Very nice! Glad to see you got it working well after all.
I'm on the Tesla Coil side of the plasma speaker fence now, and things are MUCH MUCH more complex, heh. Eventually I'll have a video up. For now, I've made an audio modulator that works decently, but sound quality isn't great and volume is low. It isn't very easy to fully modulate a 1.5kW RF signal with a couple mW of audio. Current status is that I'm in the process of redesigning my SSTC driver with alternative ICs which eliminate the need for a 5V rail, and also finding a method for increasing the primary side reactance by at least double while keeping the DC resistance low. As for the actual modulator it is just a Class-D amplifier driver used as an interrupter.
Basically this is how my TC modulator works:
TC Driver: a schmitt trigger receives the E-field from the secondary and squares up the waveform which is then fed in to a UCC gate drive chip pair which then drives the half/full bridge which powers the primary. This way the primary is always driven at the resonant frequency of the secondary.
The audio modulator: is a SG3525 which varies the pulse width of the TTL signal applied to the enable pins of the gate drive chips which turns them on and off. Unfortunately you have to set a pulse frequency for the SG3525 that is <400KHz and it just varies the pulse width with the frequency staying locked at whatever value is set by RC components. The problem is that pulse frequency modulation is the optimal method of modulating a Tesla Coil, and not Pulse Width Modulation. If you could set pulse frequency to that of the audio frequency and lock pulse width at a desired value you would have a truly perfect Tesla Coil audio modulator giving you high fidelity and variable volume (pulse width) control. With PWM modulation the volume of the audio output is locked as the average power produced by the average pulse width, which is quite low.