How the Oramics Machine Read a Pitch
Most of the Oramics machine was analogue: light through a drawn line, turned into a voltage. Pitch was the exception, and the only digital part of the instrument. Daphne Oram wrote each note as a number, in binary, on four strips of film, and the machine read the number off and held it until the next one arrived.
One strip per digit
Four strips of clear 35mm film ran side by side through the machine, and each one carried a single decimal digit of the frequency in hertz: thousands, hundreds, tens and units. Four light sensors watched each strip. A mark in front of one of them stood for 8, 4, 2 or 1, and the digit was whatever those marks added up to, which is binary-coded decimal. So the A above middle C, 440 cycles a second, was written as 0, 4, 4, 0, and what you actually drew on the tens strip was a mark at the 4 position.
This is a strange way to write music and Oram knew it. It is also the part of the machine she successfully patented, after the American patent office rejected everything else on the basis of earlier inventions. Controlling pitch digitally by reading marks off film was the one component nobody had built before.
The coding ran from 1966 until the winter of 1969 to 1970, when she re-wired the machine to read musical semitones instead, because writing every note as its frequency in hertz was exhausting. The schematics were lost in that re-wiring, which is why it is no longer recorded which of the four sensors on each strip stood for which value. The diagram below picks an order; the mechanism is right and the assignment is a guess.
The diagram
Step through the six stages with the arrow keys or the controls along the bottom. The last one lets you write your own tune: type a frequency, or nudge the digits, or click marks onto the strips directly, then add up to sixteen notes and run them through the machine's latch and delay. Turning the sound on will play what you have written.
Four pitch strips, four sensors each, and the frequency they add up to. Step 5 lets you misalign the marks and hear the result; step 6 lets you write a tune.
Open the diagramIt needs a wider screen than this one.
Why it tolerates a wobbly hand
The marks were placed by hand, which means they never lined up exactly across the four strips. That matters more than it sounds. If the hundreds digit arrives a few milliseconds before the tens digit, there is a moment when the machine has read half of one note and half of the previous one, and it will dutifully play the frequency that results.
Graham Wrench, the engineer who built much of the electronics, solved it with a delay. When the machine sees the first mark of a note it waits 10, 20 or 30 milliseconds before reading the rest, by which time the stragglers have arrived. With the film moving at about three inches a second, thirty milliseconds of waiting allows the marks to be roughly 2.3mm out of line. The circuit is still in the machine.
This is the general shape of the whole instrument. Oram had written in a notebook in 1961, more than a year before any of it existed, that additive and subtractive synthesis were "too mathematically pure & calculated", and that a freehand drawing method might produce better results precisely because of the inaccuracies inherent within it. The delay circuit is that argument turned into electronics: the machine is built to expect an imprecise drawing and to read it anyway.
The rest of the series
- The Oramics machine Who Oram was, how the machine came to exist, and what was actually new about it.
- Drawn sound on film before Oram Avraamov, Sholpo and McLaren, and the point at which Oram departs from them.
- Flying spots and photomultipliers The two components the wave scanner is built from.
- The wave scanner How painted glass becomes an audio waveform.
Common questions
How did the Oramics machine know what pitch to play?
Pitch was written on four strips of clear 35mm film running side by side. Each strip held one decimal digit of the frequency in hertz: thousands, hundreds, tens and units. Four light sensors read each strip, and a mark in front of a sensor stood for 8, 4, 2 or 1, so each digit was coded in binary. The A above middle C, 440 cycles a second, was written as 0, 4, 4, 0.
Was the Oramics machine digital or analogue?
Both, in different places. Pitch was the one digital part: a number read off film as binary-coded decimal and latched until the next note. Everything else, including loudness, vibrato, reverberation and the waveform itself, was analogue, read as continuously varying light through drawn lines and painted glass.
Why did the machine wait before reading a note?
Because the marks were placed by hand and never lined up exactly across the four strips. Graham Wrench added a circuit that waits 10, 20 or 30 milliseconds after it sees the first mark before reading the rest. With the film moving at about three inches a second, thirty milliseconds allows the marks to be roughly 2.3mm out of line and still be read as one note. Without it, a note whose digits arrive at slightly different moments is briefly read as the wrong frequency.
Why did Oram change the machine to semitones?
Writing every note as its frequency in hertz was exhausting. The frequency coding ran from 1966 until the winter of 1969 to 1970, when she re-wired the machine to read musical semitones instead. The original schematics were lost in that re-wiring, which is why it is no longer recorded which of the four sensors on each strip stood for which value.
Sources
Tom Richards, Oramics: Precedents, Technology and Influence, PhD thesis, Goldsmiths, University of London, 2018, pages 12 and 112. Graham Wrench described the delay circuit in an interview with Sound on Sound. Written for Drawing Sound.