Last month OpenAI launched GPT-Live-1 for realtime voice.
I wanted the same interruptible conversation but running fully locally, so I built an open-source version of it.
Meet OpenGPT-Voice😌
The loop stays on my machine:
Silero VAD → Faster-Whisper → GLiNER/Laya → local LLM → Supertonic TTS → WebSocket
The harder (fun) part was the messy edges of real conversation:
- Barge-in cancels the active LLM stream and queued audio immediately.
- Playback-aware history: interrupt after 2 words of a 100-word reply and only those 2 words enter memory.
- Timers and external tasks can finish and speak without derailing the current turn.
- The router waits through pauses and self-corrections instead of calling the LLM on every silence.
Built for people who want the voice loop inspectable and replaceable: local-first experiments, offline use, and edge setups where you don’t want every utterance sent to a hosted realtime API.
Still a POC. It breaks in funny ways. I’m still fixing them (constantly) 💪
Demo + GitHub repo in the comments
This might be the future of education! 🤯
I decided to test the GPT-Live-1 API and built a real-time chess instructor
It uses Stockfish under the hood to always give you accurate moves
You can interrupt the conversation at any time, and it will always provide up-to-date guidance on how to play using its backend capabilities
It is a very nice use case for it!
Created a GPT-Live-1 voice app template with @expo
- Full-duplex audio + background calls on iOS
- Choose from all supported voices
- 6 Rive animations from Vercel AI Elements
- Expo SDK 57 + native Expo UI controls
Source below ↓
GPT-Live-1 from OpenAI is now available on Phaseo through Chat.
The new full-duplex voice model can listen and speak simultaneously, handle interruptions naturally, and delegate more complex reasoning and tool use to backend models while continuing the conversation.
I recreated the old UCB Chelsea theater where I used to do improv and set up GPT-Live-1 to run improv scenes with you. Astra gives you coaching notes afterward.
The TAM of this product consists of myself, but it was so fun to make it.
Try a scene: improv.ilyaaby.chatgpt.site/
Now we're talking!
Programmers and mathematicians are replaced, now it's time for radio hosts, podcasters, sales, receptionists...
Ok holy shit. I just implemented this into Backbone and it is a game changer.
I'll post a video walkthrough tomorrow.
Very impressed.
OpenAI launched GPT Live 1 in the API. Apps can now run voice chats that keep listening even while the agent is talking.
Would this make you actually talk to an app instead of typing?
Making an AI girlfriend app that costs $14.99/week now
GPT-Live-1 API just dropped.
Just imagine the possibilities. You can now create better, more engaging experiences for your clients.
Going to test it today, will report back.
🤖AI is moving beyond prompts and into real-time interaction.
🤔For AI × Web3, this opens another door to rethink how users interact with intelligent financial ecosystems.
#OpenAI #GPTLive #tokenizedstock #Robinhood #Bitcoin #AIBank
🌈AI is moving from answering questions to taking action.
💙The next generation of digital experiences is becoming conversational, responsive, and increasingly autonomous. #OpenAI #GPTLive #Chatgpt #GPT6 #AI
Full-duplex models are the future of voice AI, across both business and consumer use cases. Everyone building voice products should + will be building on top of them.
