Abstract
In work directed toward Automatic Recognition of Continuous Speech (ARCS), there are several approaches which may be taken. The selection among these depends heavily on the purposes and, to some extent, the inclinations of the experimentalist/developer.