Voice-to-text dictation app with local (Nvidia Parakeet/Whisper) and cloud models (BYOK). Privacy-first and available cross-platform.
JavaScript aianthropiccross-platformgeminigroqlinuxmacosnvidiaopen-sourceopenaiparakeetspeech-to-texttranscribewhisperwindows
docs: drop TROUBLESHOOTING.md addition, moved into PR description instead
iSparsh61d995bb6434
diff --git a/TROUBLESHOOTING.md b/TROUBLESHOOTING.mdindex 54183b85c..b45e8eb90 100644--- a/TROUBLESHOOTING.md+++ b/TROUBLESHOOTING.md@@ -136,19 +136,6 @@ OpenWhispr tries clipboard methods in order: `wl-copy` (most reliable) → rende4. Restart OpenWhispr and start meeting transcription again5. No screen-share chooser is expected for Linux system audio; OpenWhispr captures the default sink monitor directly through PipeWire-### Linux Auto-Learn Correction (AT-SPI2) Issues--**Symptoms:** Auto-learn correction (Settings → Dictionary) never updates the custom dictionary after an edit, with no error shown, on Linux sessions other than GNOME.--**Fix:**--1. GTK/Gecko apps (Firefox, Claude Desktop, etc.) only register with AT-SPI2 when `org.gnome.desktop.interface toolkit-accessibility` is enabled — this is off by default on non-GNOME sessions (niri, Hyprland, sway, KDE). Enable it:- ```- gsettings set org.gnome.desktop.interface toolkit-accessibility true- ```-2. Restart the target app (Firefox, etc.) after enabling — the flag is read once at startup, not watched live.-3. This affects only apps that don't expose their accessibility tree; apps that do keep working as before, and unsupported apps simply skip correction learning silently rather than erroring.-### Meeting Transcription Issues**Symptoms:** Meeting detection not working, no transcription, audio not captured