call_id: 67acf3f0-a1e0-4880-a6f6-4848e7e02414 User text: Hello. My name is Nikita. I have a problem with my lamp outside. It's flickering a lot Can you help me with that? 1) Vapi transcripted "It's flickering a lot" and "Can you help me with that?" as one sentence 2) Instead of "Hello. My name is Nikita. I have a problem with my lamp outside. It's flickering a lot", we received "Hello. My name is Nikita. I have a problem with my lamp outside. It's flicker"

Hey, we process the text after each ending part like commas, dot and etc to improve the latency. You can remove it from text formatting plan.

There is no comma here, and there shouldn’t be one — it’s a single cohesive phrase. And what about "It's flicker" instead of "It's flickering"?

Can you try a different voice from 11labs and see how it performs?

Could you please clarify how the assistant voice is related to this case?

Hey august, checking if you got chance to try out Sahil's suggestion?

No, cause i really don't understand how voice is related to transcribing problem 🫠

Hey Nikita, the difference you're seeing in that screenshot are the chunks of the sentence (as they're being chunked), that then get put together to form the full transcript. In terms of the issue with "It's flicker" instead of "It's flickering" is a transcriber problem. Please try using Deepgram Nova-3, it's the current state of the art. For the voice model, I believe Sahil was recommending that because it can make a difference in the overall quality of pronunciation.

It seems to work better with a different transcriber model

Hi, checking if this is resolved/solved for you?