And thank you for the welcome. If the voltage swing, for a direct drive amplifier, is as high as the supply, then it goes into saturation (clipping). Normally the output stage has to stay within the linear region so as to have as little distortion as possible. While they can go rail-to-rail in theory, in practice it would be lower.
Still, by pure luck, I found with Google an article from TI, snaa046, which gives an example of solid-state amplifier and some numbers: for a 400V supply (class A), output has ~200Vpp and its supposed to be able to drive Stax type 1 headphones (whichever those may be, I have never had any in my hands). At any rate, that was the info I needed.
On the other hand, I see no series resistor to the output, so, even if the plates and stator are isolated inside the earcup, the wires might still present a problem if they somehow break (I have seen, many times, people feeling a sudden urge to get up, forgetting the headphones they were wearing, and this is just one example... :-) ). The resistor cannot carry high current since the impedance for a 50pF at 20kHz is ~63.7kOhms, so with the 200Vpp (from TI's appnote) there would be ~1.11mA RMS (thus ~78mW dissipated power). In theory, a 10kOhm resistor could be placed in series and the gains would be: no sacrificed bandwidth and protection against high-voltage signal (not to mention it would be the cheapest protection of output stage at shortcircuit).
Well, thank you for the answers and, if I'll have more questions, you'll know... :-)