Speaker-Dependent Isolated Word Recognition with a Bitmap Space