Experimenters have Youtube videos of a tesla coil playing music through its spark gap, or multiple spark streamers. I'm not certain whether it's (a) one high frequency modulated by audio, or (b) direct production of entire range of audio frequencies.
I see simulations of conventional tesla coils where a power transformer (at mains frequency) triggers a transformer and capacitor loop to ring at a resonant high frequency. This configuration looks as though it is simplest to build and and operate.