<![CDATA[Communication is a way for humans to convey information, ideas, thoughts, feelings, or massages verbally or nonverbally. People with hearing impairments use expressions and symbols due to hearing limitations, the term tuna means less and rungu means hearing. To bridge communication, sign language translation through the development of systems such as text-to-speech can help improve accessibility between people with hearing impairments and spoken language users in various contexts of everyday life. The Indonesian Sign Language (SIBI) translation system in this study was built the waterfall research method in which each process will be carried out in stages and sequentially, starting from problem analysis and creating a conceptual system in order to strengthen the theorectical foundation and answer the formulation of the problems raised, this study uses the Keras deep learning model and the Convolutional Neural Network (CNN) approach. The test results show that the system is able to classify hand gestures and convert them into text and voice, in addition to the average respone time of less than one second, which is around 104 miliseconds per step. This shows that the system can operate in real-time with a fast and consistent response. Video transcription and voice synthesis for an Indonesian sign language translation system using the Python programming language were successfully implemented, supported by various libraries such as OpenCV, CVZone, and GTTS (Google Text-To-Speech). Testing results on 15 movement classes with 150 trials showed an average movement classification accuracy of 87%. Further research is recommended to increase the number of subjects and dataset diversity to improve the model’s generalizability.]]>
Copyrights © 2026
