Deaf and dumb persons who want to communicate with others use sign language as an alternative method. Basically, hand gestures are used to perform sign language. With the advancement of technology, mobile applications based on Artificial Intelligence help such people to participate in social communication. The general category of applications is not suitable for a person who wants to communicate with deaf and dumb people. The systems basically lag into real-time communication with higher accuracy and less time. The proposed system is designed to instantly recognize the sign gesture used in American sign language (ASL). The image will be captured and sent to the prediction model with the help of Open Computer Vision (OpenCV) Library. For feature extraction and picture classification, the convolution neural network (CNN) algorithm is used. Using the CNN algorithm, the predictive model can recognize up to 20 alphabets accurately and achieve an 88% accuracy rate. The proposed system translates real-time sign language into text, making it beneficial for understanding and interacting with dumb and deaf individuals.