Speech-to-Text Transcription Discrepancy: Call Log...
# support
e
Hello, I'm having a problem with the speech-to-text feature in your system. Here's what's happening: In the call log, the transcription looks correct and doesn't have mistakes. But in the final transcript and in the API, the text has grammar errors or even completely different words from what was actually said. Can you please look into this problem and find a solution? I'd like to understand why there's a difference between what's in the call log and what's in the final transcript. Just so you know, I'm using Deepgram Nova for the Czech language. You can find more details in the screenshots. Thank you for your help. https://cdn.discordapp.com/attachments/1295379103830114335/1295379104056741939/download.png?ex=670e6f4d&is=670d1dcd&hm=8cd2852116e135a219822eeaf133bd72155bffa3c73a4f4bcab48048b01d2bbb& https://cdn.discordapp.com/attachments/1295379103830114335/1295379104375242833/download.png?ex=670e6f4d&is=670d1dcd&hm=b0ec80c09ee01f766a0c88399ea0a5b1ba61a519c474d3e748eabd34314f3403&
v
@EvilMax03 I'm sorry to hear that you're experiencing issues with the speech-to-text feature. It's important that the transcription is accurate and reflects what was said in the call. Since you mentioned that you're using Deepgram Nova for the Czech language, I should inform you that Deepgram does not support your language. However, I recommend trying the Talkscriber Whisper model, which might give you better results for Czech language transcription. To proceed with resolving your issue, could you please provide me with the specific details or examples from the screenshots you mentioned? This will help me understand the problem better and guide you towards a solution.
You can continue this conversation by mentioning me in the message.
Got it. Tagging @User for additional help.