TalkEasy: An Interactive Web-Based Platform for Hiragana Japanese Sign Language Fingerspelling Acquisition using K-Nearest Neighbors (KNN) Algorithm
Joseph Carlo P. Bersoto, Criselle Jose Centeno, Christephen C. De Jesus, Jemerito C. Mangahas, Elsa S. Pascual, Ariel Antwaun Rolando C. Sison · 2024
This is a significant advance in educational technology, being an innovative web-based application developed for the acquisition of Hiragana Japanese Sign Language fingerspelling. This application relies on the real-time recognition and translation process using the K-Nearest Neighbors algorithm. This site also contains features like a smart assistant chatbot, an extensive Hiragana character library along with its associated hand gestures, pronunciation guides, and descriptions that support this. There is also an advanced feature list that includes text-to-speech, speech-to-text, and speech-to-video with gesture dynamic visualization, real-time feedback, etc. The effectiveness of the system was stringently assessed based on the ISO 25010:2011 software quality model, involving 119 participants (comprising 13 I.T. experts [10.92%] and 106 end-users [89.08%]), resulting in excellent performance metrics: Performance Efficiency scored 4.59 (91.8% satisfaction), Usability achieved 4.60 (92%), and Reliability attained 4.55 (91%), culminating in an impressive overall mean score of 4.58 (91.6%). These comprehensive features and excellent test scores make TalkEasy an unparalleled learning tool that is bringing access to JSL study through an entertaining, techno-advanced approach, representing the great milestone in computer assisted language learning and demonstrates quite successful integration of machine-learning capabilities with pedagogical requirements for bridging gaps created by traditional teaching approaches over modern technological capabilities.