Call the Nose — parameter playground

Every session parameter is exposed below. Edit, press Call to apply, or Apply live mid-call. Refresh resets to defaults.

Session parameters

Model (locked at connect)

Voice (locked at connect)

Persona preset loads the brief's knowledge into the prompt + suggests a voice

Instructions (persona prompt + knowledge)

Output modalities text-only = the text-first MVP mode

Output speed 0.25–1.5

Turn detection (VAD)

Eagerness (semantic)

interrupt_response (let me talk over the Nose)

create_response (auto-reply)

Input transcription (your side)

Noise reduction

Max output tokens number or “inf”

Nose greets first on connect