Assistant Silent Despite Generated Output in Logs
# support
s
Whenever I interact with the assistant and review the logs, it appears that certain parts of the output are not being pronounced. In the attached image, you can see that there is output generated, but the assistant doesn't say anything for those sections. https://cdn.discordapp.com/attachments/1313455106561413151/1313455106879918100/Captura_de_pantalla_2024-12-03_a_las_11.39.46.png?ex=675031e4&is=674ee064&hm=0e0b4a4929f8b9ffbe25cdd2eb86537eaa14f15fb7242468138c39b2b30b59ac&
v
@s3e I'm sorry to hear that you're experiencing issues with the output pronunciation. To assist you better, could you please provide me with more details about the specific parts that are not being pronounced? Are there any patterns to the sections that are being skipped, such as special characters, certain words, or formatting issues? Additionally, it would be helpful to know if this happens consistently or only occasionally.
You can continue this conversation by mentioning me in the message.
s
@User could you help me please? 🙏
s
Hey @s3e To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help me figure out what went wrong!
s
Call id: 7a6cf3a7-28f2-465d-bfc7-0d621ff3aae1 Timestamp: 11:27:46:876 Response expected: output pronounced by the agent Response I got: silence
@Shubham Bajaj here you have it, thanks
s
@s3e can you check the call id again, the call has been ended by the customer itself.
Call Recording: storage.vapi.ai/7a6cf3a7-28f2-465d-bfc7-0d621ff3aae1-1733221677108-93148229-43f2-41fd-813d-b1da902ddeeb-mono.wav
s
@Shubham Bajaj please check this image. The issue is not how ended the call. The thing is that it seems like although the LLM model generated an output, that output was never delivered to the client through voice https://cdn.discordapp.com/attachments/1313455106561413151/1316805412179410984/Captura_de_pantalla_2024-12-03_a_las_11.png?ex=675c621b&is=675b109b&hm=00692a4c916e2270d30d3f6631a465f600434e0d9a3bb7b8a5dc8c54bc4c50a5&
s
@s3e if you look into the logs the model response was not voiced out because then the transcriber reported the user input came(from lasT) and then if you look at the logs from start another request was initated which was then voiced out. https://cdn.discordapp.com/attachments/1313455106561413151/1316816742659526706/Screenshot_2024-12-12_at_10.48.00_PM.png?ex=675c6ca8&is=675b1b28&hm=eacb916f863943439c0964c8ae4b875b9631f13b4556d0c2d84fc6efebe43c5a&
@s3e do let me know if you still have doubts.
2 Views