Parkinson disease occurs when certain clustersof brain cells are unable to generate dopamine which is needed to regulate thenumber of the motor and non-motor activity of the human body. Besides,contributing to speech, visual, movement, urinary problems, Parkinson diseasealso increases the risks of depression, anxiety, and panic attacks,disturbances of sleep. Parkinson disease diagnosis via proper interpretation ofthe vocal and speech data is an important classification problem. In thispaper, a Parkinson disease diagnosis is realized by using the speechimpairments, which is one of the earliest indicator for Parkinson disease. Forthis purpose, a deep neural network classifier, which contains a stackedautoencoder and a softmax classifier, is proposed. The several simulations areperformed over two databases to demonstrate the effectiveness of the deepneural network classifier. The results of the proposed classifier are comparedwith the results of the state-of-art classification method. The experimentalresults and statistical analyses are showed that the deep neural networkclassifier is very efficient classifier for Parkinson disease diagnosis.