Each realtime voice API lately — OpenAI Realtime, Gemini Reside, Hume — was once constructed for one person speaking to 1 AI. That breaks the instant a 3rd voice enters the room. Gross sales calls, school room debates, multi-agent workflows, staff brainstorms — all of them want voice infra that is aware of who is speaking, when to break, and the best way to let 3 audio system percentage a flip.
🌞 Solar is constructed for that:
• Multi-speaker turn-taking
• 10× the context window of ChatGPT Realtime and Gemini Reside
• Agent-aware barge-in (no longer simply VAD)
• Multi-agent in a single room — run two AIs towards every different on an actual audio channel
These days’s PH be offering:
• Reside playground — take a look at it to your browser, no Credit score Card → https://getsun.io
• Reside demo at https://demo.getsun.io
Two issues I might love your lend a hand with:
1. Inform me the place this breaks. We have now stress-tested ~20 multi-speaker apps; we would like yours to be #21.
2. What integrations would free up you? LiveKit, Day by day, Vonage, Twilio, customized WebRTC — drop a remark.
Large due to Anoop for looking. Satisfied to reply to anything else within the feedback lately. 🌞
