Skip to content

Tuning

The Tuning tab adjusts how speech is cut into lines and how long subtitles stay up.

Speech detection

  • Speech Detection Threshold — lower picks up quiet voices; raise it (0.5–0.7) for videos with music or background noise under the voice.
  • Silence Threshold — how long the server waits after speech stops before sending a line; 0.3–0.4 s feels snappier than 0.5 s. It's most of the "pause detection" time in the latency breakdown.
  • Max Speech Duration — force-splits long monologues into several lines.
  • Speaker Separation (shown while Label speakers is on) — how different two voices must be to count as two people.

These are saved with profiles.

Subtitle timing

  • Extra Time On Screen and Minimum Display Time — how long each line stays up.

These are saved with display configs.