It’s driving me mad… seriously! I really want to use the voice feature, but so often my spoken part comes out as Welsh text. They will have to fix this at some point and I really want to know the issue.
There’s a thread in the dev environment here that seems to fix it by deleting the first 30 seconds of the transcription where language is determined. But, I don’t see that as a solution.
Link below with AI summary of the thread at end. It’s a closed topic.