For most people, the default "Focused voice" mode under "Voice processing" works best in Fluxer calls. It uses "RNNoise" noise suppression together with echo cancellation. Pick another mode only if your voice sounds wrong or you use other audio software.
What are the voice processing modes?
"Voice processing" in "Voice & video" settings has three modes.
- "Focused voice" is the default and is made for speech. It uses "RNNoise" whenever your device can run it, and switches to another option while it can't.
- "Direct input" sends your microphone's sound with no processing at all, which is best if you clean up your sound in other software first. The app suggests it when your microphone's name shows that it comes from software like NVIDIA Broadcast or Krisp, because processing the sound twice can make it pump or distort.
- "Custom" lets you choose noise suppression, echo cancellation and gain.
Which noise suppression options are there?
In "Custom", you can pick "None", "WebRTC", "Noise gate", "Speex", "RNNoise", "GTCRN" or "DeepFilterNet" under "Noise suppression", and the list only shows the ones your device can run. "WebRTC" is the noise suppression built into your browser or the desktop app.
"Noise gate" mutes your microphone while the sound stays below a certain level, but it doesn't remove noise while you talk. "Speex", "RNNoise", "GTCRN" and "DeepFilterNet" remove background noise while you talk. If the one you picked is still loading or can't start, your microphone uses another one for now, and the settings tell you which one it uses.
Why does noise suppression cut off my voice?
Stronger noise suppression can mistake quiet speech or the start of a word for noise. Try a lighter option such as "WebRTC" in "Custom", or "None", and use "Mic test" in the same settings to hear how each one sounds before you join a call.
If your voice cuts out between words when you use voice activity in "Custom", the "Activity threshold" may be too high, and you can turn on "Auto-set activity threshold" instead.
If none of this helps, you can report the problem with your voice diagnostics.