What is the difference between TTS and <speak>?

[K3]

Original Discord Post by kopernikuz | 2024-11-08 21:37:30

Sometimes, i want the agent say something very specific. Unfortunetly , the FAQ styled KB did not work all the time. I set the temp to 0.01 and added “Only reply to the user based on…” in its background. Then i saw the tag and send a text with it via blueprint wrapped with the speak tag, and that kind of worked fine. Its just in the chat history though, which is not ideal and i don’t know hot it will affect the RAG.

I first wanted to try the TTS function but i am on free plan for now. I was wondering, where the difference might be, and if i only get the audio file or is the agent also speak?

[K3]

Reply by k3kalinix | 2024-11-10 22:34:47

Hello ,

The TTS endpoint is a standalone endpoint that can help you generate speech for any text.

The tag is used when speaking to a character during a session and also generates a verbatim speech.

But because it is used during a session, it becomes part of the chat history and affects future responses.