In this paper, we extend the Hopfield Associative Memory for storing multiple sequences of varying duration. We apply the model for learning, recognizing and encoding a set of human gestures. We measure systematically the performance of the model against noise.