This is odd - when the Model Output is outputting a word in caps, it is set as an acronym to the Voice Input, so the voice reads it out e.g. "NEED to succeed" LLM output is sent to the voice input as "N.E.E.D. to succeed" @Sahil how can I ensure the Voice Input is sent the caps word as it is please? https://cdn.discordapp.com/attachments/1211484481848873020/1241767833877217380/Screenshot_2024-05-19_at_15.58.40.png?ex=664b65ea&is=664a146a&hm=6cf713d49017caec0bc7ee71592572e7d042ebf67510b8928d41a6b4948174a2&

Can you share me the call_id along with the timestamp.

Yes thank you Example 1 ID: 5cecb9a8-cc82-43ea-b6f1-03755c1c0d3a Timestamp: 13:36:38:766 Example 2 ID: 04ef4f55-39ab-4323-8493-27d24ab8e18f Timestamp: 13:15:04:603 13:15:06:089 In this same call there's a 2nd issue in this timestamp, here the model output actually split the word: 13:14:40:234

By timestamp, I mean time at which this issue occurred in audio recording.

Oh I see Example 1 ID: 5cecb9a8-cc82-43ea-b6f1-03755c1c0d3a Timestamp: 05:27 Example 2 ID: 04ef4f55-39ab-4323-8493-27d24ab8e18f Timestamp: 01:30 02:48

This will be fixed pretty soon, we identified the issue and we are working on a fix.

Fantastic