The misuse of MIDI for creative purposes. MARTIN RUSS lets his imagination run riot.

I am sure many of you have heard the strange 'buzzy' noises that wasps make when trapped in jam jars or under pint beer glasses - so have I, but this has only a tenuous connection with what I really want to talk about...

I really want to discuss the strange raspy sounds which appear in the background of Bass and other sounds produced on an FM synth such as a DX7, FB01, TX81Z, EK44, as well as on many other pieces of equipment based on digital technology - ESQ-1s, Synclaviers, Fairlights, PPGs, samplers of any breed/colour, etc... (henceforth known as a 'digital sound source'). Scientists and other techno-mutterers tend to call this sort of noise by a variety of hi-tech names like aliasing, quantisation noise and intermodulation distortion. Those readers more aware of digital audio will probably be nodding their heads sagely at this point, so I will refer them to such tomes as 'The Proceedings of the Audio Engineering Society' for more details.

Instead of mathematics I want to talk about trying to get rid of that rather intrusive background buzz found in bassy sounds and this, in fact, will lead us on to a whole range of interesting side-effects...

ANALYSIS

To start with, let's listen to the distortion you get on a typical DX7 MkI Bass sound when played near the bottom of the keyboard - your ears are a very good instrument for analysing sounds, all they need is a bit of training . OK, ears aimed at the sound. Listening? Wait... Did you hear it? To my ears it sounds grainy, buzzy and reasonably high frequency/pitched. Not only that, but it is composed of several different sounds (see Figure 1). First, there is a sort of broadband noise or hiss which follows the volume of the sound (this is quantisation noise and is due to the limited precision of the 12-bit processing used in the DX7). Then there are some buzzy, ring modulator-like sounds which differ from note to note (this is aliasing, a result of inadequate filtering at the audio output, and often a problem in digitally based systems).

The 12 bits used in the digital-to-analogue convertor (DAC) in the DX7 MkI give a theoretical signal-to-noise ratio of about 72dB - comparable with a very good quality cassette deck fitted with Dolby C, or a lot of the current generation 12-bit sampling machines.

Unfortunately the story is more complicated than that - some additional level control is accomplished by amplifying the DAC output by a combination of digital and analogue circuitry, and so we get a perceived signal-to-noise ratio overall of about 90dB, measured using the loudest signal relative to the quiescent noise output from the DX7. At the loudest output level, we therefore hear the DX Bass sound we want plus the noise we don't want 72dB below it. To some extent the DX sound obscures the noise, but not entirely. You can hear a similar sort of effect on the PCM sound on Video 8 video/audio recorders - at first, you seem to have a very good dynamic range, but when you start to listen hard, you can hear the noise zooming into earshot every time the signal gets loud. (I had better stop this hi-tech talk here before we get too technical, and before you start to listen too carefully to consumer hi-fi products.)

So, hopefully, we now have a better idea of what the enemy looks like - but what can we do? You could always buy a DX7 Mk2, which uses 14-bit DACs on its output, but this only pushes some of the nasties another 12dB down...

ANALOGUE FILTERS?

At this point I will re-introduce a word you may have forgotten - VCFs (Voltage Controlled Filters, as in Moog synths, ARPs etc). Yep, you may be surprised to learn that you can use the VCF out of that old monosynth, which is probably gathering dust over there in the corner, to do all sorts of terrible and wonderful things to digital sound sources. I am sure that many of you were raised on 'impossible to tune, one note at a time' monosynths, like those used by Rick Wakeman, Jan Hammer (way before Miami Vice) and Keith Emerson, so it should be interesting to blow off the dust and rediscover it! Let's look at the basic scheme of how you might integrate a VCF into a digital set-up.

First, you need some means of transferring the pitch information from your digital sound source to the monosynth. Once you have the pitch information inside the monosynth you can feed the digital sound source into the VCF's external audio input and use the tracking and frequency controls to alter the amount of filtering you apply to the FM sound. The 'buzzing with hiss' sound we listened to above is mostly way above the pitch of the sound we want, and so we can remove it with a filter - enter the VCF, stage left.

Voltage Controlled (low pass) Filters are very good at removing high frequencies and leaving the low frequencies behind. By setting the VCF's cut-off frequency on the low side, and after a bit of experimentation, you can make a passable attempt at removing most of that 'grungy' digital noise (Figure 2). You can also use this arrangement to produce more interesting effects, as you'll discover...

SERENDIPITY

(Definition: Serendipity — invention by accident.) OK - let's introduce a bit of serendipity and try using ordinary patches or voices on the digital sound source and fooling about with the VCF controls like we used to in the good old days of 'one control knob = one parameter' (remember the Oberheim 8-voice synths?). Apart from embarrassing hiccups from the monosynth when you forget and try to play chords, you will probably find that the really crisp, thin and digital sounds don't gain much from being filtered. But the dirty, slushy sounds suddenly take on 'that analogue sound', especially if you arrange to overdrive your monosynth's VCF a little bit... Contrasting envelopes on the digital sound source and the monosynth can also give you sounds which will have you wondering where you got the ESQ-1 or PPG! You could always add in a bit of the monosynth's own oscillators for the best (or worst) of both worlds!

To bring you back to earth, let's have a closer look at some of the practicalities involved in this filtering. (See Figure 3.) The obvious monosynth to use is an OSCar - since it already has a MIDI input (although I can't remember if it has an external audio input... but I am sure any decent engineer could fit one). For most other monosynths. you are going to have to use a MIDI-to-CV convertor (Figure 4). These come in all sorts of shapes, sizes and prices: from cheap do-it-yourself kits featured in electronics magazines (Electronics Today International did one in March I think, but I seem to recall it had some problems with MIDI channel choice) to more expensive commercial products like the Roland MPU101 and the very similar Vesta-Fire/Vesta-Kozo Convertor (do they make the Vesta meals as well?), the Tantek Tanrak MIDIverter, and the UMI MIDI-CV convertors. But think of the fun you will have with a MIDI'd monosynth for just ordinary sequences, let alone filtering your digital sound sources. I am sure that you will find a new lease of life for that forgotten old friend - long live analogue (and digital)!

Particularly interesting are bass lines, where an analogue monosynth (interfaced with a MIDI-to-CV convertor) tracks a mono line played on a DX or a sampler - careful choice of similar sounds and suitable panning can give superb stereo sound fields from the resulting composite instrument. This sort of double-tracking effect works very well when used with similar sounds that are slightly detuned with slightly different envelopes, and is normally only found on those albums where groups with time (and money) to waste spend a few years in a studio, honing expensive sounds.

You will obviously have to be very careful about only sending monophonic MIDI data to the analogue monosynth - it will probably be very difficult to arrange for the digital sound source to interface to the correct channels in an analogue polysynth (Roland 106, JX8P/3P, Akai AX80 etc) with MIDI, and this is not recommended. In any case, I know of very few analogue polysynths with external inputs to the VCFs. Some polysynths don't even have a VCF for each voice - the Korg Poly 800 being the classic example (although hands up all those at the back who remember the EMS Polysynthi?!). Of course, the 'coughing' noise you can get from a confused monosynth trying to track polyphonic MIDI data can be useful, if used in the right context...

If you don't have a MIDI monosynth like the OSCar and can't afford or don't want to buy a MIDI-to-CV convertor, then there are some mods available to MIDIfy some popular monosynths - the Sequential Pro-One should be very easy to convert, since the keyboard scanning is done digitally inside anyway; and someone somewhere must surely do a Minimoog conversion kit? If you have a MIDI'd acoustic piano, you could try filtering the sound of that as well - if you can stand the sacrilege!

By using a wild, untamed imagination and a bottomless wallet, it is possible to treat almost any audio source in a similar way. You need a pitch tracker - Fairlight make one of the better ones (called the VoiceTracker) and there are several other designs of varying performance - to produce either a CV (control voltage) or a MIDI output from your bagpipes, saxophone, voice or Tibetan nose flute, and then need to feed the CV or MIDI signal, as well as the sound signal itself, into your trusty monosynth. Most pitch convertors are best treated very gently - no blatant sax harmonics or wild string bending, and coughing into the vocal mic is definitely out! So this mostly rules out Miles Davis, Steve Hackett and Ronnie Laws sound-alikes. Nice smooth steady notes will ensure accurate following of your performance, so that the monosynth can ruin it afterwards! Guitar synths fitted with MIDI outputs could be used as well, although I am unsure of how the separate strings could be routed to separate VCFs, etc. You can probably only implement a Korg Poly 800 type of situation in this case, giving the familiar 'what sort of filtered chords do you want?' sound.

Modular synthesizers can also be used to effect, again by using a MIDI-to-CV convertor. In this case, you may be able to put together enough modules to try a polyphonic system. Note that this will give you effectively separate outputs for each note, ie. eight outputs for 8-note polyphony. You could treat these outputs separately with echo, reverb etc and place them at different positions in the stereo image. Playing the same chords would probably change the assignment of the notes to particular pitches, giving very complicated dynamic stereo sound changes.

Drummers - why not post-process your drum sounds, or use your drums to trigger a digital sound source and then process that sound? All you need is a Trigger-to-MIDI convertor like the Simmons MTI, or a MIDI drumkit like the Yamaha PMC-1 or Simmons SDS9 or MTM. Once your drumming is MIDIfied, the process is the same as before: use the digital sound source for the raw sound and use the analogue MIDI synth's VCF to do the dirty work!

WHERE'S THE CATCH?

Aha - there's always a catch! In this case it is well hidden - the MIDI-to-CV conversion takes a certain amount of time to accomplish, and so you may find that the attack of your sounds is either being missed altogether or is being considerably diminished. Two solutions - either mix back in a bit of the original sound from the digital sound source, or time delay the sound from the digital sound source before it is fed into the monosynth. But remember to take account of this delay if you are using a sequencer, by advancing the time slightly. Drummers, whose timing sensitivity is normally very good, can be put off by this sort of time delay, so watch out for overly complicated and slow set-ups. Since we are talking creatively, how about using the time delay of the CV-to-MIDI convertor as part of the sound?

Free hint: how about using a MIDI processor to process the MIDI information whilst it is travelling from the digital sound source to the monosynth? The limits to this sort of thing are very difficult to find - all you need is the nerve to throw all caution to the wind! One of my favourite 'found by accident' DX7 voices sounds absolutely awful on its own, but in a mix, double-tracked with a string sound, it adds a wonderful reverby-distortion type effect which livens up the rather dry FM sound remarkably.

Synthesis has always been about using your imagination and bodging things together - I think you would be surprised to find that famous synthesists like Michael Boddicker, Jean-Michel Jarre, and Vangelis, use not only whatever the latest gear that happens to be on loan to them at the time, but also old faithful stuff like Yamaha CS80s, ARP Quadras and 2600s, even Moog Sonic Sixes (a very nasty piece of work!). Use your synthesizers and other sound sources to the full - don't fall into the trap of thinking that you need only today's gear for today's music...

This article was sponsored by the Let's Not Forget The Old Analogue Monosynth Society and the Serendipity In Music Recording Association, and is partially based on a conversation with X Series Owner's Club member C.P. Ridout at the 1986 British Music Fair. It is an expanded version of an original article published in the Yamaha X Series Owner's Club magazine, 'Feedback'. The required imagination is all yours, get out there and misuse MIDI!

ERRATA: In last month's Practically MIDI, the 5-pin MIDI sockets were incorrectly labelled on the p57/59 diagrams - the Ground connection should be Pin 2 not Pin 3. Also, in Fig 1 and Fig 3 there should be a 220 ohm resistor between the output of the 74LS05 and Pin 5 of the DIN socket.