Skip to main content

OpenAI adds custom and preset voices for GPT-Live

17 NOVEMBER 2026·2 MIN READ·3 SOURCES

OpenAI says developers can shape a GPT-Live voice with a text prompt or choose from nine new international preset voices. The OpenAI Developers post announcing the presets was published on October 16, 2026.

Key takeaways · 4

  • 01

    The nine presets cover French, German, Hindi, Japanese and Korean, and OpenAI says they were created with native-speaking actors.

  • 02

    A voice description can specify accent, age presentation, pitch, timbre and pace, within a 2,000-character limit.

  • 03

    Prompt-generated voices work with GPT-Live, but not OpenAI’s Text to Speech, Realtime or Chat Completions APIs.

  • 04

    Organizations can create up to 100 prompt-generated voices; OpenAI says they are not guaranteed to be unique.

Two ways to shape a voice

Developers can describe a voice in text or select from nine new presets for GPT-Live.[1][2] OpenAI says the presets were created with native-speaking actors and cover French, German, Hindi, Japanese and Korean.[1] The available names are Nuri, Juni and Flitz for German; Brise and Sillage for French; Nira for Hindi; Harema and Shida for Japanese; and Noeul for Korean.[2]

Creating and reusing a custom voice

OpenAI says developers can create a prompt-generated voice from a text description through the API or voice Playground and reuse it in GPT-Live sessions.[2] The API documentation specifies the POST /v1/audio/voices endpoint with the type set to prompt.[2] Descriptions can set qualities such as accent, age presentation, pitch, timbre and pace, and the quoted limit is up to 2,000 characters.[2] OpenAI says creating the voice is free, while speech generation is charged at the regular GPT-Live session rate.[2]

Limits and compatibility

OpenAI caps prompt-generated voices at 100 per organization.[2] The feature works with GPT-Live, but not with OpenAI’s Text to Speech, Realtime or Chat Completions APIs.[2] OpenAI prohibits requests for children’s voices and tells developers not to imitate a named person, public figure or fictional character.[2] It also warns that generated voices are not guaranteed to be unique: other users may create similar voices or output.[2] Teams should check these limits and restrictions before designing a voice workflow around the feature.[2]

How it fits GPT-Live

OpenAI says GPT-Live can listen and speak at the same time, while delegating deeper reasoning and actions to paired models and tools.[3] Developers can use its system prompt to shape an agent’s tone, pace and conversational style.[3] In early evaluations, GPT-Live-1 cut interruptions during learners’ thinking pauses by almost 80% compared with earlier turn-based systems.[3] OpenAI said GPT-Live-1 was available in its API at $0.05 per minute for the front-end voice layer.[3] These details provide context for assessing voice customization alongside the system’s interaction design and pricing.[3]

OpenAI says developers can use the system prompt to shape an agent’s tone, pace and conversational style. Before adopting prompt-generated voices, check API compatibility, organizational limits and the restrictions on requested voices.

Why it matters
For teams

AI training for your team

kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.

How this developed

  1. 17 November 2026

    OpenAI adds custom and preset voices for GPT-Live

Sources

AI fluency, one session a day, built for your work.