NoiseTorch

• We now display a UI even if the connection to PulseAudio fails
• We now handle the PulseAudio connection failing and re-establish without having to restart NoiseTorch. (e.g. when PulseAudio crashes)

• NoiseTorch now respects PULSE_RUNTIME_DIR and PULSE_COOKIE which should also make it work with more exotic PulseAudio setups.

• Display input sources without dynamic latency support as likely incompatible.

See Issue #30 . A pulseaudio patch is available that someone with a device affected by this would need to test.

This release is not pushed via the built-in updater, as existing users will already have noticed it not working if they're affected. Yes it is, i forgot that otherwise the updater will prompt to downgrade.

• We had some hard-coded paths. We now respect the users choice of paths set via $XDG_* environment variables.
• This also fixes an issue where for some people NoiseTorch failed to connect to PulseAudio because it was looking at the wrong paths.

We tell PulseAudio to use the sample rate that works best with rnnoise, which may improve your noise suppression

Fixes an issue where pulseaudio would sometimes crash when loading NoiseTorch.

This change makes NoiseTorch prompt for root privileges. More information on why this is necessary can be found in #11

Improve the UI when changing the Voice Activation Threshold slider, to inform the user that a reload is required.

• Fixed a bug in writing logs
• Fixed a bug in update check logic

Beta release!

• Slightly improved the UI
• Increased timeout between sending pulseaudio load-module commands to hopefully make pulse crash less.

This is the public release of NoiseTorch!

It has only been tested on one system so far. Let me know if there's any issues.