The Human Touch
The machines that stole the musicians' feel from your music are the same machines that you can now use to restore it - once you know how. Travis Charbeneau investigates the subtleties of the human touch.
A sequencer can make or break your music - but how do you give a sequence a "human" touch as opposed to a mechanical handicap?
WHAT MAKES MUSIC sound "human" or "mechanical"? Where does unfeeling mechanical precision take over from desirable human accuracy? The "soul" in music is, in part at least, a product of the techniques we humans cobbled together in order to cope with the instruments we've invented down through the ages and our inability to cope with them as we would like. The trouble with electronic instruments is that they allow us to overcome many of these shortcomings and create sounds and music that are literally too perfect.
You can easily synthesise the sound of a flute, but simulating a human playing the flute is an entirely different matter. Similarly, unless your fingers are a yard long, it's impossible to simulate a guitar strum on a keyboard over the proper register of a guitar patch. The real-time control necessary for imitating wind instruments and guitars is missing from most synths and samplers, and this situation is partly responsible for the proliferation of new and expensive MIDI wind and guitar controllers. Be aware, also, that the breath controller which drives virtually all Yamaha DX-series synths will go a long way towards solving the problem of missing expression.
Ironically, the best way of "humanising" your electronic instruments, is with the aid of a computer. If you can't get your drum machine to stutter, flam or do a press roll because you can't play them or your model doesn't support those features, remember that the computer/sequencer combination liberates us from the constraints of keyboard technique and pre-determined features. Properly used, your sequencer can simulate the playing of instruments whose sounds were never meant to issue forth from a row of black-and-whites.
Currently, even using the best controller and sound-generating gear in the world, you are unlikely to achieve the nuance and clarity of a virtuoso performance unless you ore a virtuoso. Even then, people are likely to gripe about trivia like the graininess in your sample. However, as anyone who has played in a "real" band knows, ensemble work can hide a multitude of sins. As a hardened sinner, I'll begin with the sequencer itself.
TEMPO IS THE first "humanisable" aspect of music that the sequencer brings under complete control. Various real-time interfaces are available that will follow a human drummer in live performance, but let's take a look at tempo flexibility as it comes in many sequencers. How do you get a rhythm pattern to change with a "human" feel?
Assuming your sequencer supports a tempo track, globally or by individual track, you have all the tempo control any human group of players has. Better, in fact. Your drummer won't get excited and speed up, causing bitter recriminations about whose "job" it is to guide and keep time. And global tempo control is just fine. Unless you're into experimental music, the idea is to keep "the group" together. If your sequencer supports offsets for individual tracks, you can always simulate the bass player who needs sleep or the drummer who's anxious to get through a rehearsal before the pubs close.
After you've got your basic tracks laid down, say for a rock instrumental, open up the tempo window, listen to the piece and let your imagination in. Basic tempo: 120 beats per minute. But, it's a hot night at Club I' Go Go. The band starts at 120bpm, however, after the first eight bars, people are crowding onto the floor. On the eighth bar the drummer rolls into the second verse and, responding involuntarily to an adrenalin rush, the tempo goes up: bar 9: 122, bar 10: 123, bar 11: 125. But at 125 you smartly regain control and sizzle up to the chorus. Drummer does a nice, broken tom roll, and you immediately flip back to 120bpm. Then comes the guitar break: another abrupt jump up to 125bpm. As the guitarist burns to a climax, the tempo climbs just perceptibly up to 128. Third verse: 125. Then, something you could never get right at the real Club I' Go Go: a perfect ritard on the vamp: 123, 120, 117, 111, 106, to a perfect dead stop - right down to the reverb dying on the beat. Something that would have taken years to achieve playing in garages and clubs, tempo control perfectly in accord with musical content, can be accomplished by the thoughtful entry of a few digits.
The uses of tempo variations can be a lot more subtle or a lot more obvious. Try raising the tempo just a couple of bpm for each bar in a two-bar drum roll heading into the middle eight, then resuming the original tempo. The roll will build a head of steam "just like a real drummer". In the "obvious" department, picture a classical conductor threatening the orchestra with his baton as he pushes and pulls them 12 times in as many bars. Your variations may run over a very narrow range, maybe a change of one or two bpm from one bar to the next, but the effect can be surprisingly convincing.
For random variations in tempo, make a tempo track of some uneven length that has nothing to do with the structure of your song - say nine bars of narrow variation running 120bpm, 122bpm, 121bpm, 119bpm, 118bpm, 119bpm, 121 bpm, 123bpm, 122bpm. Loop it and forget it. Barely perceptible, these slight variations can help humanise a piece of music in a way deliberate changes of tempo will not.
THESE SAME OBSERVATIONS apply to velocity and MIDI volume (MIDI controller 7) assigns. Even if your keyboard puts out velocity information you are unlikely to achieve a convincing performance for your sampled string section. This is what the computer was built for - play the part in at a straight velocity value of 64. Then go to the event editor and lower the velocity on individual notes. Suddenly you've got a "performance".
Many sequencers (my Voyetra Sequencer Plus III included) allow you to crescendo or decrescendo velocities. Phrases or passages can thus build to dominate others, or give way to others. The truth is, they can fade in or fade out a part more convincingly than most "real" players. Try combining a little crescendo with the tempo build-up in the two-bar drum roll we talked about earlier. A tired pattern taken right off your machine can be sculpted in this way to fit your expression, even if you've never hit a drum in your life.
If all your gear supports controller number 7, you can go a long way towards an automated mix. At the very least, you can achieve "human" dynamics which may exceed, but still resemble, those of human players. In most of my early bands we were so alternately preoccupied and then thrilled at playing all the notes right that dynamics were forgotten - properly used dynamics are often the trademark of a seasoned player. Changes in dynamics, even in heads-down, no-nonsense, mindless rock 'n' roll, are essential parts of human performance. Again, after all the tracks are down, go to the mixer and "audition" bars for forte and piano (loud and soft, to you). Then go back and insert number 7s or velocity assigns until you arrive at the right feel.
WHAT ABOUT THE precision with which a musician plays - what your sequencer calls quantisation? For years musicians have striven to be tight (in one sense or another). Now (musical) tightness is possible at the push of a button. Usual result: mechanical music. Quantisation is obviously useful as a repair tool, perhaps to put a bassline back in time with a bass drum, but it can also be a great creative tool, particularly if you cultivate a sensibility for human laziness. Say you've got a nice, loose groove leading up to an instrumental break, try quantising the break to match both the bass and the drums. In this case, the abrupt move into tight sync from your carefully-cultivated looseness will sound truly impressive, just like "real" musicians who've suddenly tightened up.
But quantisation is like any powerful new tool: you'll probably tend to over-use it at first and finding an appropriate place for it in your creative palette is worth spending time over.
More subtly, offsets - which slip tracks or parts of tracks forward or back in relation to the others - can also help humanise a performance. My Sequencer Plus has two offset functions, one which offsets in playback only, allowing you to audition various degrees of slippage, and a permanent offset that will take any part of the track and "physically" slip notes so that they're permanently re-positioned after you've decided on the desired result.
"Aggressive" and "laid-back" are not normally considered adjectives for use with machines. But these two attitudes can be simulated with uncanny effect by a computer. Of course the chances are, that no-one will consciously notice that anything is happening. In all music there are a great many subconscious ingredients and the effort spent studying them and implementing these with your sequencer will pay off when it comes to "humanising" the machine's performance.
YOUR SEQUENCER MAY support various types of transposition. These can also be used to add variation and colour. Say you've got a single-line sax vamp, built on a repeating phrase - it can only repeat for a limited amount of time before it sounds unrealistic. Try adding a harmony line to a few bars of the single-line phrase. Try straight fifths for starters and experiment from there.
You may have a sequencer that supports harmonic inversion as well as harmonic transposition. Try telling your key signature window that the song is in C Dorian instead of plain old C-minor and your vamped sax can suddenly switch modes for the fade-out. Alternatively, phrase inversion can turn a phrase inside-out around a selected note axis for some nice surprises. If the facility doesn't exist on the sequencer it may be worth the effort of working the inversion out and inputting it as a separate part - isn't that what real musicians do?
IS ALL THIS key-punching, this microsurgery necessary for a few bars of music? Having served a 20-year apprenticeship in bands and writing music, I can confirm that this is no spontaneous kick-out-the-jams enterprise. It would be fair to say it's more closely related to the writing experience. In terms of artistic legitimacy, computer composition/performance is just as valid as writing music but is perhaps more akin to painting or making animated films. Both pursuits lack spontaneity and do not yield immediate results - but they do yield results and are perfectly valid approches to what you might loosely call artistic creativity.
The MIDI/computer encounter is also very much like that of taking a band into a recording studio. The bottom line in both cases is the final mix. I challenge the general record-buying public to tell the difference between a record made by a well-produced group using "conventional" studio practices and a sufficiently resourceful one-man MIDI band working from his back room.
Next time you're drooling over the advertisements and reviews in these pages, look seriously at your current rig. Have you really exploited it to the full? Look what The Beatles did with two guitars, bass and drums. There's a lot to be said for knowing your instrument and this applies as much, if not more, to hi-tech gear as to "conventional" instruments. Adding more and more little black boxes to your arsenal of sound sources is going to make it correspondingly more difficult to "humanise" your music.
On the other hand, the capacity to better simulate human technique is being built into a lot of newer gear. Some of it, the new R8 "Human Rhythm Composer" from Roland, for instance, employs "artificial intelligence" to do this. Instead of real AI (which is not yet out of the lab) I'm guessing these machines use variations on the types of algorithms found in algorithmic compositional software, where you set parameters and the algorithm simulates a controlled randomness. (See feature on computer composition elsewhere in this issue.) But improvements in sequencing is where the real action will be. The Alesis MMT8 dedicated sequencer already supports different types of quantisation to adjust note ons, note offs, note durations, changing the note ons without affecting durations and so on. In brief, more flexibility.
More significantly, and probably more cheaply, look for software upgrades for your present computer-based sequencer. The new Master Tracks Pro, for example (it wasn't the first, nor will it be the last), supports a more sophisticated approach to quantisation where you set a margin of error or a quantisation "window". Notes that fall off the click, but are within the margin of error, are left unquantised while any real howlers are re-positioned. This way, some of the original human inaccuracy in the performance is retained.
Part of the beauty of making music with MIDI and a computer is that the technology continually brings ever more sophisticated, creative tools. That said, I don't think we'll ever have a true "human" button - you'll still need to know how to apply a technique to an instrument, and how and where to apply it. But better "humanising" tools are almost certain to come along and will, no doubt, be made very welcome.
Humanising a garage band is no problem; it's true, there's nothing like a live jam involving live musicians and their abilities, limitations, moods and inspiration. Nobody is going to deny that, and long may it live. But the result of an inspired performance and a carefully engineered sequence is the same - an emotive piece of music on a piece of tape. Right now there are A&R men up and down the country turning down demo tapes they believe to be out of fashion because they're obviously made with the help of machines - the same A&R men who are signing bands using technology a little more subtly, claiming that music needs to be made by "live" musicians. What did an A&R man ever know?
The bottom line is that once the computer has done your bidding, it's your music. That old saying about computers, "garbage in, garbage out" easily and validly translates to "human in, human out".
Feature by Travis Charbeneau
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!