sound clipping when exporting to audio

I wonder if it is related to whether you're using the "Use High Quality Settings" on export.  Toggle it and test it between two different exports.  I honestly don't know what exactly the difference is between "Use Current Settings" and "Use High Quality Settings" but my guess is that the current settings will route in realtime through the selected audio device (and pick up any artefacts in the process), while the high quality will use the cpu only with no quality or polyphony compromises and won't render in realtime.

The clipping happens to me on both high quality settings and current settings, it also happens on a Demo version of pianoteq standard as well as on my licensed pianoteq stage.

I do have the limiter enabled on both. regarding OS volume. does changing the volume on my computer have any effect on the output audio file?
There is one point I haven't mentioned that might serve as a hint, and that is the fact that for some reason the output audio file has a higher volume than the pianoteq program that generated it, dunno if that is how it should be or not though.

What are you using to play back the audio file? I gather it's some software media player on the same computer...? Have you looked at the file in an audio editor to see if it's visibly clipped (i.e. consecutive samples at 0dB)? MP3 encoding can introduce some clipping if the signal goes over about -0.3 dBFS but you'd only need to lower it by that much to eliminate the problem and the reduction in loudness would not be noticeable. .WAV should be fine right up to 0dBFS unless Pianoteq's metering is not 'true peak'. I haven't yet tried exporting from the Standalone app.

Yes, I expected something like that. Exporting at -1dB should not be a problem, and you should not concern yourself with whether some listeners may use a player (or monitoring system) that mangles the playback. That's their problem. I can't comment on the suitability of ubuntu's default player, but just FYI, there's been some noise on the Interwebz lately about VLC delivering malware. You might want to look into it.

More important is whether the audio exceeds the loudness standards of the various streaming services, but a solo piano piece that hasn't been massively compressed is unlikely to exceed any of those standards no matter how high it's peaking.

Out of curiosity, I did a quick test exporting audio from the Standalone Pianoteq, and comparing it to audio rendered by the VSTi in Cakewalk, and the two were very close - enough to null significantly when one was phase-inverted. Interestingly, I found that Pianoteq does not play back exactly the same way twice, probably due to deliberately random elements in the modelling that are intended to help it feel more alive (or possibly unavoidable...?), so the renders from two different hosts are never going to be identical even with all other settings being equal. Incidentally I disabled the reverb and delay FX for testing, and the level was low enough that the limiter should not have been engaging.

P.S. Given the nearly non-existent metering in the Standalone Pianoteq, I would highly recommend running the VSTi in a DAW to get a better grip on levels and loudness for distribution purposes.