Audio Tuning
Configure pre-roll, post-roll, speech cleanup, and low-latency capture in Vorn Voice.
Audio settings control how Vorn Voice captures and prepares audio before sending it to the transcription model.
Pre-roll
Default: 350ms (range: 0–1200ms)
Pre-roll captures a small buffer of audio from before you press the hotkey. This helps catch the first word if you start speaking immediately when pressing the key.
If the beginning of your speech is getting cut off, increase the pre-roll value. If you hear unwanted noise at the start of transcriptions, decrease it.
Post-roll
Default: 80ms (range: 0–1200ms)
Post-roll captures a small amount of audio after you release the hotkey. This prevents the last word from being clipped if you release the key slightly before you stop speaking.
If the end of your speech is getting cut off, increase the post-roll.
Speech cleanup
Default: Balanced
Speech cleanup processes the audio before transcription to reduce noise and silence.
| Mode | Description |
|---|---|
| Off | Raw recording sent to the model. Safest for quiet speech or very clean audio environments. |
| Balanced | Light rumble and silence removal. Recommended for most users. |
| Aggressive | More silence trimming and noise reduction. Best with a clear microphone in a noisy room. |
If you are getting empty transcriptions or the model seems to miss quiet speech, try switching to Off or Balanced.
Low-latency capture
Default: On
When enabled, Vorn Voice keeps a warm microphone stream running in the background. This means audio capture starts immediately when you press the hotkey, with no startup delay.
When disabled, the app spawns a new audio capture process each time you dictate. This uses slightly less power but adds a small delay before recording begins.
For push-to-talk usage, keeping low-latency capture enabled is recommended.