Signtalk: Sign Language to Text and Speech Conversion
C.Uma Bharathi, G. Ragavi, K. Karthika
Abstract
People with hearing and speech impairments are facing lots of difficulties while communicating with the public. The proposed work provides a helping hand for hearing/speech-impaired and even blind people to communicate with others. For a sign input, the proposed model provides a speech/text output, thus providing a user-friendly platform for users. Being a minority, the sign language used by them is not known to most people. So, the idea proposed is a system which converts American Sign Language (ASL) to text and speech output. This work uses convolutional neural networks (CNN) to extract efficient hand features to identify the hand gestures according to ASL. The proposed model offers an accuracy of 88%. Hence, this system helps in recognizing the hand gestures of special people and converting them to text and speech to communicate more effectively with normal people.