Skip to content
Neural Daily

ToolsNews

GPT-Live-1.1 brings new voices and better accents

OpenAI began rolling out GPT-Live-1.1 to voice in ChatGPT, Codex and dots on October 15. The update brings 13 new voices and better accents in several non-English languages; API access comes later.

By: Mikola Máté Patrik · · 1 min read

Illustration of layered sound-wave lines on a violet to teal gradient
Image: AI-generated illustration / Neural Daily

What's new

GPT-Live-1.1 updates OpenAI's real-time voice model. The announcement lists 13 new voices and considerably better performance in French, German, Hindi, Mexican Spanish, Korean, Japanese and other non-English languages (OpenAI Community).

OpenAI says delegation is more natural, steerability has improved and there is less 'checking…'. It is rolling out to all users; voices are picked under Settings, Voice in ChatGPT (OpenAI Help Center).

For developers

The model is not in the API yet; OpenAI says it is coming soon. A follow-up forum post explains that GPT-Live can call the Decisions API natively, so decisions can start while the user is still speaking, skipping a round trip through your app.

Early feedback is mixed: several forum users reported Standard Voice Mode issues such as wrong voices, interruptions and language-recognition errors, which OpenAI moved to a separate thread.

Key takeaways

  • 13 new voices and better accents in six highlighted languages.
  • Less waiting, better delegation and steerability.
  • Rolling out in ChatGPT, Codex and dots; API later.

    Produced with the help of AI, reviewed by an editor. How we make our articles

    Reviewed and edited by the Neural Daily desk. About

    Related articles