Is the velocity curve same with both instaces? Is there offset in reading midi values?

I think this was the difference.

Modartt support contacted me and asked me to do another test. This time I copied the same midi file, fxp and mfxp to both instances and exported the audio.

And they sound identical!

So it was my error. I thought I had the same settings everywhere but I didn't.

Thank you all for your inputs and sorry to waste your time

I Since the original post is getting older, you probably want to update the versions you use for pianoteq on each machine. Good luck.

Forgot to mention that both are running 8.4.0. (registered on windows, trial on rPi) (I can't edit the original post)

I did originally test with the same audio hardware (Volt 1) which had the same difference in audio.

I'm now using the "export to audio" function in Pianoteq for two reasons:
1st: it should eliminate all audio hardware differences 
2nd: it allows for easier comparison and I can let others hear what I'm hearing

Interesting thought. Luckily I know someone else with Pianoteq so I'll ask them to make the same audio file and compare.
For the ones playing along at home: V8.4.0 - Shigeru Kaway SK-EX Player - Render to audio - lossless wav - standard demo midi file

Thanks everyone for your replies. I'll now wait for some hints from support.

Although, it is possible Pianoteq renders differently on the different machines it seems more likely to be a missed setting. I might also try a different rendering option when creating the file. No doubt, you have already considered some of this so Pianoteq devs may be your last option and rest assured they will get to the bottom of it. Since the original post is getting older, you probably want to update the versions you use for pianoteq on each machine. Good luck.

You are right, only the Pianoteq settings should make a difference when rendered to file. Could there be some filter on the windows machine? Try
playing back both renderings on the same machine. Is there still a difference?

I've listened to both tracks on my windows machine through a Volt 1 and Beyer Dynamic headphones. And the difference in volume and brightness is obvious.

For testing I switched both machines to the same instrument preset without any changes (Shigeru Kaway SK-EX Player) . Also the volume setting in Pianoteq is the same.
The difference in the volume is also very surprising to me.

I'm a software dev myself, so I'm quite surprised by the difference in sound if the code is the same. But it could always be a library that behaves differently on the different architectures.

I searched the forums but couldn't find anything so I've mailed support for help.

Thanks for the reply levinite.

My assumption: If I use the "export to audio" function, all hardware is bypassed.

The obvious difference is that the level (volume) in the Windows version is much higher which I suspect makes it closer to digital clipping. Not excessively so to be highly distorted, but close enough to make the sound brighter.

Have you tried increasing the volume on the RPi to match, or (IMHO better) decreasing volume on Windows again to match?

Unfortunately my issue isn't solved yet.

I just tried a complete clean re-install of the raspberry pi 5.

My assumption: If I use the "export to audio" function, all hardware is bypassed.

I've exported the demo on both my windows machine and my rpi:
https://kruissen.com/Preset%20demo_rpi.wav
https://kruissen.com/Preset%20demo_win.wav

Both have an internal sample rate of 48KHz.

I think I'm going to ask for some support by modartt.

In my opinion they should sound identical as it is purely rendered by software. But they are quite different.
Because I'm going to be using both, this difference makes it unusable for me on the rpi.

So I've got the demo version running on an Raspberry Pi 5. Performance seems good, but it sounds muffled compared to my normal PC.

Soundcard on the rpi is a Pirate Audio headphone amp (hifiberry-dac) but I've also tried my UA Volt 1 I normally use on my PC with the same results.
Using the same pair of headphones for all tests.

Pianoteq version: pianoteq_stage_linux_trial_v813.7z

Some settings that might be relevant:

 <VALUE name="engine_rate" val="48000"/>
 <DEVICESETUP deviceType="ALSA" audioOutputDeviceName="snd_rpi_hifiberry_dac, HifiBerry DAC HiFi pcm5102a-hifi-0; Direct hardware device without any conversions"
                 audioInputDeviceName="" audioDeviceRate="48000.0" audioDeviceBufferSize="128"
                 forceStereo="0"/>
  </VALUE>

I must be doing something wrong.
Has anyone experienced something similar? Any ideas what to try next?

I've done an export (render to file) on both the rPi and my PC and the difference is noticeable in that file as well: blues demo rendering (first the PC then the rPI. Preset is the YC5 Basic)

Pianoteq sounds great on my Pi 5 via HDMI. Indeed, Organteq 2 might be workable (see my post on the Organteq forum). I suspect the new PiOS the the issue, there is a new audio driver and pulseaudio is no longer default. You can try setting pulseaudio using raspi-config.

Soundcard on the rpi is a Pirate Audio headphone amp (hifiberry-dac) but I've also tried my UA Volt 1 I normally use on my PC with the same results.
Using the same pair of headphones for all tests.

Pianoteq version: pianoteq_stage_linux_trial_v813.7z

Some settings that might be relevant:

 <VALUE name="engine_rate" val="48000"/>
 <DEVICESETUP deviceType="ALSA" audioOutputDeviceName="snd_rpi_hifiberry_dac, HifiBerry DAC HiFi pcm5102a-hifi-0; Direct hardware device without any conversions"
                 audioInputDeviceName="" audioDeviceRate="48000.0" audioDeviceBufferSize="128"
                 forceStereo="0"/>
  </VALUE>

I must be doing something wrong.
Has anyone experienced something similar? Any ideas what to try next?

I've done an export (render to file) on both the rPi and my PC and the difference is noticeable in that file as well: blues demo rendering (first the PC then the rPI. Preset is the YC5 Basic)