Speech-to-text openai with Next js returns incorrect text in Safari

I have a project, and here’s a small copy of it, specifically the logic for record voice (GitHub - ZaharBerku/openai-speech-to-text). When I turn it on in Chrome, everything is fine, it transmits the text word for word, but when I do the same thing in Safari, it returns something completely different, what could it be?

Safari and Chrome, while similar, have implementation differences, a quick Bing/Google for Chrome and Safari compatibility should provide you with detailed lists of the issues.