide an ending. A quick search through the sample Categories on the optical disk provided a thunderbolt and an explosion - which we beefed up by layering bomb sounds on the Partials. Nick panned the initial strike of the thunder so that it started in one speaker, moved all the way to the other and back to the centre. The digital memory recorder remembers all pans as part of the sequence.

JUSTIFICATION TIME

So far, the music had all been recorded in free tempo. Now I wanted a strictly timed section with drums, bass, and all the other things that make up dietary fibre for one's musical digestion. Quantisation on the Synclavier is performed by pushing the Justify button and works as you might expect it to, together with a programmable click.

I led into the new section with a couple of helicopters. One flew from one speaker to the other, the other idled on the ground. The first needed a little gain riding (have you ever tried achieving a constant level on a helicopter in flight?). Gain riding can be incorporated easily into the sequence. As we had already assigned the modulation wheel to pan on this track, we used the foot pedal to control volume. I overdubbed this onto the track.

Fortunately, if you get a controller overdub wrong the first time, you can go back and do it again. The control information is kept internally separate from the notes. It took me several goes to get it exactly right.

So far, all had gone well, even though it was being improvised bar by bar. But now, after a quick sandwich lunch, inspiration wasn't coming with as much clarity as it might have been. I emerged from the helicopters with a hi-hat pattern, followed by bass drum and snare. I allowed one bar of hi-hat and three bars of full drum kit before introducing a Hammond organ and Steinberger bass (not to be confused with Steinberg!). This, as it turned out, wasn't enough to balance the sections properly. But unaware of this, I marched on into a 12-bar blues chord pattern with drums, bass and a string pad. The strings were excellent. They were stereo and really filled out the track with a minimum of notes. Unfortunately - a second problem - the hi-hat pattern was so busy that I began to run out of voices and 'note-stealing' was occurring. Yes, this happens even with a 32-voice Synclavier.

We didn't really sort out this problem until the second session, but one major contributing factor appeared to be that the decay on the hi hat sample was such that each 16th note hi-hat strike took up a lot more than a 16th note's worth of time. Cutting back on the decay reduced the note-stealing. A little wizardry from Nick on the Mac II cured it completely.

HOMEWORK

By now, it was time to pack up and go home. Nick reassured me that he could save absolutely everything we had done on disk and it could be recalled exactly as we left it. I wouldn't dare do this on my MIDI system unless I had dumped the tracks to tape first. There are just too many floppy disks and too much note-taking.

I decided to have a live guitar playing over a sequence of three repeats of the 12-bar pattern. I could have recorded this directly into the Direct-to-Disk system linked to the Synclavier, but I wanted to spend more time making it as good as I could than would have been available in the studio. So I took a DAT cassette of the track in progress home with me, transferred the section in question to multitrack, and overdubbed my guitar to that, playing through my all-valve Fender Champ.

Back in the studio on the second day, we started by transferring a DAT copy I had made of the guitar track into the Direct-to-Disk system. We topped and tailed it (ie. removed the unwanted bits at the start and end) before getting back to the track itself. We were at a stage now where the structure of the music was looking a bit shaky. I hadn't been taking an overall view of the link between the sections on the first day. On multitrack, this is the 'let's start again' time. But on the Synclavier, changing the structure was a fairly straightforward cut-and-paste job. We inserted three bars of empty space before the bass drum and snare came in, then copied three bars of hi-hat into it. The helicopters were also extended and all the following tracks were slid back in time to make way. Simple.

Of course, you can do this on any decent MIDI sequencer, but this was already becoming a big track, both in terms of notes and controller events, and also sample memory. Out of an available 32 Megabytes on this system, I had about three or four left by the time we finished.

Fitting the guitar into the track was less problematical than I had anticipated. Direct-to-Disk tracks can be integrated to work like any other sequenced track on the Synclavier. It was just a matter of finding the right start point. This took two minutes, literally. We finished the track with some more sound effects. I had intended to bring back the woodland atmosphere I used at the beginning of the music, but Nick remarked that it should be night-time as it was the end of the track. So a couple of cicada samples were sourced from the optical disk, and I mixed them with a few final paddles.

Although the system we were using had 16 separate polyphonic outputs, plus another eight on the Direct-to-Disk, we did a simple mix to save time. (I often spend more time mixing than I do recording. It did me good to have a bit of time discipline). It was obvious that some tracks would need different amounts of reverb - the only external effect we used. The sound effects were all allocated to outputs 1 and 2, the drums and bass guitar to 3 and 4, and the rest to 5 and 6. The live guitar had its own outputs from the Direct-to-Disk unit. The amounts of reverb added were respectively, none, a bit, a bit more, and none - since I had included some echo on the guitar at the recording stage. Apart from slight initial tweaking on the mixer, all the track levels were controlled from the Synclavier. The final mix was a straight hands-off pass through the track.

CONCLUSION

The first conclusion is that I want one, together with the optical disk library and Direct-to-Disk system! But since I know I can't quite afford it out of my pocket money this month, I'll limit myself to saying that the Synclavier is a hell of a system. It's true that many, perhaps most, of the features are available elsewhere. Also, given enough time and the same samples, I could have recorded the same track with a more modest 16-bit sampler, together with sequencer and multitrack, to almost the same level of quality. But the most important feature about the Synclavier is that it is all there, in one place and in one system (and I didn't get around to using any of the FM synthesized voices). There is nothing that I could have done with my MIDI system and multitrack that isn't possible on the Synclavier in half the time.

But as a parting thought to this installment, if the bad news is that most of us can't afford one, just think of the good news: Synclavier power and capability is bound to filter downmarket to more affordable levels over the next few years, and we will be able to have 'Synclaviers' of our own. The only trouble is that the real Synclavier will have developed even more and there will still be the unattainable to yearn for. Oh well, that's life isn't it? More next month...

FURTHER INFORMATION

NED UK, (Contact Details).