Roger Beaujolais • Bill Bland • Winston Clifford • Paul Jayasinha • Roland Perrin

In the final part of the series, Simon Zagorski-Thomas takes us through the last two MIDI controller sessions; now all that remains is the trivial task of editing the resulting MIDI data...

Just to be obtuse, this month's 'Wot, no keyboards' features, amongst others, Roland Perrin on... you guessed it: Keyboards. Also, after all this prattle about how brilliantly everyone played, you finally get the chance to judge for yourself — available on cassette, DAT and, uniquely in the history of humankind, also simultaneously on MIDI file for the Atari — The Wot, no keyboards mini-album. For details of how to snap up this incredible bargain, turn to the end of this article.

Tuesday 5th October 1993

ROGER BEAUJOLAIS: KAT MIDI percussion, Roland D110, Akai S1000.
BILL BLAND: Simmons SDX.
WINSTON CUFFORD: Ddrums 2, Akai S1000.
PAUL JAYASINHA: Akai EVI, EV2000, Oberheim Xpander.

This session was the one that most resembled a 'normal' multitrack recording session. For a start, we did it in machine time — using the bar numbers and metronome of the computer. In fact, Roger played most of the parts in using a sampled drum loop as his time guide before Winston played the drum parts. Again, we used Notator on a 2Meg Atari to do the sequencing but this was done completely using overdubs, and simultaneously sequencing and recording onto an Alesis ADAT digital multitrack. The reason for this was that Paul Jayasinha's EVI does not have a MIDI In to the EV2000 module (the EWI Midi's module does have one but that was only made for the saxophone version — the Electronic Wind Instrument rather than the Electronic Valve Instrument). Therefore getting the exact sounds that he wanted — a mix of the Oberheim Xpander and the two-oscillator analogue synth in the EV2000 — was only possible by putting his performance down to tape. The tune is called 'Slipping Through The Net', and was specially written for this cassette by Roger. Its rhythm is deliberately ambiguous; a 6/8 feel superimposed on a 4/4 rhythm, and based around a chord sequence using a lot of flattened fifth chords. Everything in the track was left unquantised — as are all of the tunes on the cassette — but with this superimposed polyrhythm it would have been pretty well impossible to find a suitable quantisation algorithm in any case.

We started by loading up a drum loop that Roger had used when writing the track; it had a 'four on the floor' bass drum with a 6/8 hi-hat feel over the top. Then Roger played a guide electric piano track on the KAT. The KAT mallet percussion controller is sold in octave-long modules, of which you can link up to four together — Roger has three. There are no internal sounds in the KAT, but it has a comprehensive set of MIDI output parameters and can send on up to three MIDI channels at once. The basic setup is for two channels (A and B) each of which has a set of parameters to change pitch (chromatic transposition or a +/-7 octave shift range), MIDI channel, highest and lowest note range, minimum and maximum velocity values and a choice of velocity response curves. You can also store program change numbers to be sent on each MIDI channel when a patch number is selected, and also select a MIDI channel and program change number to control your effects unit. The other variable is note length, which you can vary from 0.025sec to 6.375sec, and also whether each channel sends MIDI sustain pedal commands when the sustain pedal is used. Apparently KAT have plans to introduce a breath control system to allow pitch bend to be sent, but Roger was still waiting for news about that at the time of going to press. I mapped a MIDI foot pedal onto pitch control for him to try out but it didn't have a very intuitive feel and he didn't use it. There are, though, three pedals that come with the KAT that can be programmed to do different things — sustain, preset number up or down, and a third, user-definable option which can perform specific patch editing tasks — an octave shift, a volume change, a MIDI channel change, and so on.

There's also a third MIDI channel available (X), for which you can assign any note to any key — which allows for drum mapping. In fact, Roger sometimes uses this live to trigger backing vocal samples off the very top of the 'keyboard'. When using channel X you can also get the volumes or velocities of the channel X notes to be affected by the volumes and velocities of channel A's notes — a sort of real time interaction where both the settings of channel A and channel X are affected by each other.

Roger has had the KAT for a long while now and uses it extensively both live and recording; on his Vibrophonics album on Acid Jazz Records, all of the keyboard parts were played on the KAT. He went to the launch of the Simmons Silicon Mallet (see Louis Borenius' session in part one of this series) — "I've never seen so many tuned percussionists in one room at the same time before. It was an exciting new thing and everyone wanted to know about it" — but he was put off by the lack of attention to detail, as he saw it. "It's a three-octave instrument except that it goes from a low C up to a high B — it's crazy, it stops just short of the top of the octave. You can buy an extra octave to stick on the end as an add on module, but that stops at the B above as well". So Roger went for the KAT instead; as far as he's concerned, its main drawback is the lack of performance parameters like pitch bend, aftertouch or MIDI controllers to counteract the fact that it has a far narrower dynamic range than an acoustic instrument. Because of this lack of expression, he finds it frustrating to play live — it's great for backing but not so good for soloing.

Once Roger had laid down the guide electric piano (triggered fron a Roland D110), he played the bassline in (also D110), some string lines from his S1000PB, and then he redid the electric piano. Then it was Winston's turn. He had never played pads of any sort before and, indeed the Ddrums were new to all of us because I'd only borrowed them from the percussionist Preston Heyman the night before, and our tape op, Marcus, had struggled through the manual during the day whilst I recorded Roger. We used the Ddrums to trigger samples from the S1000PB rather than using its internal sounds, for no other reason than that I knew my library and thought it would be quicker.

Winston started out by playing along to Roger's keyboard stuff with a click, but we soon discarded this and he simply played along with the other parts. He found the Ddrums took some getting used to — initially he had a tendency to try to dampen the cymbal sample by grabbing the pad like a cymbal. To me, the main difference between the Simmons SDX and the Ddrums is that the latter has real drum heads and so I assumed that it would feel somehow more 'drum-like'. Winston said that the difference for him was greatest in the tangible sense — the sound wasn't coming back up the stick at him but was coming from somewhere else, and the dynamic response just doesn't compare with that of a real drum. That said, he enjoyed playing them and liked them for what they are — a completely different instrument. He thought that if he used the system in future, he would keep his cymbals and hi-hat acoustic, because it just didn't feel right playing them off pads or even the tubular metal percussion trigger that we had the hi-hat on.

Just as Winston got a really good take down, I remembered that I hadn't saved for a little while. Yes, that's exactly what happened... we lost it all. My Atari has an external drive and the connecting lead had been knocked loose — the bloody thing crashed halfway through saving. It was 10.30pm and the last time I'd saved was at about 5.00; apart from losing Winston's drum take, we also lost Roger's replacement electric piano part. I rebooted and Winston did another couple of takes — and got an even better take than the lost one in no time at all.

In a fine show of locking the proverbial stable door after the proverbial horse had proverbially bolted, I then proceeded to save and back up the tune every five minutes until the session finished. Next up was Bill Bland, who laid down a conga pattern that we later turned into log drums. He was playing the part in from the SDX using two of the tom-tom pads. On one we had a low open tone and the other had a positional separation of a slap and a hard muted rim sound — one in the middle and one round the outside of the pad. He tried to play it in using his hands at first but it didn't work naturally, and he found it easier to use a pair of sticks. He saw the possibilities of the SDX as an instrument, but found that the positional stuff was so unintuitive as far as conga playing was concerned that it was better for him to treat it as something completely different.

Next day, Roger re-did the electric piano and Paul came along with his EVI and Oberheim Xpander. He bought the EVI in 1987 and spent a concentrated six weeks mastering it. The main differences between the EVI and a trumpet are that you just blow through it instead of making that peculiar farty noise that trumpeters have to make into their instruments, and that you have to move a slider with your left hand to jump the intervals that you would normally change by making higher or lower pitched farty noises.

Paul finds that he's using the EVI less than he used to but he still plays it with Steve Lodder's band. He describes it as his "way into MIDI" and finds it to be very responsive and natural to play, especially when he uses it with analogue synths. He has used it with a D50, and with an S950, routed through the audio envelope shaper which uses the breath control to drive a VCA.

Paul used three sounds on the track — two for the different sections of the melody and one for the solo — all broadly pulse wave-based sounds with different degrees of filter sweep on the aftertouch that the EVI was putting out as breath control.

Paul has spent a great deal of time programming the Xpander to get the best response from the EVI and it shows up in the expressive movement that he gets within the sounds, as they respond to his breath control. It's a remarkably expressive instrument in his hands, but if this is inspiring any trumpeters out there to go out and get an EVI, they'll have to look on the secondhand market, because Akai stopped making them in 1989 due to lack of sales.

Paul played the melody lines in first and the solo afterwards. After that, he added a mute guitar part from his Emu Proteus I, a Roland SH2000 Vibes sample mixed with a D110 shakuhachi on the intro and outro, and a cabasa part played off the KAT but. triggering the SDX. Then we simply changed the sounds of the congas into log drums, added a fretless sound to the bass, got the drums up on separate outputs from the S1000 and put a mix straight down onto DAT. We didn't have to edit any of the parts at all!

Tuesday 12th October 1993

ROLAND PERRIN: Korg SG1 Sampling Grand Piano, Yamaha TX802.

So what exactly is there to say about keyboards in the light of all this? For me, the main thrust of this piece of journalistic exploration has been to find out how various MIDI controllers work as instruments in their own right. For those of you who have persevered through the whole series of articles, you may have noticed that the recurrent gripe about controllers from the people who regularly play them is that their dynamic range is so limited. Musicians have been quite prepared to alter their technique to get over glitching problems and the like, but what has dampened their ardour the most is the expressive limitations of MIDI controllers. Perhaps not surprisingly, it was the wind controllers that came out best in all of this — after all, they allow more expressive range by deploying three continuous controllers to add expression to a monophonic stream of notes: MIDI volume, breath control and pitch bend. How, then, do keyboards fare in this respect? Roland Perrin is a pianist and not a keyboard player. He uses electronic keyboards (his usual setup is Kurzweil but he used our setup in the Premises Studio for this session), but only because even if there is a half-decent piano available at a gig, he usually plays in Latin and African ensembles; with the volume of three percussionists to contend with, a miked-up piano will almost never give enough volume or a decent sound. Yet, for Roland, the dynamic range of an electronic keyboard doesn't bear comparison with a real piano. In the hands of a great player such as Herbie Hancock, electronic instruments can be used for 'tone painting', but as they stand at the moment, they don't allow the 'expression of touch' that an acoustic piano does.

As I mentioned briefly last month, I want to explore the possibility of making a MIDI instrument behave more like an acoustic instrument. Unfortunately, even the wealth of real-time transformations available on Notator were not enough to allow me to try what I wanted in real time as Roland played, so he sequenced a couple of pieces for me to tinker with afterwards. The first was a South African based tune called 'Lucky Strike' and the second was a Latin tune called 'Que Milagra'. Unfortunately the studio's sustain pedal chose this moment to shuffle off this mortal coil and so, after a fruitless manic tour of Hackney's music emporiums to try and find a replacement, I mapped a MIDI volume pedal onto the sustain controller (MIDI controller 64) and subtracted 60 from its value. Despite Notator's contention that a value of 0 is off and anything above it is on, values of 0-63 are off and 64 and over are on. By subtracting 60 from the value, when the pedal was fully down (127), it gave a value of 67 and the sustain was on. Anything under 124 and it was off — ie, as soon as you lifted the volume pedal more than a fraction. Fiendishly clever stuff, eh? Still, a working sustain pedal would have been even more fiendishly clever. In any event, Roland overcame the handicap of having to use a volume pedal for sustain and we sequenced the two tunes. I haven't had the chance to work on them yet, but let me explain the ideas that I have for trying to get a little more expression to show up in the performances.

Basically, I want to try and stretch the dynamic range of MIDI by using two synths to play the same note — a quiet one and a loud one. Obviously, to work together, to sound like a single instrument, they'll have to use the same kind of building blocks (we've got a DX7II and TX802, which are both 6-operator FM synths with almost exactly the same internal architecture). First, I'd copy the track and then choose a limited range of velocities and delete the rest. For example, in one track I might delete all the notes with a velocity of over 85 and in the other all those under 43. Then in the first track I would multiply all the velocities by 1.5 — so, for instance, a value of 85 becomes 127. In the second track, I would subtract 42 and then do the same thing. I would now have two sets of notes that retain the relative ratios of the velocities of the notes in the original performance, but they would be stretched out. If I put these tracks on different MIDI channels and sent one to the DX7 and the other to the TX802 and edited together two patches — one for the first track with fewer overtones, to be more like a naturally quiet piano tone and one for the second track with more brightness — and if I get their relative volumes right, I should have what sounds like a coherent whole and yet has a wider dynamic range than the original part played on a single sound.

If I also programmed a sound consisting of the natural harmonic partials of a given note, without any discernible attack portion to the sound, then I could use this in conjunction with the sustain pedal. When a piano has the sustain pedal depressed, the effect is not only to allow all the notes to continue when struck but also to allow other strings to start sympathetic vibrations, adding to the richness of the sound. If I therefore had a third track on a different MIDI channel, edited such that only the notes played with sustain pedal were left, I could add extra subtle harmonics to simulate sympathetic resonance.

How might all this sound? Will it work? I'm afraid your only way of finding out is to try it yourself or buy the cassette...