Last week I animated the Maico Chromalyzer, a brilliant device that I’d never seen before. It should have been far more popular. It was simple enough that a radio amateur accustomed to dealing with tubes could have built one in a few evenings from available parts.
The Kay Sonagraph was opposite in every way. It was thoroughly familiar, found in every acoustics lab. I’ve operated it dozens of times, and always wanted to model and animate it. The Sonagraph was huge and heavy and complex mechanically and electrically. Like everything in speech and hearing, the Sonagraph was developed at Bell Labs. Kay Electric bought it and commercialized it.
Operating the Sonagraph required two stages. First you would record the sound from microphone or other transducer. A huge heavy turntable directly under the drum carried a wide strip of magnetic tape around the edge. The turntable rotated at 25 rpm during recording, so the loop held about 2.5 seconds of sound.
Here Happystar is recording the familiar sentence Joe took father’s shoebench out, which was a sort of quick brown fox for phonemes. It doesn’t contain all phonemes by any means, but it holds a wide range of frequencies and formants.
The printing stage ran the turntable at 78 rpm, with a specially treated paper wrapped around the drum. The stylus was charged with a high voltage that varied with the filtered signal. As the recorded sound played over and over, a variable bandpass filter moved up with the stylus, so each height represented a different frequency band. The treated paper was burned by the discharge of voltage to the grounded drum. This burning was SMELLY and SMOKY, so the Sonagraph needed a well-ventilated room.
When done, the paper was unwrapped to reveal a spectrogram, with dark burned areas where each frequency band was strongest.
The Sonagraph was replaced by cleaner and more flexible digital devices starting in the ’80s, but remained in use longer for high frequency sounds like birdsong or ultrasonic recordings. Digital systems have strict frequency limits because the sound is discretely sampled. Faster sampling gets more difficult and more expensive. A fully analog machine like the Sonagraph has no intrinsic limits on either frequency or dynamic range.
Polistra's Mill 