When speaking into the voice AI platform, my audio is captured in the logs and is present in the recordings, yet it never appears in the transcriptions. It seems as though the audio is being removed from the conversation, resulting in the assistant not processing my speech. Consequently, I receive an erroneous response ("I'm sorry. I'm having a hard time hearing you. Are you still there?"). The conversation flowed normally in the logs, but it did not appear in transcriptions or in the messages. (And the Assistant Speeches were not added in the call) CallId: f5d93cd6-6a58-4cb5-bfcb-5982e5e4baf7. (00:26+)

I'd like to help troubleshoot this issue with missing transcriptions. Based on the information you've provided, I need to understand more about your specific setup to provide targeted solutions. Could you share which transcriber provider you're using (Deepgram, AssemblyAI, Talkscriber, etc.) and any specific configuration settings you've applied to your voice AI agent? Also, have you made any recent changes to your setup before this issue started occurring? Additionally, it would be helpful to know if you've already tried adjusting any endpointing settings or language configurations, as these are common causes for the symptoms you're experiencing. Sources: - [Troubleshooting missing transcriptions in voice AI platform](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1340235729204285510) - [Possible causes for missing transcriptions in voice AI platform](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1303981615344779325) - [Audio loss in conversation within voice AI platform](https://discord.com/channels/1211482211119796234/1211483291191083018/threads/1316042862630146058)

Hi @Shubham Bajaj what was the failure in the tool call request that caused this behavior?

As you can see in both of the screenshots, when the LLM generated the output, it was sent to the 11Labs voice provider, and then there was no audio which came back to us from the telephony side, resulting in voicing out these idle message. https://cdn.discordapp.com/attachments/1354839054671286403/1355866078202564820/Screenshot_2025-03-30_at_4.55.05_PM.png?ex=67ea7c2f&is=67e92aaf&hm=c6d71a508366613b6385df7d562d01d874af03038430806e489e1621f38718ed& https://cdn.discordapp.com/attachments/1354839054671286403/1355866078600757338/Screenshot_2025-03-30_at_4.56.26_PM.png?ex=67ea7c2f&is=67e92aaf&hm=94915642c0134d61a54ca2b0d6c670113dff8b7a77ffd80d6e4cf54a8a97f2c9& https://cdn.discordapp.com/attachments/1354839054671286403/1355866079020453909/Screenshot_2025-03-30_at_4.55.05_PM.png?ex=67ea7c30&is=67e92ab0&hm=b1de7fc7a70a9f7db19c598cff479fadf7df22e771848dd3dad40178f5787663& https://cdn.discordapp.com/attachments/1354839054671286403/1355866079460589701/Screenshot_2025-03-30_at_4.56.26_PM.png?ex=67ea7c30&is=67e92ab0&hm=cbcdcae7c33f7a65880c66f0a4c237a5848b160a3db7ad24444d860a221a3a5f&

Thanks @Shubham Bajaj . My follow up question is regarding how to recover from this error. Is there any way we can avoid this? Should a fallback mechanism help in these cases?

Sorry, we don't have any mechanism or process by which we can add some fallbacks in case telephony fails to deliver the audio.