FSK synthesis

I'm in hardware now! Well, firmware at least. I posted a little comment to the electro-music.com DIY synthesizer forum about using FSK (Frequency Shift Keying) as a synthesis method, and Scott Gravenhorst took the idea and implemented it as an 8-voice, dual oscillator polyphonic synthesizer-on-a-chip using an FPGA (Field Programmable Gatorade)! How cool. He has posted some audio.

I had developed a software version (VST, using SynthEdit) and tried out a few configurations, but I didn't expect to make it into hardware.

FSK is a technique from communications. Applied to synthesizers, what you have is

1) A master oscillator (e.g. a sawtooth) —the modulator— and a slave oscillator (e.g. a sine) —the carrier— synced to the master

2) One or more comparators on the master's sawtooth which select different frequencies for the slave oscillator

As you can see in the graphic, the result is a waveshape which has a total period the same as the master, but which is made of two (or more, potentially) segments which have higher frequency periods. (In the graphic, the two traces don't quite align.)

At the electro-music.com site I put up an audio sample to demonstrate the idea: at subsonic frequencies of the master you can hear the two tones alternating, but at audio frequencies of the master you get a dual formant sound (like sharp twin resonant peaks.) However, with FSK is much simpler to get this effect than, say, two very sharp-peaked band-pass filters.

FSK looks a little like FM however the use of a square wave on the modulator means it has much more direct and predictable sonic properties.