Parkinson's disease generates a special interest in factors such as gait patterns, posture patterns, and risk of falls. The human gait pattern has a basic unit called the gait cycle, composed of two phases: stance and swing. Using gait analysis it is possible to get spatiotemporal variables as walking speed and step number derived from stance and swing phases. In this paper, we explore the feasibility of wavelet techniques to analyze gait signals, we use a member of Daubechies family to distinguish automatically gait phases, this approach allowed us to estimate spatiotemporal variables that shows significant differences between Parkinson patients and non-Parkinson patients, this result aims to allow clinical experts to easily diagnose and assess Parkinson patients, with short evaluation times and with non-invasive technologies.