We define production gesture starting from specific acoustic criteria. The characteristics of these gestures are essentially dynamic: they deform the vocal tract area function by operating on regions that are acoustically the most sensitive. Gestures generated using a model that incorporate these criteria are validated in two ways. First, they are compared to the natural gestures generated by the human speech production mechanism. Second, the model gestures are tested in listening experiments to determine the relationship between production dynamics and perceptual dynamics.