Hi Google AI team,
We are currently testing gemini-3.8-live for a production real-time voice agent.
The new Voice Design API released with Gemini 3.8 TTS works very well for our use case, especially because we need a consistent Spanish (Spain) voice.
However, it seems that custom voice_... IDs created through /v1beta/voices cannot currently be used with gemini-3.8-live.
Are Voice Design / custom voices planned for Gemini 3.8 Live?
If so, is there any estimated timeline or preview / allowlist we can request access to?
Our use case is a real-time conversational voice agent for business phone calls, so using the same custom voice directly in the native audio Live model would be extremely useful.
Thanks!