In modeling speech recognition, the sound and the image can be assumed to be independent of each other