@inproceedings{8b6fe9b9c9be47e98127c2949f504643,
title = "Personalized Conversational Audio Descriptions in 360° Virtual Reality for Blind and Low-Vision Users",
abstract = "On-demand, conversational audio descriptions in 360° VR empower blind and low-vision users to actively explore immersive visual content. We present a Meta Quest demo that integrates head-pose-based view snapshots, real-time speech recognition, and GPT-4o-powered chunked text-to-speech streaming directly on-device to support multi-turn Q\&A with personalized voice profiles. Our pipeline leverages chunk transfer encoding to play AI-generated audio as it's produced, minimizing perceived delay. Unlike prior VR accessibility demos reliant on static or author-crafted descriptions, our multimodal system delivers dynamic, user-driven narration for inclusive and interactive VR experiences.",
keywords = "accessibility, audio description, Conversational AI, virtual reality",
author = "Khang Dang and Sooyeon Lee",
note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 2025 IEEE International Symposium on Mixed and Augmented Reality Adjunct, ISMAR-Adjunct 2025 ; Conference date: 08-10-2025 Through 12-10-2025",
year = "2025",
doi = "10.1109/ISMAR-Adjunct68609.2025.00268",
language = "English (US)",
series = "Proceedings - 2025 IEEE International Symposium on Mixed and Augmented Reality Adjunct, ISMAR-Adjunct 2025",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "961--962",
editor = "Ulrich Eck and Gun Lee and Alexander Plopski and Missie Smith and Qi Sun and Markus Tatzgern",
booktitle = "Proceedings - 2025 IEEE International Symposium on Mixed and Augmented Reality Adjunct, ISMAR-Adjunct 2025",
address = "United States",
}