Conversion of Indian Sign Language to Speech by Using Deep Neural Network

Sign language is the most used non-verbal communication method. It is mostly used by persons who have hearing and speaking disabilities. In this world, normal people find it difficult to interpret sign language, so it is very difficult to communicate with mute and deaf people. Mute and deaf people aren't able to express their thoughts to the world. To interpret the meaning of sign language an interpreter is needed. To eliminate the need of an interpreter a system is proposed using neural network. The proposed system can easily interpret the meaning of human hand sign to convert it into relevant speech. Our system has been implemented by using a deep learning technique, Convolution Neural Network (CNN). The hand images are pre-processed using canny edge detection technique for better accuracy. After pre-processing, the classification and prediction steps are applied. The speech synthesizer is used to convert text into speech using Google text to speech API. The proposed system achieves accuracy of up to 98%.