What about recording the audio of your output? device and then transcribing it