MIDI in the Mix
The scope of MIDI control has expanded beyond that of simply allowing one synthesiser to control another. Chris Meyer explains how you can now shape a whole mix using the five-pin DIN plug.
MIDI has expanded beyond the boundaries of playing notes and changing programs into the realm of studio automation - now you can use MIDI devices to help perfect your mix.
TIME TO UPSET a few people. Let's say you've just written your best song ever - perhaps it's the best song anyone's ever written. It's even recorded, either on multitrack tape or into a MIDI sequencer. Fine. You're only half done.
The debate over which is more important - the composition or instrumentation and mix - probably started when the first caveman found two logs that sounded different when he hit them. Nowadays there's talk of producers "playing" the recording studio like an instrument. The problem is, they're the ones doing it - not you, the musician. But MIDI has finally integrated the music and the mix so that you can now have control over both. Now the studio is an instrument.
OBVIOUS, BUT STILL ignored by some, is the fact that MIDI allows you to go back and change which instrument played which part. Would a sampled piano sound better than that FM Rhodes? But what about changing the orchestration during a song? With all the freedom that programmable synthesisers and MIDI sequencers allow, it's distressing to hear how many musicians limit themselves to one set of sounds throughout a song.
Admittedly, part of this is due to increased use of samplers (which often have only one good sound per disk) because of the time it takes to load a new sound during a song. If you're using synthesisers, try spicing up your arrangement by changing patches for the second chorus or for every other bar - the latter is an old technique called "hocketing". At last, you can justify the 3000 DX patches you've collected over the past three years.
On samplers, you can use alternating versions of a sound by changing just a few parameters of the basic sound - amplifier decay and filter cutoff are probably the easiest. Horn disks tend to come with several instruments of the same family on one disk - these are particular fun to hocket with. Changing whole drum kits or a single percussion sound inside a kit is just as effective.
Making program changes during a sequence often takes a little more than the anticipated amount of thought. Many samplers choke sustaining or releasing voices when you change sounds. Some machines also have a delay when a new sound is loaded. Placing the program change a clock beat or two before the first note to be played with the new voice gives the sound the maximum time to die away and gets the change over before the downbeat of the next bar.
You can simplify a tricky synthesiser orchestration by saving the patch data with the sequence itself. It's not unusual to have more patches for a synth than fit inside the machine's memory at one time. This would normally require careful notation of which patches to load to replay a particular sequence (something that you'll still have to do with sampler disks until hard disk systems become more common and less expensive). However, it's not unusual for today's sequencers to record system exclusive data - the stuff synth presets are made of (or can be stored as). Consider a synth's internal program memory to be a scratchpad, and keep your patch library in cartridges or librarian programs. Load the patches you use for a particular sequence into the internal memory. Once happy with your orchestration, save the patches used in a couple of blank bars at the start of your sequence (or as a "setup" sequence you play once into your synth before starting work on a particular song) and you've one less thing to worry about.
Many synths and samplers these days also transmit and receive the standardised MIDI master volume controller. Anything from a simple balance to a rudimentary mix can be created using this feature. More instruments receive it than transmit it, so check your owner's manuals. If you have one instrument that transmits it, you can "ride" its master volume knob to mix down all the other devices in your setup that receive it. Otherwise, you'll have to resort to one of the all-too-few MIDI performance boxes (Yamaha MCS2 and its like) that can send it, or use a mapper (Yamaha MEP4, Axxess Mapper and so on) that can translate something such as a mod wheel into master volume. If the worst comes to the worst, there's always hand-typing volume information into the sequencer (it's controller 07, for those with sequencers that let you directly access the MIDI data).
THE SONG AND orchestration are complete. The next step is playing with the processing of the sounds. It should be noted at this point that the MIDI automation of processing and mixdown we're going to be talking about from here on doesn't apply just to MIDI sequenced material. It's perfectly legitimate to use a MIDI sequencer synced to tape for changing details of the mix. MIDI has progressed beyond what notes get played.
Almost all reverbs and multiprocessors (except for the cheapest and, oddly enough, most expensive) have MIDI on them. These MIDI implementations typically cover program changing, saving patch information and perhaps changing the output volume.
Sound familiar? They're all the things we talked about in relation to creative orchestration of the sounds in a song. And the same principles apply to signal processing. Effects levels can be changed and presets remembered in the same way as instruments, though more mileage can be had from changing the treatments.
"MIDI has finally integrated the music and the mix so that you can now have control over both - now the studio is an instrument."
Just as some synths cut off the old sound and trigger the new sound when a program is changed, most reverbs mute and cut off the old signal, switch programs (the Lexicon PCM70 in particular takes a long time to do this), and start their new effect as if sound has just been fed to them. The best way to deal with this is to find a place where the instrument being effected has been silent for a few beats and slot the program change in there. If no such space exists, prematurely end the effect by fading it out with the master volume control, change it, and push it back up immediately. Some effects change fast enough that they can be placed right on the downbeat without any unpleasant side effects - you only really hear an effect like chorusing on the body of the sound, not the attack, so missing the very start isn't a problem. Experiment with different effects.
Here's an interesting trick to try with programmable delays. Delay an individual drum sound or the bassline (without letting any of the original through) to alter the "feel" of the rhythm section. Doing it for a break or chorus is an effective way of reviving interest in a flagging song, and is a hell of a lot easier to do with a DDL and MIDI than altering the line in the sequencer.
Delays or gated reverbs can also help play with the rhythm of a song. Stretching a gated reverb on a snare or tom from an eighth note to a quarter note appears to slow a song down; going to a sixteenth note tightens it up. Delays mixed with the original can be manipulated from a straight timed echo to a swing or triplet figure and back again. This will liven up an otherwise repetitive rhythm track. Making these changes as part of the MIDI sequence makes the treatment as much a part of the song as the notes.
To figure out what delays you need, dig out your calculator and invert the tempo so it becomes beats per minute to minutes per beat; multiply it by 60 (for seconds per beat); multiply by another 1000 (to get milliseconds). Now divide it by two for eighth beats, four for sixteenths and three for triplets.
Programmable equalisers are a relatively new phenomenon. An EQ change can work as well as a subtle (or, of course, drastic) program change for adding sonic variety. For example, two entirely different sets of equalisations and ambiences can be swapped every few verses - one restricted to the mid range with some distance to it, and the other right in your face and pumped to the ends of the frequency spectrum.
SO FAR WE'VE discussed fairly static changes in sound - similar to an instrument being able to play notes and change programs, but having no performance controllers - like mod and pitch wheel, aftertouch, and the like. Some newer signal processors (like the Lexicon PCM70, ART DR1 and Eventide UltraHarmonizer) are starting to allow real-time changing of their parameters via MIDI.
The most common use of dynamic MIDI in signal processing so far is harmonising - selecting what intervals are created based on what MIDI notes are played into it. Creating additional notes is more a part of the song than the mix, but what dynamic use of MIDI is doing is blurring the distinction between the two.
Other dynamic applications of MIDI might be changing depth of chorus or to have reverb room size changing with the pitch of the notes being played - say, smaller for the high notes; larger for lower ones. The ultimate goal of a dynamic MIDI implementation is to make any parameter accessible from the instrument front panel alterable over MIDI. A new range of effects and ideas from the sublime to the atomic suddenly appear.
WE MOVE FROM the creation and modification of sounds to the mixing of them. As with signal processors, there are both static (referred to as "snapshot") and dynamic MIDI level mixers. Snapshot mixers are mixing desks that have the ability to remember settings of levels, pan, EQ and so on, and have those settings recalled via MIDI program changes. This can be thought of in exactly the same way as recalling a reverb or EQ preset.
"The ultimate goal of a dynamic MIDI implementation is to make any parameter accessible from the instrument front panel alterable over MIDI."
The most basic use of a snapshot mixer is muting and unmuting audio channels. Inflexible? It's actually useful enough for outboard mute boxes (like JL Cooper's MIDIMute) to be in demand. At the very least, they can be used for cutting out tracks while they're unused. This gets rid of unwanted tape hiss and quiescent synthesiser hums. I was sceptical about how useful this actually was until I used one - it sounded like I was using far more expensive equipment than I actually was. Mutes are also useful for covering thumps from switching programs or punching in and out on tape. More artistic applications include muting instruments to either thin something out or do a full "breakdown" - again, easier than editing a sequence to remove an instrument for a few bars.
With the exception of the Yamaha DMP7, all dynamic MIDI mixers currently on the shelves are intended to be used as add-ons to existing mixers. In various incarnations, they are all VCAs (Voltage Controlled Attenuators) linked to MIDI. Typically they plug into the insert points of the desk and control the level of the signals before they reach the faders. Hence the "VCA boxes" become a set of ghost faders that control the mix of the song. Multipass mixing is just like multitrack recording or sequencing - it's often more convenient to be able to work on one line, get it perfect, and go on to another instead of trying to "play" it all live on the desk at every pass.
VCA boxes offer an advantage over using the master volume on the instruments in that they tend to operate a lot more smoothly than the instrument itself. Internal volume controls also may attenuate a voice before it gets to the noisiest part of the output stage, leaving a hiss behind even when they're supposedly right down.
The price of VCA boxes is largely governed by three things - how they are controlled, the quality of the VCAs, and how intelligent they are. Some packages allow just an alpha wheel for changing VCA levels. Others take the form of VCAs controlled by a computer program (running on a personal computer) and have graphic faders which you move with a mouse to perform the mix. If these offend your sensibilities, you'll be best off looking for an alternative fader (Yamaha MCS2, or the data slider on many master keyboards) to do your mix. Others (JL Cooper's MAGI and MixMate, MegaMix, Yamaha DMP7) have "fader boxes" that are a second set of faders for the VCAs (the Yamaha goes as far as to have motorised faders to show precisely where the VCAs think they are).
A direct relationship exists between the cost of an individual VCA and how much distortion and noise it adds to the signal - a critical listening test is very much in order before deciding which one to buy. They're just like any other piece of recording gear - forget the hype and weigh what your ears are telling you against the price.
Each audio channel (VCA) of one of these boxes is typically assigned to a MIDI continuous controller, all on the same or different MIDI channels. Each channel should be thought of as a separate instrument as far as MIDI recording goes. Recording one channel's data is like one instrument line - make recording passes until you get it right. Punch in and out of the middle of a move to make smaller corrections. However, when channel 2 comes up, the fun begins, just like an instrument pass, you don't want to lose what you did on channel 1. So, if you're using a normal MIDI music sequencer and a VCA box with no onboard intelligence beyond which MIDI controller goes to which VCA, you'll either have to put each channel on a separate track, or put each VCA on a different MIDI channel and be careful which channel you erase if you want to re-record a move.
To counter this, some boxes have a degree of intelligence built in. They'll take the MIDI information from the previous pass, merge it with a channel's move made with their local controller, and feed the mix back out to be recorded on a new sequencer track. The idea here is to keep alternating between two tracks on the sequencer - one playing back the previous pass, one recording the new pass. This has the advantage of backing up each pass but the disadvantage of you having to remember to ping-pong between tracks on successive overdubs.
Overdubbing an automated mix also differs from a musical line in that you'll often want to match the previous level of the ghost fader before starting to overdub to avoid sudden jumps in level. This is called "update mode" and the process of matching the previous level is called "nulling the fader". No current musical sequencer has this mode. Some packages (JL Cooper's SAM and MixMate, MegaMix) step all the way around the MIDI sequencer and take care of recording and updating themselves.
So how useful is automated mixing at this level? Let me put it this way: I recently mixed a series of radio plays that each had dialogue, music, sound effects, and unwanted noises (punch in/outs) recorded at full volume across seven tracks. I'd like to meet the person who thinks they can remember almost a half hour of continuous fades, mutes, pans, and EQs. I ended up using a JL Cooper's SAM and MAGI to automate the levels, mutes, and effects returns, leaving my hands free for the fairly menial tasks of panning and punching EQ in and out. I simply couldn't have managed a manual mix.
I really only scratched the surface of automated mixing; the revolutionary Yamaha DMP7 has brought the dynamic MIDI concept to a full mixing desk, with every parameter being potentially automated.
MIDI GAVE US power over musical compositions from its inception. Now it's giving us the same power over the mix. Sadly, very few people seem to have chosen to exploit these possibilities so far. Perhaps it's technological overload from all the options available (note how well preset devices such as the Alesis line are doing); perhaps it's too new a concept to have been fully assimilated just yet. I hope so.
Feature by Chris Meyer
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!