If you didn’t want to use a plugin, for whatever reason, then you could simulate the effect with a curve and feed that into a dynamic material instance parameter. Then, you would have to manually (that’s the lame part) map the peaks and troughs of the curve to the audio.
I suppose if you opened the graphical representation of the audio, you could mimic the waveform directly.
The other obvious downside is the necessity to make a curve for every piece of audio… I’ll let you decide whether that’s worth it or not!
I did something similar to emulate blinking on one of my characters. If you need more info, then I’d be happy to provide it!