Settings → General
General settings covers your default transcription preferences, how exported subtitle lines are wrapped, and the app's color theme. Everything here saves automatically as you change it — there's no Save button.
Default Transcription Settings
- Default Language — pre-selects the source language on the Dashboard's transcription controls. Full Whisper language list, alphabetically sorted.
- Default Model — pre-selects the transcription model. Only shows models you've actually installed in Transcription Models; if you have none installed, it prompts you to install one there first.
Subtitle layout
Controls how exported/generated subtitle lines are wrapped — applied automatically whenever you export or save to SRT, not as a per-export choice:
| Line length | Short (≈32 chars), Medium (≈42, default), or Long (≈52 chars) per line. |
|---|---|
| Maximum lines per subtitle | Single line, or up to two lines (default). |
| Split long subtitles (off) | When text exceeds the line limit, split it into multiple timed entries instead of allowing extra lines. |
| Force Precise Timestamps (on, recommended) | Prevents subtitles from appearing too early or lingering during silence. |
| Voice Activity Detection (VAD) (off) | Filters non-speech segments before transcription. Can improve accuracy on videos with long silences, but may produce empty output on quiet/soft audio. Requires the Silero VAD model installed. |
| Reduce hallucinations (off) | Disables cross-chunk context to stop Whisper looping the same phrase. Recommended for music-heavy content; may slightly reduce accuracy for continuous speech. |
| Post-process subtitles (on) | Cleans raw Whisper output: removes speaker markers, fixes punctuation, deduplicates repeated lines, fixes timing overlaps. Turn off to inspect the raw model output. |
Appearance
Dark Mode toggles between light and dark theme, defaulting to your system preference on first launch.
FAQ
Where do I actually pick the language/model per video?
The Dashboard has its own Source Language and Model dropdowns for each transcription run — the ones here just set what's pre-selected there.
My VAD toggle isn't doing anything?
It requires the Silero VAD model to be installed separately in Transcription Models.