June 30, 2025
Yes, it's fairly common to have a voice agent say a specific phrase when you start a function call or other task that will take some time to return.
In Pipecat you typically do this by pushing a `TTSSpeakFrame`.
Example here[1] https://t.co/Qg7tdBqGSN
@kwindla I'm gonna dive deeper into this, but it seems to me that you could buy some time by playing a sound and telling the user that something is happening. Like ChatGPT's voice does. This would degrade the UX vs doing it async, but in some cases, it might improve the overall UX. WDTY?