Very interesting. This is, as the name suggests, a hybrid device, with regular antennas, not optical (on the next iteration - on this iteration the volume is optical) that quantises notes and feeds them to an onboard DSP based synth. (Forgive any inaccuracies - we covered quite a lot of ground over a few hours and I wasn't taking notes.)
The design is in response to a perceived demand for, essentially, a low cost synth with the keyboard removed and replaced with pitch and volume antennas. One of the aims of the prototyping process is to find the best way to determine note events, and this was the focus of much of our discussion.
But first the synth. (I'm not going to dwell on the antennas - they are currently the least developed part of the instrument and the known deficiencies with them have straight-forward solutions. Likewise the noisy power supply. The antennas are self-calibrating. Which is good.) Two oscillators, switchable between sine, square, sawtooth and variable width pulse, one of which can be detuned. (I would prefer both to be detunable. I recall reading that we perceive the average of two very close frequencies to be the note.) A ring mod with a sine oscillator. A phase sweep (not currently implemented), ADSR envelope shaping (Attack triggered my note-on event, Release by note-off) and a choice of quantisations - currently: chromatic, C Major, C Minor and atonal (i.e. Off.) but to include the other major and minor keys too. Also a simple delay (er, I think - hopefully Richard will be along soon to set me right.)
Note events are currently triggered by moving from one quantisation to the next, giving a device with the same functionality as running the back of your hand up and down a piano keyboard. Vibrato is not possible, but it does a wicked semitone trill, a fixed tremolo can be achieved with the ring-mod (digital, so clean) and a variable speed celeste can be achieved by detuning. The volume antenna also creates note events by transitions to and from silence and determines the maximum volume of a note.
A lot of this may sound familiar - eh, Fred? - and is an instance of the difference between what the customer [i]thinks[/i] he wants and what he actually needs. You may think you want your synth to be more like a theremin, but actually you need your theremin to be more like a synth. :-)
So, the next prototype will incorporate ideas from a device he designed previously - a voice to MIDI converter (sorry, can't find a link) which seems a good starting point - the theremin is very voice-like. I demonstrated how a rapid jump from one pitch to another without muting constitutes a note event, and how the volume loop can be used for envelope shaping. I also mentioned the general dislike amongst players of a fixed vibrato.
Had I thought of it at the time I would have explained that the choice of rate and extent of vibrato is an expressive device for the thereminist, dependant on the mood of music currently being played, so really can't be automated. It should be possible - subject to available processing power - to separate the small movements of a vibrato from the note being played and use them to modulate the pitch so that the top of the vibrato is in tune (I recently learned that we perceive the highest part of the vibrato at as the note being played, not the average pitch as above) whilst still recognising larger movements as note changes - and maybe even glissing nicely towards the new note during the current note's release. And - I just thought - you're going to want a pitch preview with a short releases to be sure of arriving at the right note at the right time.
The other idea I had was that it could play atonally and without the envelope shaping, but when you settle on a note (with or without vibrato) it gently nu