When I record audio files and send them to a storage bucket in Supabase this works fine on all devices. The audio files are there (all mp3 extensions) and play my voice. However, when I send these links to openai whisper, only the ones created by windows chrome seem to be recognized and transcribed. The others don't transcribe. There appears to be an issue with these files but I don't know what issue. Are these actual mp3 files or is there an issue with these files in fact being m4a files.. Not sure what is the issue here.