Overview

Tune STT, TTS, VAD, LLM, turn detection, interruption, and input collection. Configuration cascades from template to node.

Configure how Breeze Buddy hears, thinks, and speaks. Most tuning lives in a template’s configurations object; a subset can be overridden per node.

Pages in this section

• STT — Soniox, Deepgram, Sarvam, OpenAI, Google. Turn detection modes.
• TTS — ElevenLabs, Cartesia, Sarvam, Google. Voice presets and multilingual output.
• VAD — Silero parameters. Start and stop timings.
• LLM — Azure OpenAI, temperature, Langfuse tracing.
• Audio — background sound, noise filter, keyword filter.
• User idle — silence handling, prompts, and graceful termination.
• Input collection — multi-segment capture for phone numbers and structured data.

For the raw schema, see Configuration reference.