Depends on what you are watching? If it's a scripted show the transcriber would have a copy of the script so they KNOW what the words are supposed to be. In fact the closed caption is already imbedded at the factory.
Sometimes a character will ask a question and the response of the other character is a disgruntled, or annoyed stream of mumbles designed to be obscenities that you can't say on TV. Hence "unintelligible." If your hearing good enough to sort of keep you in the game you can get the wrong impression.
If it's a sporting event or other live thing people will talk too fast, talk all at once, be overridden by a surge in crowd noise etc. A court stenographer can interrupt and ask for a restatement. On live TV with a transcriptionist OR a speech recognition machine you can't get that.