Discussions

Ask a Question
Back to All

Low latency

The realtime API is audio-only, but I’d like to send input via text (I believe that’s not possible at the moment).

Is there any best practice for sending partial words or sentences without making the speech sound choppy? I’m using OpenAI as the LLM for the interactive avatar.

What’s the recommended approach, and what is the expected latency that would be considered good for this type of interaction?