I've been testing my assistant with the web call functionality and everything works correctly, as you can see in this call

d24c445e-60a3-4ea3-8a49-64af21426a8b

. However, when I call the same assistant using a real phone the transcriber output and the model output seem completely garbled and I'm not sure why that is. Everything should be the same, but it's not. Here's an example call id for that

ed53ce6f-413e-42dc-b9b3-d28ffed2ca3f

. I expect it to work the same as the real call, but I'm just getting random stuff like this https://cdn.discordapp.com/attachments/1351305768657489993/1351305768955150427/Screenshot_2025-03-17_at_5.26.32_PM.png?ex=67d9e511&is=67d89391&hm=342bf0a2f50545b0b9ef904e175f017670a1488ba3d629ea632cd1694eae1228&

@neelmehta247 I've a similar issue. I wonder if they are related. Are you r audio transcript disjointed?

Yeah the transcript doesn’t line up with what the audio is saying

https://discord.com/channels/1211482211119796234/1351349443756036156 like this

@neelmehta247 @~Stu~ CoreAspect.Agency looking into this.

That seems to work, however we cannot use deepgram since I need multilingual support which only google seems to offer right now

@neelmehta247 @~Stu~ CoreAspect.Agency Hey guys, the issue is on our end, and the team is working on it. It will be fixed by the end of the day. Till then, to use

smartEndpointingPlan

, please switch to Daily Channel for your organization.