I tried using protontricks to install sapi and messed with some .NET versions with no improved results.
this also effects voice attack 2
They have updated the voice menu a bit: There's now three language packs to choose from, but that seems to correspond to the windows language packs you would have to install.
Which one would be the correct one for the "Proton Voice Files" ?
They also seem to have switched their voice recognition provider as you can select "New" or "Legacy" voice recognizer model. But both don't work on Linux.
There's also a microphone check where you can test if your microphone is recognized and play back the recording - this works without any issues.
Still the voice recognition feature in this game does not work.
New Log file with up-to-date Proton Experimental:
steam-2176130.protonExperimental.log
8348.493:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 8
8348.495:0548:0574:trace:mscoree:wine_mono_assembly_preload_hook_v2_fn "System.Speech.resources, Version=4.0.0.0, Culture=en-US, PublicKeyToken=31bf3856ad364e35"
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 9
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 10
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 11
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 12
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 13
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 14
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 15
8348.495:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 16
8348.496:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 17
8348.496:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 18
8348.496:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 19
8348.496:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 20
[0000000000000574:] EXCEPTION handling: System.IO.FileNotFoundException: Could not load file or assembly 'Z:\home\joni\.local\share\Steam\steamapps\common\Proton - Experimental\files\share\wine\mono\wine-mono\lib\mono\gac\System.Speech\4.0.0.0__31bf3856ad364e35\en-US\System.Speech.resources.dll' or one of its dependencies.
8348.496:0548:0574:trace:unwind:RtlGrowFunctionTable 0000000004AA0B30, 21
When I look up the FileNotFoundException file it doesn't exist (the directory structure ends prematurely):
'Z:\home\joni\.local\share\Steam\steamapps\common\Proton - Experimental\files\share\wine\mono\wine-mono\lib\mono\gac\System.Speech\4.0.0.0__31bf3856ad364e35\en-US\System.Speech.resources.dll'
joni@linuxjoni04 ~> ls -l /home/joni/.local/share/Steam/steamapps/common/Proton\ -\ Experimental/files/share/wine/mono/wine-mono/lib/mono/gac/System.Speech/4.0.0.0__31bf3856ad364e35/
total 676
-rwxr-xr-x 1 joni wheel 690688 Jan 17 19:34 System.Speech.dll*
I'll probably have to try something like this when I have time:
https://gist.github.com/vsTerminus/bf4f0247d75b7c0b747ab04bb34a0999
@GamerKing122: Looks like they got VoiceAttack working with that guide.
It looks like they have introduced an alternative to their Windows-based voice recognition
and now offer some type of local AI model for voice recognition:
It will still complain that you haven't set up Windows voice recognition yet, which I think can be ignored?!
Unfortunately when you start a round and the game tries to initialize the AI voice recognition, the game freezes (pause & return to menu still works and also the ambience sounds continue playing) so it's not playable yet.
This is where it freezes:
New Proton Log with Proton Experimental:
I am able to get into the game, however it stays at "Building the grammer..." and framerate is just stuck at "1FPS". I am running Resonite in the background, however, even setting the niceness to -9 has no effect on the game. I am using the Cachy version of Proton with the game.
IT WORKS!!!!
I'm using Proton Experimental and suddenly AI voice recognition works?!
@joni999 Great news! Could you share a few details so others can reproduce it?
It works for me using Proton Experimental, Latest Game version, and Steam Launch arguments
OMP_NUM_THREADS=8 KMP_AFFINITY=disabled %command%
The full details from https://www.protondb.com/app/2176130
Add the startup command args of "OMP_NUM_THREADS=8 KMP_AFFINITY=disabled %command%" (replace the 8 with the number of CPU threads you want the game to use)
Without the startup command args, the game will throw (silently, shows in the logs onlly) an error when trying to startup the voice recognition module.
You can now Test the voice recognition in the settings and it should work correctly.
When trying to start a new game (with the Flexible AI selected) you will still see the warning about a "Compatible Language Pack not installed". If you click "I'll Fix It", the game will continue and the voice recognition system will work correctly
It works fine for me without any launch arguments but the voice recognition is quite a bit slower than on Windows. I'll have to check if it runs better with them included.
The game seems to launch an entire AI model with python or something. There's probably immense overhead.
For me the AI voice recognition has up to 3 seconds delay after letting go of the recording button.
Sometimes it also fails to register a command if I give another command without waiting a few seconds between voice commands.
The launch arguments do not help unfortunately.
Can anyone confirm if they experience it too or is it just me?
My Specs:
i7 13700k
RX 9070XT
proton experimentalx5 2026-09proton 9.0x1 2024-09proton 9.0-3x1 2024-09resources.dllx1 2026-01speech.dllx1 2026-01system.dllx1 2024-09
Compatibility Report
System Information
I confirm:
PROTON Logfile:
steam-2176130.log
This seems to be one of the main issues:
Symptoms
The speech recognition does not work.
In this game you can use your voice to give commands to airplanes, similar to real life (same as in the previous game of the series: #8109)
Unfortunately this does not work on Linux as it seems to use some windows components.
Unlike described in #8109, synthesized/artificial voices actually do work here.
Ingame you get additional warning / error messages prompted by the game like these:
Warning message before trying to start a round:

Error message popup upon starting a round:

Speech recognition test in the settings menu is unavailable / greyed out:

Reproduction
Big THANK YOU to everyone contributing to proton and wine!!