Currently, there are more than 300 million hearing-impaired people worldwide, and hearing impaired people communicate using sign language. However, people who do not use sign language have great difficulty in communicating, and hearing-impaired people become uncomfortable in social life. In this study, a sign language translator based on motion recognition is developed and the performance of the proposed system is evaluated through experiments. The proposed system recognizes the sign language motion and expresses it by sound and sentence. A device that can measure EMG signal of lower arm and classify pattern using pattern recognition algorithm based on artificial neural network is devised. Through the proposed system, it is confirmed that the classification accuracy is 98% on average for 8 sign language motions.