Modes of Confusion
Chris Meyer attempts to unravel the mysteries of MIDI's various operating modes, and dispels the myth that some recent instruments have new ones built into them.
The various modes in which MIDI can operate should have been simple to understand; instead, they've plunged the music industry into disarray. We guide you through what's happened so far.
MIDI MODES - SOURCE of confusion since the birth of MIDI, and on the verge of becoming even more confusing today. Let's try to straighten out where they came from, how many there really are, what they are supposed to be, how they are used, and what manufacturers mean when they claim to have invented new modes - modes that may not really exist.
As we all know by now, MIDI allows for 16 channels (or, as I prefer to think of them, "pairs of hands"), each capable of carrying separate sets of musical information (note messages, program changes, performance wheel movements, and so on). The MIDI mode an instrument is in determines which pair of hands it "listens" to, and how.
In the very beginning (namely, "MIDI" before the MIDI 1.0 Specification), there were three modes - Omni, Poly, and Mono. They were fairly simply defined as follows:
Omni meant listen to all MIDI messages regardless of MIDI channel, and act as if all of those "hands" were trying to play the physical keyboard attached to the synth.
Poly meant listen to only one MIDI channel (ie. one pair of hands), but otherwise act like Omni.
Mono was the mode that looked forward to multitimbral applications, by suggesting that one synth voice be assigned per MIDI channel - ie. one channel and voice per "finger", as opposed to per "pair of hands".
These definitions actually served two purposes in life: as definitions of how to react to incoming MIDI data, and as names for three different three-byte messages to send via MIDI to put a slave machine into a particular mode - a seemingly reasonable thing for sequencers or master controllers to want to do.
Along with the specification of these modes was the rule that receiving a Mode Change command would also force the instrument to turn off all notes currently held on via MIDI, since changing, say, from Omni to Poly mode could cause a synth to miss subsequent Note Off messages (the Note On received in Omni mode could have come on a channel - pair of hands - now ignored in Poly mode).
An additional message - All Notes Off - was specified to do the same, without changing the receiver's mode - it was a sort of MIDI "kill" switch. Since no one knew in the beginning how reliable or unreliable MIDI hardware would be, it was invented to ease paranoia that missing Note Offs may be a fairly common thing for a synth to do.
To round out the rules, it was suggested that all MIDI devices power up in Omni mode, so that any user (or salesman) could turn on any two synths, hook them up via MIDI, and they would then work regardless of which channel they were on. (Another interesting provision was that the ability to receive any modulations such as pitch-bend or pressure was to be disabled on power up, for fear that different scalings of modulations - and just plain too much going on - would confuse the early MIDI user.)
Alas, several forces at work in the world today (and in the music industry in particular) conspired to throw a series of monkey wrenches into the whole affair. Briefly, these were as follows.
1) MIDI is a "specification", as opposed to a 'standard". This means that there is no professional organisation - no "MIDI Police" - to ensure total compliance with the spec (a set of rules that tight would be considered restriction of free trade, and therefore illegal). So, it is ultimately in the hands of individual manufacturers to comply with the specification as best as they see fit. The MIDI Manufacturers' Association (MMA) and the Japanese MIDI Standards Committee (JMSC) were eventually created to suggest the "best way" of doing things - and this is all they can do. But before then, manufacturers were not always fully aware of what others considered to be this 'best way".
2) The language gap between American, Japanese, and European manufacturers can mean unintentional misinterpretation of the "best way" laid down by the MMA and JMSC.
3) Manufacturers can be over-zealous in both the "protection" of users and the attempt to anticipate what in the world users are going to try to do.
4) Users tend to start out more stupid, and end up more clever, than anybody gives them credit for.
5) Engineers love to invent things. And marketing people love to market them, whether they fully understand them or not.
The first hole appeared when Yamaha and Sequential had different ideas of what "Mono" meant, with Yamaha leaning towards "monophonic" and Sequential towards "multitimbral". The result was an official redefinition of the MIDI mode messages in the MIDI 1.0 spec, so that there were (and are) four messages (Omni On, Omni Off, Poly, and Mono), with two messages (one each of Omni On or Off and Poly or Mono) required to define a mode - leaving four combinations, and therefore four modes. The translation of the old modes to the new is as follows: "Omni" = Mode 1, "Poly" = Mode 3, and 'Mono" = Mode 4, with a wildcard Mode 2 falling out of the combinations that no one I'm aware of uses. (Yamaha, as shown by their FB01, have finally come to understand what this "multitimbral" stuff was that Sequential were talking about.) The next series of holes came with some hardware and software manufacturers being paranoid about notes left hanging on over MIDI. Roland are perhaps most famous in this area for making its machines transmit an All Notes Off messages in sequence tracks whenever they detected that all keys were indeed up on the local keyboard. The venerable MSQ700 sequencer would even insert All Notes Off messages in sequence tracks whenever it detected that all Note Ons had been accounted for with matching Note Offs.
Many other sequencers would simply transmit an All Notes Off at the natural end of a sequence, to make sure everything ended up clean (once again, in case something got lost because of an overflown buffer or whatever).
Now, on the surface, protecting against things accidentally getting lost, either by software or hardware gremlins, all sounds innocent enough - commendable, even.
But Murphy's Laws tend to cut both ways. First off, when some older synths received an All Notes Off message, they chopped all lingering notes off to silence, as opposed to letting them go through their programmed decay and release stages - so, whenever you lifted your hands off certain keyboards, all the notes would suddenly get choked off on certain other slaves.
Manufacturers caught on to this one fast, and merely acted as if the proper Note Offs were finally received for the lingering notes - thereby allowing them to go through their natural release, and sounding less abrupt.
The second problem was trickier and ultimately more insidious. By strict definition of "Omni" mode, any All Notes Off message, regardless of MIDI channel, should be listened to. This caused problems when something such as an MSQ700 started inserting All Notes Off messages - all the notes being off on one MIDI channel caused the creation of an All Notes Off message for that MIDI channel, even though notes may be sustained on a different track and MIDI channel. Therefore, a receiver in Omni mode would end up chopping notes that were rightfully sustaining on that other channel.
Eventually, the MMA and JMSC issued a notice that All Notes Off should be ignored in any Omni state and that such a message should be sent only in an "emergency" condition - such as the user hitting Stop or some other "panic" button (bringing it in line with Meyer's Second Prime Axiom - A Machine Should Not Take An Action I Didn't Ask It To).
Yamaha's way of avoiding All Notes Off problems in the DX7 was to ignore them altogether - a perfectly valid approach, since in theory you'll never need them anyway (the MIDI specification is very strict on the subject of sending a matching Note Off for every Note On, and the MMA and JMSC rejected a motion as recently as Summer NAMM 1986 to abolish that rule).
So, some sequencer writers thought they would outsmart Yamaha by sending a Poly mode command instead - the DX7 did listen to that, and according to the rules receiving any mode-change command, also executed an All Notes Off. And, they reasoned, don't all synths really want to be in Poly mode anyway? Well, no they don't, not if you're trying to use multi-timbral synthesisers as slaves - they would much rather be in Mode 4 ("Mono" mode).
So, aside from all of the normal problems associated with the All Notes Off message, MIDI users now found sequencers mysteriously switching their synthesisers into Mode 3, when they knew they had just set them to Mode 4 - as good a proof of the Second Prime Axiom as any, if you ask me.
In short, some sequencers were making some synthesisers do things that neither the manufacturers nor the users wanted or intended them to do. And as users' demands became more sophisticated, it was quickly apparent that Omni mode (Mode 1) wasn't much use for any configuration of two synths or more - and sometimes not even for two (for example, if you're trying to get them to ignore each other without resorting to plugging and unplugging the MIDI cable). Some users were even so bold as to want their synths to remember the last mode they put them in, instead of resetting all the time on power-up.
The inevitable result of all this was several manufacturers throwing their hands up in the air and saying: "Fine - the MIDI mode is merely how we listen to MIDI; we're going to remember it just like another patch or performance parameter, and ignore it if we receive it - let the users decide it for themselves".
Which brings us to the present. Software and firmware engineers have got smarter, and learned more about what they can trick a synthesiser into doing.
"Split" mode - the ability to divide a synth into two smaller synths over MIDI and from the local keyboard - came early. Oberheim's Xpander soon followed, allowing three 'zones" in which to divide the available voices. Later came what's known as "dynamic allocation" - the trick of appearing to be as many as 16 polyphonic synths over MIDI by throwing the voices where they're asked for on demand ("OK, voice 2 - right now channel 3 needs you to look like patch 5!"). Most samplers are dynamic-assign beasts by nature, since it quite often takes more than one sample to cover a keyboard, and each sample can be thought of as a different patch in its own right. Ordinary synths learned this talent too, and eventually this ability came to be known as Multi mode. But this mode is not a new MIDI mode command. It's a definition of a different way to react to incoming MIDI data - namely, that an instrument be able to react to two or more channels of MIDI information polyphonically, and also be able to respond to patch changes, pitch wheels and so on individually for those different channels (at least, to the best it can).
Many machines now use "Multi" mode where they used to use "Mono" mode (Mode 4) before, since many applications suit using a synth or sampler as several polyphonic beasts as opposed to several monophonic ones.
However, this issue has been confused by several people referring to Multi mode as the "new" Mono mode or Mode 4, and even switching into Multi mode when they receive a Mode 4 command over MIDI. It may, just possibly, be what the user really wanted, but it's technically wrong.
Beyond that, we have some machines claiming to have new modes. The best example of this is perhaps the Akai X7000, which claims to have no fewer than five new MIDI modes: Omni on/Special mono, Omni off/Special mono, Multi program/Poly, Multi Program/Mono, and (phew!) Multi program/Special mono.
But again, these are not actually new MIDI mode command messages, merely different (and very flexible) keyboard/voice control modes - "polyphonic", "unison", and "split" are a few simpler ones.
The point of all the above is this. Try not to be confused by all of these supposed "new" modes. If you're looking at an advertisement or brochure, translate the words "MIDI mode" to either "keyboard mode" or "voice control mode", and judge them in that context alone.
In the meantime, unless you are trying specifically to set up some automated "master control" system (such as having a sequencer or mother keyboard automatically reset the modes on all your synths - some of which you won't even be able to reach, since MIDI mode commands do not exist for them yet), you may be wiser (and in the end, a much less frustrated person) to disable reception of them whenever possible.
Trust me. Reading about it and trying to understand it now is a lot less confusing (and less expensive) than tripping over it in a session. Or worse still, with your wallet hanging out open in a music shop.
Feature by Chris Meyer
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!