Settings

/demo/kpn/settings
voicefleet
Organization
Realtime voice backend

Select the AI model powering voice conversations

Voice persona

10 voices available for OpenAI Realtime 1.5

Voice activity detection

Barge-in and turn-taking behavior

0.6
0.1 — Sensitive1.0 — Strict

Lower = detects speech more aggressively. Higher = requires clearer speech.

400ms
100ms — Fast1500ms — Patient

How long after silence before the turn is considered complete.

150ms
0ms — Tight500ms — Buffered

Audio captured before speech detection. More preserves context.

Generation

Response creativity and variation

0.8
0.0 — Deterministic1.2 — Creative

Higher = more varied responses. Lower = more focused and predictable.

Active config
backendgpt-realtime-1.5voicemarintemperature0.8vad_threshold0.6silence_ms400prefix_ms150accent#00A651
Knowledge base
No documents. Add SOPs or scripts to inform the AI agent.