The Sin In Synclavier

Mutant Timbres, Stupid Guitar Controller Tricks, & Typing Through The Tuplets.
As Told To Dan Forte.

Guitar Player, June, 1986


My history of working with guitar synthesizers goes back to a guitar-following device called the EWE, which stands for Electro Wagnerian Emancipator. There's only one of them; it was designed for me by Bob Easton at 360 Systems. If you played a single note, all 12 notes of the chromatic scale would be ringing, and you could make a decision as to which of those 12 to leave on and which to leave off – and thereby select a chord that would follow parallel to whatever you played on the guitar. It worked; the only problem was the timbre of the synthesizer sound that came out was, I would say, fairly unattractive – a real square wave sound. That is now gathering dust in the warehouse. I tried to use it on "Big Swifty" from Waka/Jawaka - Hot Rats, but it didn't end up on the final track.

On "Be In My Video" [Them or Us], a couple of cuts from Thing Fish, and the entire Francesco Zappa album, I used the Synclavier with no sampling – just the synthesizer sounds. The first time I used the polyphonic sampling was on Frank Zappa Meets the Mothers of Prevention; all of the material on side two, except for the little instrumental section with Thing Fish talking over it, was done with the machine. And on side one, "Yo Cats," which sounds like a little jazzbo lounge group, has real drums and Ike Willis' voice, but everything else is out of the Synclavier.

I recently tried out the Modulus Graphite controller, and it had problems similar to the guitar controllers I've played before – it just felt better as a guitar. It seemed to be a better instrument than the Roland I tried out originally with the Synclavier. The difference is that the Modulus Graphite's neck is supposed to be more stable, and you're supposed to have better isolation, less false triggering, and so forth. There was less false triggering, but for the way I play, there was still too much. I've only tried the early Roland model they had with the Synclavier, and it just wasn't right for me. If it were, I'd have the thing up and running right now. I wouldn't dissuade somebody else from buying one - apparently there are other people who can play it and make it do wonders – but for the way I play the guitar, and for the uses I want it for, it just seemed wrong to buy it. Instead, I enter all the data through the keyboard or the typewriter.

The Roland electronics won't read things like hammering on the string with a pick; it just chokes on that. The way in which I finger the instrument apparently is too slovenly for it to read. I don't mute every string after I play it; there's no Berklee technique involved here – I grab it and whack it. You can adjust the sensitivity within certain parameters, but if you adjust the sensitivity higher, that means it's going to pick up fewer nuances – so where do you draw the line?

Depending on what data I'm entering, it can be inconvenient to have to do it all on keyboard or typewriter. For example, I can't just sit there and play a solo on the keyboard, because I don't think in those terms – certainly not in real time, although I can slow the sequence down and play stuff and get some styling and phrasing. But the way in which the data is entered has a "drier" feel than if it had been played on the guitar. For one thing, on the guitar, you get to wiggle the intonation a little bit to make more subtle things happen. The tradeoff there, though, is that as you wiggle the strings and intonate it – all that nuance stuff – you generate masses and masses of numbers that fill your sequencer up very fast. That's certainly a liability because you can't do as many tracks of information. The bulk of the sequence is being filled up with inaudible data dealing with microtonal pitch adjustments. So a single line you might play of a minute-and-a-half duration, if you're using raw pitch, will fill up your sequence of 30,000 notes. Whereas if you type the pitches in, there's no dynamic or nuance pitch information entered that way. So you can have many more notes in your sequence and fill up all 16 tracks and still have space left over.

Allan Holdsworth came by the house with his SynthAxe, and that had some similar problems as well as some different ones, for me. One similarity is the string delay. The delay in the SynthAxe is caused by the MIDI delay, since the instrument doesn't have to count frequency like the Roland does. One set of strings tells you that the note has been initiated, and the other set tells you what the pitch is. It's insignificant delay on that end, but the MIDI delay is something I can feel. This doesn't bother some people, but for my ear and the way I want to use it, it just seemed inconvenient. Any time I can hear the sound of the pick and then after it the sound of whatever's supposed to come out, it bothers me. Even though it's just a few milliseconds apart, it makes me feel awkward.

I would have preferred it if the angle of the SynthAxe's neck had been adjustable by some sort of pivot. Everyone's body is different – arm lengths are different; trunk lengths are different. The engineering principle is interesting and has a lot of merit, but it would be more useful if you could readjust the neck angle.

The main advantage for me with the Synclavier is that I can imagine rhythms that human beings have difficulty contemplating, let alone executing. When I'm writing for a live band, I'm constantly limited by the physical liabilities of the people who are going to play the parts. On the one hand, you can say, "You just keep practicing, be insistent, and you'll eventually get the rhythm." The truth of the matter is, the more you practice, the more the musician hates it; you're never really going to get it spot on if the person is suffering to play the rhythm. Why subject the musician to that punishment and torture when you can just type it in and get the thing mathematically exact? I'd say that a musical ideal would be a combination of the things live musicians do best and the things the machine does best blended into a type of composition that lets each element shine.

In terms of the sounds, the samples we use are mostly homemade. They're pure digital samples done in my studio. Often the factory samples originally start off on analog tape, and sometimes you can play a sample up and down the keyboard and hear this residual, unwanted, blurry gunk traveling with it. In the case of the drum set, we recorded each drum and cymbal in isolation, so that there's no residual hardware ring from other objects on the drum set. And we recorded each drum in stereo – the snare, the kick drum, each tom, everything. The net result when you put this together into a patch and play it back on the keyboard is something surrealistically clean and perfect. I also have samples of lots of different types of guitars – a really nice classical guitar, several steel-string guitars done with pick and with fingers all in stereo, both ambient and close-up recording. I've got tenor sax done with full-length, eight-second tones, with natural vibrato, in stereo, with the Ben Webster "flaw attack" [laughs] - a really unctuous tenor sax sound. We also have samples of whole orchestral chords, from theLondon Symphony Orchestra – Zappa album. The LSO album was done on a 24-track digital multitrack, so you can single out string, brass, or wind sections to get really high-quality, isolated digital specimens.

For the sounds that don't resemble any other instruments, we have a whole classification of noises: one being the Evolver, where a sound starts off to be one type of an instrument, and by the time the note is finished, it's been turned into maybe two or three other instruments, all with a smooth transition. Then we have Resolvers, where different types of resynthesized vocal or instrumental timbres are located on each of the four partials, and by depressing a single key on the keyboard, you get a four-note chord that is actually four independent melody lines that resolve against each other to a final payoff. Then if you depress the key at the end of the payoff, you get a bonus of another bent. So you can have little melismas, little eight-note melodies, that occur, and all you do is push the key down, and it sings some kind of Renaissance cadence or whatever. Two of the partials could be resynthesized voices, one could be a resynthesized violin, and the other a resynthesized bassoon. Instant Renaissance ensemble when you hit each key. So you can imagine what happens when you play a chord [laughs] – it gets very ab- surd. It enables you to write things that you couldn't deal with under any other circumstance.

An example of one of the "impossible" things I've done is: While one instrument is keeping a steady pulse at 120, another instrument is doing a ritard, where each successive note is five milliseconds later than the note before, over a period of time. There's no accurate way to notate that, but it has an interesting feel to it because it slows down so gradually and so mathematically. You can do things like that within a bar – little accelerandos and ritardandos in five-millisecond increments inside of one bar.

Eventually, everything goes to tape because that's how the album is manufactured, but prior to that you can do all sorts of things; there's no question that that's a big advantage. Most of the editing I do after the basic composition is entered is done on what Synclavier calls the "G Page," which shows you three tracks, each with three columns of information. One column tells you the start time, another tells you the pitch, and another tells you the duration. Once you learn how to read that, you can edit on that page very fast, and you still have access to playing things on the keyboard while you're doing your editing – which you don't have when you're editing in the music-printing function. The keyboard is disabled when you're dealing with that; you have to do it either on the screen or in your head, or keep bouncing back and forth. If you do it on the G Page, you don't actually see any notes or staff or anything – you see numbers that represent in fractions where the beat is located in the bar.

When you're typing in the music printing section, there's a process called tupletization. It's not a real word, but it's an accurate description of what the machine does. You push a button, and then it says "tuplet." Then it asks you what kind of a tuplet you want. You give it a flavor, like 11. Then you tell the machine 11 over how many beats – for instance, 11 over 3. Let's say it's all in 4/4: You type in tuplet 11 over 3 beats, and hit Return. The screen re-draws, and now the first three beats of a 4/4 bar have been restructured. The way you deal with entering information in the music printing is, each bar of music is divided into what they call "edit blocks." For example, if you're in 4/4 and choose edit resolution 32, every time you move the cursor one degree to the right, you're moving it one 32nd-note edit block to the right. So in 4/4 with edit resolution 32, there are 32 edit blocks in the bar. If you do this tuplet thing that I just described, there are now probably 44 edit blocks over the first three beats. If you want to enter an 11-tuplet in there, you just give a couple of commands and it locates pitches inside this imaginary framework of an 11-tuplet over 3. And you're not limited to just entering 11 notes; you could enter four 32nd-notes for each eighth-note in the 11-tuplet, if you wanted, thereby winding up with a 44-tuplet. Also, if you decided that right in the middle of your 11-tuplet you wanted to have a quintuplet that began on the third note of the 11-tuplet, you give a second instruction and it gives you a second-level tuplet. We're building what is described as a nested polyrhythm – one polyrhythm living inside of another polyrhythm. With this machine you can nest three of them. After you've entered your quintuplet starting on the third beat of the 11-tuplet, you could then decide you wanted to have a septuplet that began on the second beat of the quintuplet inside of the 11-tuplet. Or any kind of tuplet you wanted, up to the maximum resolution of the machine. After you've typed in all the stuff, you push the Play button, and by golly, there it comes – and it's on time.

If you really want to get abstract and build your composition just on the G Page, instead of dealing with tuplets, you can deal in milliseconds. The rhythm on the G Page is determined by the start time of the note – that's the data that lives in the left-hand column. So you can read data on the G Page in three different modes: in terms of seconds, beats, or SMPTE [time code] numbers. If you're looking at a 4/4 bar at 120 in the mode that shows you beats per bar, beat 1 is the first quarter-note, beat 2 is the second quarter-note, etc., and beat 5 is the downbeat of the next bar. But inside of that, you have resolution down to five milliseconds. You can add and delete notes on the G Page. So you can build a list that would say: There's a note on beat 1 and there's another note on beat 1.005. Then the next one could be on any arbitrary number – you can just enter any kind of numerical scheme you want for the rhythm. For some of the kinds of rhythms I type in, my G Page tuplets look like beat 1, beat 1.07, beat 1.14 – in other words, this whole series of notes is going to be 70 milliseconds apart. You can have a series of nine notes 70 milliseconds apart, followed by a series of five notes 90 milliseconds apart, followed by any number of notes you want any number of milliseconds apart. You don't even need to worry about tuplets anymore – just go for the flow. You can have these notes be totally distinct from one another, or you can have them overlapping each other to make chordal arpeggios, just by changing the duration in the far right-hand column. In other words, if you want the notes to overlap – if they are 70 units apart, and you want every three of them to overlap – on the page it would look like the third note would last 70 units, the second note would last 140, and the first would last 210. The first note that plays would last the longest. The effect is like a little three-note arpeggio. That's what I do for 12 and 14 hours a day – sit there and deal with those kinds of numbers. It is the only way to write that kind of music.

MIDI needs to be faster, because if I'm writing in those small increments of time just for the rhythm, with the types of delays that are built into using MIDI to hook things together- it gets a little tedious when things [i.e., synth modules] talk late.

With any of the systems that are pitch-to-voltage, before the voltage counter can figure out what frequency the string is vibrating at, it has to wait until the decay of the blast of white noise that occurs when the pick hits the string goes away completely. There's no way around that that I know of The problem with fret-switching is that it's taking us all the way back to the Guitorgan. If the fret-switching is going to tell you what the pitch is, then you've kind of got what's happening in the SynthAxe.


This interview was reprinted in Best of Guitar Player.

Read by OCR software. If you spot errors, let me know afka (at) afka.net