Start point (SP) is used to compute the first frame to be
considered by the system. Similarly, end point (EP) is used
to compute the last one. To identify the first frame in a given
utterance, we multiply SP by the total number of frames
contained in the whole utterance under processing and,
likewise, we multiply EP by total frames to get last frame. In
both cases we must round multiplication outcomes to the
nearest integer. SP must have a value less than EP. This
algorithm is outlined