Hi! When I use the TTS nova-2-general-pt-br model in Vapi, it detects the information incorrectly. However, when I download the same recording from Vapi and send it to Deepgram, it transcribes the conversation correctly. Could you please explain why the same model from the same provider works in Deepgram but fails in Vapi? ssistant id:8113c4cd-c9f1-4523-847a-15be5fbf0850 17:29 calls example: 0199c575-5dd7-7bb5-8e6e-eb0dfb7ff535 17:29 0199c574-995f-7aa0-a241-6e4afd39dc58

That happens because Vapi streams audio in real time, while Deepgram transcribes full audio files. Streaming often loses small audio parts or context, so the model can misdetect words. Check your transcriber settings in Vapi make sure the encoding, sample rate (16kHz or 44.1kHz), and language code (pt-BR) match Deepgram’s setup. This usually fixes it.

Hey, I’ve handled similar cases before where Vapi and Deepgram gave different results using the same model. I can help you trace where the issue happens—usually it’s with how Vapi streams or encodes the audio before sending it for transcription. Could you share the audio format, bitrate, and how the data is being passed to Vapi? @ClaraThays

We apologize for the delay in responses lately. We are in process of making some changes to our workflow in how we handle support requests. The good news is that we are starting to get back on track of having a 48 hour response time and we would like to start cleaning up anything old. With that being said, just wanted to check in with you to see if you have made any progress or changes to your project since support has last reached out.