OpenAI released prompt-to-voice for all API customers on October 16. The feature lets developers create a custom voice from a text description through the API or voice Playground, then reuse it in GPT-Live sessions. Creating the voice is free; speech generation uses regular GPT-Live session billing.
The same update added nine natural voices for GPT-Live: Nuri, Juni, and Flitz (German); Brise and Sillage (French); Nira (Hindi); Harema and Shida (Japanese); and Noeul (Korean). Native Decisions integration in GPT-Live also shipped on October 15, allowing an app to act while a user is speaking.
These API changes sit alongside the broader GPT-6 rollout that began for ChatGPT users on October 7. The voice tools expand the practical surface for developers building real-time audio experiences. Background on the content stack that incorporates such APIs appears in the AI content stack overview.
Key takeaways
- Prompt-to-voice released for all API customers on October 16; custom voice from text description is free to create.
- Nine new GPT-Live voices added across German, French, Hindi, Japanese, and Korean.
- Native Decisions integration in GPT-Live shipped October 15.
What the October 16 prompt-to-voice release changes
Developers can now generate a reusable voice from a short text prompt without supplying audio samples. The nine new voices expand language coverage for full-duplex sessions. The Decisions integration lets an application respond mid-conversation without waiting for a full turn. Primary details are in the OpenAI platform changelog for October 15–16.
What to watch next
Further GPT-Live voice and latency improvements are the logical next incremental updates. Enterprise admin controls for the newer ChatGPT models remain the gating item for larger deployments. The next OpenAI release notes will show whether additional languages or pricing adjustments follow.