Greetings! for some reason the TTS layer unexpectedly stopped working in the telephony service, the TTS model i am using is Cartesia Sonic 3 and the voice is Tessa, the agent works perfectly fine in preview but when i pushed it to the telephony service and added the dispatch rules and set it up correctly according to the docs, when i call it, it won't speak back at all and when i checked the session agentic insights to see what went wrong, all these errors came up which indicates there is an issue with the TTS layer not functioning, here are the errors:

-tts_node: APIError: no audio frames were pushed for text: This is Alex, could you connect me with the owner or the manager, please? (body=None, retryable=True)

-tts_request: APIError: no audio frames were pushed for text: This is Alex, could you connect me with the owner or the manager, please? (body=None, retryable=True)

-tts_request_run: APIConnectionError: Connection error. (body=None, retryable=True)

-Error in _tts_inference_task

-failed to synthesize speech, retrying in 2.0s

If this is working in your preview, perhaps it is an issue with your TTS key in production? The first thing I would check is that the key got deployed successfully. You could also try swapping out for the inference provider: https://docs.livekit.io/agents/models/tts/inference/cartesia/ (I believe Tessa is available via inference)

I'm using it on livekit's cloud directly, doesn't it already handle that on its own?, its not self hosted

What is your project ID? (it will start with

p_

). I may have made some assumptions about your setup.

p_5sq2bcz9mje

Hey, I am getting this same issue @crooked-accountant-63197 did you fix it?

Sorry @crooked-accountant-63197, I missed that you had provided the project ID - I see many successful sessions in your project, so I hope this issue is now resolved. @cuddly-cartoon-47334, what is your error? Might be easier to start a new thread and tag me there if it's not identical. (never mind, I see you already did 🙂 )