How can I switch from text generation to audio generation?

[j.wischnat]

I had this issue initially.

When sending a payload, make sure to still include BOTH modalities (text AND audio).

This will make the AI respond with audio.

Good luck!