I have such an app on the iPhone. If I speak alone into the microphone (also fast) then works wonderfully. But a conversation of 3-4 people is not possible because an app can not selectively hear. This means that our brain can follow a conversation of two people without problems, even if ten other couples are speaking in the room. An app can not do it. Then there are slang, dialects, laughter between and other sources of interference.