Speech disorder is a condition where a person can-not properly create speech sounds required for communication. About 5% of children in the world suffer by speech disorders. Dysarthria is a motor speech disorder which results in slurred or slow speech that is hard to understand. It is a neurological disorder, that affects the muscles which help to produce speech. In this paper we propose a method where machine learning can be used to predict the actual word spoken by dysarthric patients while the dysarthric speech may not be clearly understood by the common man. We have also proposed a method to predict the gender of the speaker and severity of dysarthria in patients. We have also built an application which has also the above mentioned features along with the option of saving the details and also emailing the data.