what is the best practices to have faster voices conversation?

[jianming]

I've built a simple voice agent using the latest Agent SDK and VoicePipeline with the default configurations. However, the voice conversation feels noticeably slow. Can you explain how the speech-to-text (STT), workflow, and text-to-speech (TTS) processes work behind the scenes, and offer suggestions on how to improve the speed or reduce latency in the pipeline?

Our case is via phone line (twilio).

[rm-openai]

cc @dkundel-openai - could be a good idea to write a doc and/or example about this?

[revmag]

In the tracing config, the Text to Speech starts after the whole POST request has been made and text has been retrieved, why does it happen if text is streamign and TTS is cofigured to output the partial text?

[github-actions]

This issue is stale because it has been open for 7 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 3 days since being marked as stale.