OK -- going out on a limb here. I have a Physics degree, but I haven't done any experiments with this particular apparatus at all.
It is plausible for light to have some audible effect, particularly if the amplification factor is high. The relevant physical process is the Photoelectric Effect. Very well known and characterized -- a single photon can excite and dislodge a single electron, which is then effectively free to move under the influence of electric fields.
This effect will happen whether or not there is gas in the tube. What you're thinking about with a gas-filled tube, as in a Geiger-Muller tube, is in fact something different -- in a Geiger counter, you are relying on the photon (or other radiation) to ionize an inert, otherwise insulating gas between plates at very high voltage, and when it does so it briefly creates a circuit path through the gas. What happens here is not that. Here the photon would strike the cathode directly and dislodge electrons to create a current, and once the electron is free it doesn't care if it's in a gas or in vacuum; vacuum is probably slightly better.
Every material has a "cutoff frequency," i.e., a minimum photon energy below which no electrons will be emitted. Metals, in general, require fairly low energy -- light in the UV or even visible light are often enough to do it. Light in the IR range, probably not. The metal in vacuum tubes is also (a) heated, and (b) doped with metals to encourage electron liberation, both to encourage thermionic emission which is how tubes work in the first place. I'm going to guess (don't know for sure) that these steps also make it easier for the photoelectric effect to take place.
If electrons are being shaken loose by the photoelectric effect, they will create a current, as the free electrons will then proceed directly to the anode. Our tube systems are forward biased, creating an electric field which will always push them in that direction, so they really have nowhere else to go. So there will be an actual signal. How strong, I don't know, but it's not much of a stretch to think it could be audible.
Supposing this is at the heart of it, there are other factors that might amplify the effect. Like amplifiers, for instance. Or really sensitive speakers. There could also be things like instability in an amplifier downstream that magnify the effect, but I can't see this upsetting any decent amplifier, particularly a zero-feedback design at a good operating point. It's also possible that a photoelectric effect could create enough current for other effects, like microphonics, to become audible, particularly if those other effects on their own are not strong enough to start current flowing, but together they can.
If you want to try it further, here's a couple of experiments:
1. Does the sound vary with light intensity? Try holding the light closer and farther away.
2. Is there a "color" below which there suddenly is no sound? LEDs are pretty monochromatic but glass filters and such sometimes let weird frequencies through.
3. If you create a narrow beam of light, does the effect change if you shine it on different parts of the tube? Shining it on the anode should do nothing at all, shining it on the grid shouldn't have much effect, but shining on the cathode should have the greatest effect.
4. (Advanced) Does the effect change with heater current?
And finally,
5. If you're an audio purist and worried about light screwing up your music (!), tubes with dark glass, reflective interior, or screens should be immune. Try mechanical ways of blocking the light.
In my own system, I keep it all in a wooden cabinet, and I rig up LED power indicators always using deep, deep red just because I like them... but maybe they sound better, too...
Sounds like a fun experiment. All I can tell you is that physically it's plausible. I have no idea how strong the effect is under normal conditions, or whether it should be audible without going to great lengths to make it happen.
