In speech vs vision tasks, how many parameters that describe these events, and do you integer all ranges of these parameters? If yes, by same approaches?

Are there any feedback loops to analyze the data?