1 min read

Voice Session Start

Overview

The Voice Session Start node creates an ephemeral OpenAI Realtime API session for voice-based interactions. The frontend uses the returned session token to establish a WebRTC connection for real-time audio streaming.

Category

Voicevoice/sessionStart

Settings

SettingTypeDefaultDescription
providerStringopenai_realtimeVoice provider
modelStringgpt-4o-realtime-previewAI model for voice processing
systemPromptStringSystem prompt for the voice agent
voiceStringalloyVoice selection (alloy, echo, fable, onyx, nova, shimmer)

Inputs & Outputs