Why MIDI Music Stinks
Just as MIDI has made it as easier to produce great music, so it has facilitated the production of worthless garbage. Paul D. Lehrman explains the problem and how to avoid it.
MIDI has become so pervasive in the world of music composition and production that it's hard to think of any composer who doesn't use it in one form or another. But there are musicians who eschew MIDI, and electronics in general, not out of any 'purist' ethic, but because they are convinced that MIDI-based sounds, no matter how good, cannot produce 'real music'.
For example, an article appeared recently in an American film journal by an experienced film composer, who angrily and categorically declared that no decent film score could ever come from a MIDI studio. Furthermore, he said, it wasn't any cheaper than using live musicians.
When I read this I tried, after picking myself up off the floor, to read between the lines of this extraordinary document. Obviously, the gentleman had had some horrible experiences with MIDI studios mangling his work and shooting his budgets through the roof. (Had he come to me, of course, his experience would have been quite different, but that's another story...) He also was under some basic misconceptions of what the MIDI composition process is all about (for example, you don't need copyists, and that saves a lot of money!)
However, after thinking it over for a while, and spending a few days listening critically to music on radio and television, I realised that although his position goes against the very essence of my being, I had to agree with him in part. I too find that the great proportion of MIDI music — and not just music for films — is predictable, mechanical, and boring.
But while he blames MIDI itself, I don't. To borrow a slogan from the American gun-nut lobby — another bunch that promotes the use of potentially deadly technology — "MIDI doesn't kill music, musicians kill music."
MIDI has made it possible, even easy, to create perfect music, precisely programmed and absolutely identical every time it's performed. But music from MIDI that sounds 'live', as if it's performed by human beings instead of machines, is a rare commodity. And it's only music which can communicate human values that listeners find worth listening to, especially in the long run.
Those of us who've been involved in MIDI development for a while have successfully accomplished some very important goals: making synthesizers sound better; designing software that improves creativity; bringing professional tools down to affordable price brackets; and coming up with elegant and cheap ways to link sound and pictures.
These have allowed MIDI to become a wonderfully democratising influence, encouraging many musicians who otherwise could not possibly afford to produce music that sounds so good, to try their hand at making records, performing, scoring images, or simply enjoying themselves with music. It has fuelled a tremendous technological revolution, has inspired countless of software and hardware designers and independent record labels, and created new genres of magazines, like this one.
But unfortunately MIDI has also inspired a high degree of sloth in both those who make the tools and those who use them. When everything sounds so good so easily, why should we have to work at all at making music? That's the trap that MIDI users have fallen into, and it's a trap we have to pull out of.
Although many folks think it's fashionable to complain that MIDI "isn't good enough" for doing 'real' music, in reality, despite the inevitable compromises that went into its adoption and the very real limitations against which some users have bumped, MIDI is an extremely well-designed descriptive language for musical performance.
Before electronics, all music was created by physical activity: singing, whistling, beating, scraping, blowing, sucking, rubbing, etc. MIDI, with its carefully defined but wide-ranging command set, happens to be an excellent modelling system for physical gestures. A musical sound starts, it grows and changes, it becomes something else, it stops. All of these actions are reflected in the MIDI command set.
So what's the problem? Because MIDI tools do so much right away, and bring so much power to even the poorest (in both senses of the word) musician, it's incredibly tempting to let the tools do all the work. There's an old joke about a guy who hears a certain machine will do half his work for him, so he buys two. When the tools do all the work, however, the musician has given up the responsibility of creating, and what comes out will not speak to other humans, but only to other tools.
Rather than take advantage of the new horizons MIDI technology has opened for them, too many musicians simply use it to go after the same old goals, only faster and cheaper. In the process, their horizons get even smaller, and the genres they work in become even more restricted.
The most commonly cited fault of MIDI composition is over-reliance on quantising. Quantising robs music of any rhythmic subtleties, and while it can often help compensate for a lack of technique, it has the secondary effect of eliminating much of what we perceive as musical phrasing. Phrasing is more than note durations and volumes; it's also small changes in timing. A note that's delayed slightly will sound emphasised, while notes that are played ahead of a beat tend to be de-emphasised. Quantising wrecks all of that.
Step-time entry is even worse, because durations are quantised as well as attacks. If you ever feel like taking a glorious melodic line by Mozart or Tchaikovsky and squeezing all the life out of it, enter it into a sequencer in step-time.
Another familiar bugbear is factory sounds. A representative of a synthesizer company once told me a horrifying but telling story: they would send the first production model of any new synth to a Famous Rock Star. This gentleman would listen to the factory patches, pick one, and write a number-one hit around it. Soon, that patch was everywhere, and you couldn't get away from it. Even users of other synths would fall all over themselves trying to emulate it. Good for the company, good for the star, devastating to anyone who values individuality and creativity.
Factory sounds are designed to show off an instrument in a music store, to grab your attention and wrench your gut. While these qualities can be useful, they are not all that you need for a composing or arranging project — subtlety, expressiveness, and ability to blend are also major parts of the equation. At the same time, over-reliance on factory sounds robs music of any of the composer's personality. Clarinetists all use basically the same instrument, but the clarinet is a sufficiently flexible tool that every clarinetist can have his or her own unique sound. MIDI synths are far more flexible, but if you use the same sound as everyone else, you've lost any hope of uniqueness.
Just as neurologists theorise that we normally use only a small part of our brains, most MIDI users take advantage of only a small part of the MIDI spec. How many of us have ventured beyond Note Ons and Offs, the occasional program change and sustain pedal, and maybe a little mod wheel here and pitchbend there? On a violin, knowing when a note starts and stops is only a small part of what that note is all about. So it should be with a MIDI note. But how many of us bother to use footpedals for anything besides volume, or to use breath controllers, joysticks, or aftertouch for timbral changes? For that matter, how many of us know how to programme our synths to respond to that kind of control?
Instead, many of us have been concentrating on finding 'the perfect sound', which we can just play on a key, and walk away from while the music unfolds by itself. But music is not what happens when you push a button: music is what you do with the sound after the button is pushed. Every note that a saxophone player plays is different: pitch, vibrato, tone, volume contour. How dare we play a few notes on a keyboard and call it a sax line! Note Ons just go 'Now!' They don't sing, or whistle, or draw a clump of hair across a string, or spit into a tube, or make a piece of cane vibrate.
"When everything sounds so good so easily, why should we have to work at all at making music? That's the trap that MIDI users have fallen into, and it's a trap we have to pull out of."
A fascinating sound will only get you so far. Stravinsky had a philosophy about strange sounds: if you repeat the same chord enough times, no matter how dissonant it is, it becomes a tonic, or tonal centre. But there's a corollary to that, which I came up with after sitting through too many concerts at a computer music conference in the early '80s, which I modestly call Lehrman's Law: No matter how fascinating and complex a sound is, if you repeat it enough times, it becomes boring.
Over-reliance on Note Ons means that music must inevitably become pattern-based, with lots of repetition, and no room for nuance or expression. Everything that isn't dance music ends up sounding like Philip Glass. It becomes non-physical, non-human music.
It takes a traditional musician years to master a single instrument, to develop the ears, the micro-muscular responses, and the aesthetic sensitivity to produce music with it that others want to hear. Think about the first time you picked up a flute or a trumpet — could you make any sound at all? The first time you pick up a new synth, it may sound great, but you are still a long way from using it to its full capabilities.
The MIDI studio is an instrument, and needs to be mastered, like any other instrument. We don't give ourselves enough time to learn its parts. Every synth, every patch on every synth has a playing technique. (Fortunately, we can alter the patch to suit techniques with which we're already comfortable, or we'd never get anything done.) A good patch should provide plenty of room for exploration and expression, but most of us would rather go for something quick and simple. When was the last time you saw an advertisement for canned patches that didn't say "Realistic! Useful! Just like on TV and hit records!", but instead said "Lots of Controllers!"?
Samplers are particularly seductive. A sampler records the sound of a specific instrument playing a specific note in a specific way. Many users think that's all it takes to capture that instrument. But to convincingly re-create a real instrument requires many samples, as well as ways to choose among them that make sense musically, and a playing (or sequencing) technique that reflects an understanding of how the instrument is played. Proper use of velocity, aftertouch, and mod wheel can effectively simulate a real violin performance. But if you don't have a firm grasp on what a violin is supposed to sound like in the first place, you won't fool anybody.
Another trap into which MIDI users fall is that, because the technology gives them such a rich vocabulary and so much control over so many aspects of the music, they believe they can handle not only the composing, but also the arranging, mixing, recording, and editing all by themselves, with equal facility. Some of us, let's face it, just can't deal with all that. MIDI is empowering stuff, but it's not magic. I've been mixing music for nearly 20 years, but when I have an important project to do, I still hire a mixing engineer.
I'm often amazed, when I work with students and other young musicians, that they can be so well-versed in certain aspects of production, and totally ignorant of others. I once borrowed an analogue tape recorder from a friend who is a very well-known up-and-coming MIDI programmer, who does very impressive work, and loves to relate how he has made his samplers do one impossible trick after another. When I asked him whether the deck was 1/4 or 1/2-track, he had no idea what I was talking about.
A lousy mix can be a dead giveaway of what would otherwise be an acceptable MIDI-produced track. In an acoustic session, an engineer would never spread out a piano so it takes up the entire stereo image, and then put a 5-piece horn section in mono, dead centre. But an inexperienced MIDI musician, equipped with a brass patch on a single-output synth and a stereo piano sample, might do just that.
On-board processing, while it can make a synth sound much better in the store, can get you into trouble in a mix. A half-dozen synths mixed together, each with their own peculiar reverb, will often result in a mishmash of acoustic spaces that have nothing to do with each other, sounding totally unreal and out of control.
The other side of the coin is that MIDI has made it too easy to sound good, at least on a superficial level. In the past, a killer piano sound required a great piano, expensive mics, a real room, and a competent engineer, but all you need today is a £300 box. So a major criterion by which we were once able to determine whether a piece of music was any good — ie. does it sound good? — is now useless, because everything sounds good.
This means the distinctions between good and bad music are much more subtle. The audience is challenged to listen more carefully, and most listeners aren't interested in making the effort. Instead, they've handed the responsibility of deciding what they should be listening to over to a group that is particularly ill-equipped to handle it: record company executives. These people aren't interested in what's good, they're interested in what sells. And the general level of artistic expression falls. Is this situation new? Of course not: bad music has always overwhelmed good, and art has always had to fight the forces of commerce. But in its own way, MIDI has exacerbated it.
Before you start feeling guilty that it's all your fault, remember that there are other forces involved here. Manufacturers who are lazy or frightened of implementing radically new features, who respond only to market pressures and don't lead users into new realms of creative expression, are just as responsible for stagnation in MIDI music.
Very few manufacturers are producing anything that challenges the status quo of MIDI. Yes, wonderful new products are coming out all the time, but few of them offer any interesting new ways to exploit MIDI, or if they do, they are difficult to find, because nobody knows how to support or sell them.
Taking an old synthesis technique and jazzing it up with a new name and some new patches is no excuse for innovative technology. A manufacturer who dips his toes into a new type of expressive controller and then bails out when it fails to turn a profit after six months is doing more harm than good — other potential innovators will look at the experience and say, "Well, it didn't work for them, why should it work for me?"
Here's an example: an important characteristic of string and wind instruments is that the speed and depth of vibrato are independent of each other. In the vast majority of digital synths, however, a patch's LFO speed is fixed. Even if you have a synth that lets you control speed and depth separately (like the Kurzweil 1000 series and Oberheim's Matrix series), chances are you've never tried to set up a patch that takes advantage of this. This isn't because you're dumb: it's because the manufacturer has buried it in the software, not bothered to explain it in the manual, and failed to provide any presets that illustrate it.
Another example: real singers and wind players can only produce one note at a time. There are a number of ways they can get from one note to the next: sliding, re-attacking, breathing, tonguing, etc. Every synth has a 'mono' mode, but how many of them actually behave differently when they are in this mode, and how many offer you a choice of articulations? A 'legato' controller is in the final stages of being written into the MIDI specification, but it says nothing about how instruments are to behave once they are in 'legato' mode — each manufacturer will have to take his own initiative on that.
In fact, there have been many useful additions to the MIDI spec in the last few years, and except for MIDI Time Code and MIDI Sample Dump, few of them have come into common use. Instruments that use general purpose controllers, or registered and non-registered parameters, are still quite rare. Controller matrices are finally appearing in synthesizers, samplers, and processors, but few of these devices include among their factory patches any with complex controller maps, which would not only be useful themselves, but would inspire users to roll their own.
"Users who feel pressured into always having the latest, hippest, noisemakers never have a chance to get past the surface of each new toy before they have to move on."
There is also an argument to be made that constant technological innovation actually harms the cause of MIDI music. Users who feel pressured into always having the latest, hippest, noisemakers never have a chance to get past the surface of each new toy before they have to move on. Today's synthesizers are really, truly, deep, and only a tiny percentage of users ever get to utilise more than a tiny percentage of their capabilities.
Older synths still have plenty of life. My students, brought up on D70s, Proteuses, and VFX-SDs, pooh-pooh the Yamaha TX rack in the school's studio, until they hear what can be done with multiple modules, aftertouch-controlled brightness, and a little chorussing. Then they want to know why we don't have two racks.
Sequencer manufacturers also rush to include new features, but they don't always consider carefully enough how they will be used. A program as complex as a modern computer sequencer has to achieve a delicate balance between comprehensiveness and accessibility, or else it stifles the user's creative flow instead of increasing it. If the features are presented in such a way that users are forced to deal with all of them at the same time, many will get turned off altogether.
There are still some features that could help enormously to achieve a sense of realism in sequenced music, and which could be implemented in an intuitive way, but have yet to show up at all, as far as I know. For example, while many sequencers let you slide a track in time, to make it lead or lag other tracks, I don't know of any that will let you progressively rush or drag over a period of time, which is something a real instrumentalist is called upon to do very often.
The exploitation of System Exclusive information for its musical abilities is also not as far developed in sequencers as it could be. Many sequencers allow SysEx to be spat out in a 'chunk' at the beginning of a sequence so as to re-programme a synth, or even a whole studio, but have few provisions for including small packets of data for timbre alteration within a sequence. Many synthesizers only reach their full expressive capabilities when controlled by SysEx. To exploit this, sequencers need to record SysEx, give the user a chance to edit it like text, and be able to send it back at pre-determined times like any other MIDI event.
Although I'm painting a gloomy picture, it's obvious that thousands of people are doing great music with MIDI in spite of these problems. There are plenty of things you can do right now to improve the quality of your music, and move it away from the realm of the, as our film composer friend put it, "mechanised, quantified, [and] predigested".
• Master your instrument. Take time to learn what your studio can do, and get good at doing it. Try to avoid spending all your time looking for the latest and coolest hardware, but work with what you've got. As one successful but frustrated composer told me, "I have all the tools I ever wanted. Now I wish they'd just leave me alone and let me make music!" Maybe you can get the complex sound you want by layering two instruments, or using processing, instead of searching for that one synth that you think will do it all for you. The principle to keep in mind is "Get what you like", not just "Like what you get."
• Here's a good exercise: try working with classical pieces, and see if you can make them sound convincing on your equipment. It doesn't matter if you don't have a string sample. Realism is only one goal — it can be just as valid to try to make it sound like something the composer might have liked if he were alive today. (I have a file of a Casio CZ101 playing 'Anitra's Dance' from Peer Gynt that I like to think would make Grieg smile.) You don't have to have a specific acoustic sound to make an orchestral point, but you can instead use something that fulfills the same function as that sound. For example, where an orchestral composer would use a viola section to fill in a chord, you might get the same effect with a vocal pad. It won't sound exactly the same, but it could end up fitting in just fine with the rest of the sounds you're using. Consider the type of texture it produces — a pad, a sting, a fall-off, a thunk — and then make it come alive in its own terms.
• Learn how to make music breathe. Sometimes it helps to think of each instrument as a human voice, and give each line the same kind of shape you would if you were singing it. Whether you're imitating an instrument or creating something entirely new, imagine how that instrument feels to play, and figure out how to translate that into finger and controller movements.
• Learn about orchestration. The principles of orchestrating for electronic instruments are largely the same as for acoustic instruments: different timbres should complement each other, and stay out of each other's way. Make sure each line has its own distinct spectral space. Keep things from getting muddy.
• Don't take what they give you for granted. If a patch is labelled 'String Thing', don't assume it will be a great solo violin. Try to judge what patches sound like in context with other tracks. The best horn sound I've ever found for a DX7 is called 'Moog-2A', and my favourite fill-in woodwind sound is 'MiniPad'.
• Learn about mixing, and how to use processors, reverbs, compressors, EQ, etc. Too many MIDI musicians think that just because they've figured out how to run 16 synths into their mixer without overloading it, they're mixing engineers. You don't have to know as much about mic technique as professional recording engineers do, but you still have to know how to mix if you don't want your music to sound amateurish. Mixing goes hand in hand with orchestration. Learn how to create spaces with processing, and how to keep elements distinct even as they blend.
• Try alternative controllers. Get a couple of drum pads, and try a MIDI guitar if you play guitar at all. Even if the tracking is slow and your chops went South years ago, you will be able to create note and chord combinations that are clumsy or impossible to do on a keyboard, and that will add to your catalogue of musical gestures. If you sing, spend some time with a pitch-to-MIDI convertor. If you play woodwinds — even just a penny whistle or a recorder — try a Casio horn, or one of the Yamaha or Akai wind controllers. You won't be able to make it behave perfectly, but that's why sequencers have editing capabilities. These can add levels of expressiveness to your music with a minimum of effort, and if you really get into them, you might find you've created a unique compositional voice for yourself. If you can devote the time needed to learn them, look at the really alternative controllers from Don Buchla, or alternative composition programs like M or Music Mouse. Try to expand your vocabulary as far as possible beyond the note-on/note-off totalitarianism of keyboards.
• Get out of the studio once in a while. Go out and listen to some 'real' music: not Milli Vanilli, but chamber music, jazz, folk, blues, or ethnic. Pay attention to how the performers use subtle changes in time, pitch, timbre, and vibrato to make their expressive points, and think about how you can adopt some of those techniques to your music.
• Collaborate. MIDI is making one-man bands out of many people, but frankly not all of us are up to the challenge. Even if you do have the best instrumental, orchestration, and engineering chops in the world, you can gain a lot of perspective by working — playing, programming, or mixing — with someone else who has a different point of view or set of experiences.
I hope I've shown that it's not MIDI that has proven itself inadequate: the problem is that too many of us have used it as an excuse to create bad music. But I believe in MIDI. I have spent the last eight years immersed in it, and it has changed my life in more ways than I can count. I intend to keep working with it for a long time. Needless to say, it drives me crazy to hear "It isn't good enough," or "It isn't pro," or "I want real music".
We, as users and developers, have not done all that we can to realise MIDI's potential. But it's not from lack of trying, or lack of desire. Unfortunately, we all (well, most of us) live in the real world, and are constrained by real-world economics. Manufacturers have to sell to fickle customers, compete in a highly fashion-conscious market, keep to production schedules and budgets, and constantly introduce new products so as not to appear to fall behind the competition. Users have to justify the costs of their equipment, meet client schedules, and feed, house, and clothe themselves and their families. Few of us on either side of the equation have unlimited time to experiment, to play, to continuously seek out new forms of creative expression.
But unless precisely that form of play is encouraged, and unless MIDI is convincingly shown to be the creative — not just commercial — force it can be, it will continue to be categorised, and vilified, as only capable of producing 'machine music'. And that will be a great loss to humans.
Paul D. Lehrman is a composer, consultant, author, teacher, and long-time contributor to Sound On Sound, who lives in Boston, USA. He currently sits on the executive Board of the MIDI Manufacturers Association.
Feature by Paul D. Lehrman
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!