JIITA, vol.8 no.2, p.951-967, 2024, DOI: 10.22664/ISITA.2024.8.2.951
Truong Quoc Thang*, Pham Quoc Bao, Nguyen Van Vi, Than Thi Thao, Nguyen Trieu Tien, Tran Kim Sanh,
International School, Duy Tan University
Abstract: This project presents the development and implementation of a specialized voice assistant tailored to support visually impaired individuals operating Windows PC systems called Beacon. The voice assistant is designed to enhance accessibility by facilitating seamless interaction with integrated functions, including music playback, news consumption, receiving usage instructions, and volume control, catering to the unique needs of visually impaired users. Notably, the assistant is equipped with support for the Vietnamese language and employs a user intention recognition model based on PhoBERT, and FastAPI, contributing to the accurate interpretation of user commands. Evaluation through testing demonstrates an impressive accuracy rate of up to 90% for the provided test data, showcasing the efficacy of the voice assistant in empowering visually impaired users to navigate and utilize Windows PC functionalities independently and efficiently.
Keywords: Voice Assistant, Visually Impaired, Windows PC Accessibility, User Intention Recognition, PhoBERT-based Interaction
Fullpaper: