With the real-time method I can check immediately if the "speech-to-text" conversion has gone well, instead if I upload a file and i have to wait some minutes in order to get the response it is more difficult to check the result...