Fortunately, automatically generated subtitles are often good enough.
How would you know if you don’t speak the language?
How do you know they don’t?
How do you know you can comprehend someone's English? You assume you do because the information conveyed isn't self-contradictory or nonsensical, doesn't contradict your observed reality and things you know to be true, and is unlikely to do so by mere chance.
I noticed if the speaker has a thick accent, it will often interpret them as speaking their own native language, and generate nonsense.