ARTICLE  


Instrument Controlled Synthesizers: Technology, Technique, and Music

by Mike Beigel

Reprinted from Sound Arts Journal
Parts 1 and 2, September and November 1980, Volumes 3, Numbers 9 and 10

AN INSTRUMENT YOU PLAY WITH AN INSTRUMENT

The relationship between a musician and a musical instrument is beautiful and subtle. Years are spent to learn musical techniques and musical material. In the best musicians, the instrument becomes an extension of the body's capacity for expression. Now that electronic modification of musical sound has become commonplace, it is time to look at musical electronics as an extension of musical instruments.

The newest additions to the musical electronic "arsenal," Instrument Controlled Synthesizers, merit special attention and an expansion of our concepts of musical expression. Instrument-controlled synthesizers are musical instruments that are activated by other musical instruments. In the same way that a guitar is the link between a musician's body and the outside world, an instrument-controlled synthesizer is "played" by the output of a musical instrument. This concept is vitally important if you wish to extract the maximum musical potential from these new instruments, or to effectively market them.

Many people think of musical effects as pushbutton boxes that simply "work" when used with a musical instrument. This conception, which is weak even when applied to musical effects, breaks down when applied to instrument-controlled synthesizers. It is almost like expecting a saxophone to just "work" the first time you pick it up. Even a kazoo is not that accommodating. A musician who understands that the instrument-controlled synthesizer is itself a musical instrument, with all the subtleties of any other musical instrument, will be willing to devote the time and attention to perfect a new technique in his musical language.

Much of the initial confusion about (and hence, resistance to) instrument- controlled synthesizers centers around the notion that you have to "play into" them. But you even have to "play into" a fuzz-tone, and you certainly have to "play into" your native musical instrument. That is one of the beautiful aspects of making music: the musician's skill in controlling the instrument, or basic musicianship. So I would ask you, if you have a notion that "playing into" any musical device is a drawback, alter your thinking into the much more useful notion of simply "playing" the device. As with other musical instruments, the identity and charm of instrument-controlled synthesizers will derive as much from their peculiarities and limitations as from their inherent virtues.


EXTRACTING CONTROL SIGNALS FROM MUSICAL SOUNDS

If you've read this far, you are probably curious about how one musical instrument can "play" another. It's a fascinating subject. The human ear can discern the many aspects and subtleties of musical sounds with great precision and in many dimensions. But how does an electronic "box" decipher the complex information in a musical signal, and turn it into control voltages and parameters for a synthesizer? Moreover, what kinds of control information can we hope to derive from a musical signal?

If we consider a single note played by a musical instrument, we can characterize it by a few different aspects: time length; loudness; pitch; tone color. Obviously, there are more aspects than these, and there are subtleties in defining each aspect. We have to come up with working definitions that can be translated into "sense-able" quantities: reduction of a musical quality to a number or a voltage. Let's look at the four aspects just mentioned.

Starting with a note's time length: The basic question is, "Is a note being played or not?" With some instruments, particularly wind instruments, the question is fairly easy. Wind instruments have a limited dynamic range and a fairly quick decay after a note is released. But with a guitar, the note fades gradually into inaudibility. When can we say it is "off"? Let's say if the signal goes below 1 millivolt we will call it "off." Well, wouldn't you just guess that in the process of dying out, the guitar note sometimes gets a little louder before it gets a little softer! So a note enable detector, as we call the note "on " or "of" sensing device, must be sophisticated enough to take into account the subtle amplitude variations in a musical signal without producing "false" outputs. To roll this all up into a definition: A note enable detector senses the amplitude of a musical signal and produces an on output if a note is audible and an off output if a note is not audible. This is a "digital" output signal.

The loudness of a note is also related to the amplitude of the musical signal. Loudness is a relative or "analog" quantity, and is approximately measured by the musical signals "amplitude envelope." The "amplitude envelope" is essentially what is read by a VU meter on a mixing console: a continuous indication of the "average" AC signal voltage. Instead of using a meter, we convert the average AC signal into a corresponding DC signal. The envelope voltage is then used as a control parameter relating to signal loudness.

There is another way to measure loudness. In a guitar, the loudness is determined by the magnitude with which a note is "struck." All of the information about loudness comes at the beginning of the struck note. We need a "transient detector" to determine when a note is struck, and something like a "peak follower” to determine how high the peak value is before it dies down. So we have two new pieces of information-gathering hardware about the note: the attach detector (a digital signal that records the moment of a new note's attack) and the peak follower which records the amplitude at the time of attack.

We might also assign the attack detector to the function performed by the enable detector, since it tells when a new note has been initiated. Actually it does function in conjunction with the enable detector. Imagine the case of slurred but accented notes in quick succession. The enable detector would indicate that only one note was played, because the signal amplitude never reached the "off" level. But the attack detector would record the beginning of each new note. Together, the attack and enable detectors accurately specify note time-lengths. The attack detector, peak follower and envelope follower specify the note's loudness dynamics.

Pitch, of course, is simple. You merely have to find the fundamental frequency of the note being played, and produce a control voltage relating to it. Unfortunately, the fundamental frequency is often masked by other overtones of the musical signal, attack transients, wind noise, or interfering signals from other notes. Other notes!! You mean you can only extract the pitch of one note at a time? Well, yes. With presently available technology, only one "pitch" can be extracted at a time, if a "single input" system is used. A single input system is a system that accepts the musical instrument input- whether the instrument is polyphonic or not-through one input jack. Instrument-controlled synthesizers which are polyphonic must essentially use a separate isolated "pickup" for each voice employed. These concepts and instruments (notably, specialized polyphonic guitar synthesizers) would greatly expand the scope of this article, so I will deal with only the theory and implementation of "single voice" synthesizers. We will define a pitch follower as a device which tracks a musical note's fundamental frequency and converts it to an appropriate control voltage. A pitch follower is not simple.

[Continued...]
Top· Page [1][2][3][4]


Home | Services | Project Profiles | Technologists | Business Terms | Product Development | Articles | Contact | Links