• the probability of occupying a certain state at a certain time (state occupancy probability) can be computed efficiently by taking the product of the forward and backward probabilities

• total probability of a speech utterance given an HMM can be determined at any time by simply summing the forward backward products over all states