The reason this keeps getting reinvented is that the physics only half works. Sound disrupts the flame's boundary layer and separates it from the fuel, but it removes almost no heat. The fuel stays above its ignition temperature, so anything hot nearby relights it the second the speaker stops. DARPA demoed this in 2012 and it stalled there. Impressive on a small liquid fuel fire in a lab, very hard to scale to anything you'd actually want to put out.
I wonder if one could use a thermal camera (or some other, high speed sensor) and some kind of feedback loop to squelch it and keep it disrupted until it cools. Or conversely, manipulate the flame for more complete combustion (in industrial processes).
But granted, I have trouble seeing this work on a large scale without much more sophisticated sonic wave tech.