Everybody is focused on the tubes... What about the diffuser with pyramid shaped lenses? the highly directional light projects a beam (flourescent lights are known for awfully harsh shadows) through the thin, lightweight 24x24" or 24x48" lens(diffuser). as the light is projected down a minimum of 8 feet the tiny deflection from sound hitting the diffuser is amplified. Then, at an interface of light and shadow on the floor, the moving projection becomes a brightness signal that has audio encoded in it. they could use a lens and a photodiode. Didn't exactly have 5000fps cameras falling off trees 40 years ago.

If you get a high frame rate video of a space, there's plenty of other objects that can act as visual transducers: https://people.csail.mit.edu/mrub/VisualMic/.

I as skeptical as the next guy when it comes to the claim that ambient audio modulates the fluorescent light. But I could sure use more details on what analysis was done to the photos. Sounds like there were several photos taken, say 100ms(?) apart. Then the photos were subtracted pixel-by-pixel to look at the variations in time (between photos) or spatially (neighboring pixels)? I still think a photo-diode AC-coupled to a high gain amplifier fed into an oscilloscope would be the more conclusive test. Do a audio frequency sweep from 100-10kHz. Both with the fluorescent light on, and then again with the light off (as a control, in case something else in the setup is sensitive to the audio).

There is also the problem that that impedance mismatch between the gas in the tube and the glass surrounding it is huge so very little power is coupled. Interesting things would be for the tube length/resonance to be closer to the sound frequency of interest (eg 200-2000Hz) and close to critically damped. I have no idea if that's the case.

The second question is whether one can simply modify the tube/ballast in a visually imperceptible way. For example, one could insert an IC in the ballast that modulates the power draw of the lamp dependent on the ambient coupled acoustic/vibration power (eg possibly using the lamp as a vibrational microphone). Then pulse modulating the power draw of the lamp at a relatively low noise part of the power spectrum and within the band of the main power transformer filters. The lamps easily draw 40-80W so there's lots of room to work, and there are already Power Factor Correction implementations that use this simple technique to make lamp's inductive loads "look resistive".

So what's notable is that either the ballast of the lamp could be tampered with or one of the bulbs. Either one could allow transmission out along power lines, if standard transformers and capacitors are used for power delivery. This is how home powerline ethernet works and it transmits Gb/s rather than the few kb/s needed to get voice out over the power lines.

Case in point. I saw a single Korean Samjung lamp interfere with AM radio stations (<1MHz) from 100 meters, by just clipping the power draw of a small fluorescent with a thyristor and oscillator at the ballast of a desk lamp to provide PFC so that thousands of lamps could be placed in huge open plan offices.

One way I could see this working is with old bulbs - when you’re near the threshold for it flickering, perhaps sound can modulate the light.

Or perhaps the contacts need to be realistically cruddy and intermittent, and the long glass tube acts as a sounding body to trigger an intermittent contact, encoding sound into light.

Would be interesting to do a FFT of pixels in the sequence of images and see if anything is in there.

Just because the brightness fluctuations aren’t noticeable doesn’t mean they aren’t there.