Although much more widely used than ever before, the ability of electronic instruments to produce numerous different sounds at the same time is still not always fully understood. Paul White explains the underlying concept of multitimbrality.
An all too brief decade ago, the first multitimbral synth was only just going into production, yet today virtually every synthesizer or expander module has some multitimbral capability. Put simply, multitimbral means the ability to play several different sounds at the same time, each different sound or musical 'part' being controlled by a different MIDI channel. If you wanted to create a mind's‑eye model of an 8‑part multitimbral synth module, you'd probably be thinking along the lines of eight separate synthesizer modules crammed into one box. Up to a point, this is an excellent analogy, but what you do have to remember is that these 'virtual' modules are not really separate, and in many ways, their operation is linked.
At the simplest level, they all operate from the same power supply, so you can't turn them on or off individually. Also, even though the sounds may appear to be generated separately, on the more affordable models, these sounds will be mixed to stereo inside the machine so there is no way of accessing them separately. If you're lucky enough to have a machine with separate outputs, then the individual 'virtual' synth modules can be patched into separate mixer channels just as though they were physically independent, but with newer machines quite commonly offering 16‑part multitimbrality, building a machine with 16 separate outputs, possibly stereo outputs at that, is rarely practical.
To help overcome the limitations imposed by having all the sounds pre‑mixed inside the module, the user invariably has control over the sound level and pan position of the individual parts, though on some of the cheaper modules, these parameters may only be accessible via MIDI from a sequencer or a suitable editing software package. In addition, simple effects such as chorus and reverb are often included even in very inexpensive modules, and though the same effect setting applies to each part, the effect level can be set by the user, allowing some creative freedom. Less basic synths and expanders may incorporate effects processing sections to rival those of stand‑alone multi‑effects units.
A more important consideration is that of polyphony, and this is one area where the 'virtual' modules really are linked. Polyphony is a term used to describe how many music notes may be played at the same time, and in the case of a multitimbral module a mental picture helps clarify the situation far better than mere description. Imagine that the musical notes the unit can produce are physical objects — say marbles — stored in a single communal shoe box. If the unit has 24‑note polyphony, then there are 24 marbles in the box. When one of the parts is required to play a note, it borrows one of the marbles when the note starts, and it keeps it until the note has ended at which point, it returns it to the communal shoe box. Clearly, this is fine — until all 24 notes are playing at once, because now there are no more marbles left in the box.
So, what happens if a 25th note is played while the other 24 notes are still being used? You might expect nothing at all to happen because the marble box is empty, but that would be musically disastrous. Instead, the usual mode of operation is to 'steal' a note from one of the already playing parts, and to help disguise the theft, it's normal for the system to steal whichever of the notes was used earliest.
Even this apparently fair system can cause problems, especially if your song starts with a long drone note, because in a note robbing situation, this is the one most likely to be taken — and that would leave an obvious hole in the music. For this reason, most modules use a system whereby a minimum number of notes are reserved for use by each of the parts, so that if robbing does take place, it is less likely to be noticeable. For example, if one part is playing a pad chord, and one of the notes in the chord is robbed just as it's dying away, that's less likely to be a problem than if the whole chord went. It is often possible for the user to change the number of 'sounds' reserved for each part so that important parts have more notes at their disposal.
Multitimbral synth modules have many advantages, though they need to be used with a sequencer to make the most of their considerable features. The main advantage is cost; it's obviously cheaper to buy a single multitimbral module than eight or 16 separate MIDI instruments. There's also a convenience advantage in that a single unit is compact and easy to wire up, and if it has a mixed stereo output, as most have, you can still get by with a relatively small mixer.
The newer breed of GM/GS compatible modules enable users to be able to exchange MIDI song files in the knowledge that the song will play back on any GM module with fairly appropriate sounds, regardless of the manufacturer, and equally important, the drum mapping will be consistent so you won't find that a snare drum on one machine plays back as a cow bell on another.
On a more practical level, just because a machine is 8‑part multitimbral, it doesn't mean you have to use all eight parts. It may be better to use two or three different modules in a composition to avoid the possibility of note robbing and to provide a wider variety of available tone colours. This is particularly true of earlier machines, some of which had very limited polyphony.
Voices And Tones
One of the problems facing anyone coming to terms with multitimbral synths is the manufacturer's jargon — and different manufacturers tend to use different jargon. It's fairly logical to call the various virtual synth modules Parts, because these equate fairly well to the different musical parts they may be called upon to play. But what do we call the sounds of the notes that these Parts play? Indeed, Sounds seems like a good word to me, but Roland call them 'Tones' — and life still isn't simple, because a Tone may be built up from more than one basic sound. Roland call their basic building blocks of sound 'Voices', and their quoted polyphony figure relates to Voices, not Tones. In other words, if you play a sound or Tone which is made up from two Voices layered together, that counts as two marbles out of the box, not one.