Answers are grounded in this paper and cite their sources.
This paper describes Football for All , a modular system that transforms football broadcast video into an immersive, headphone-based spatial audio experience for blind and low-vision (BLV) spectators. The system integrates: (i) a computer vision pipeline for detecting and tracking players and the ball, estimating pitch calibration, and converting image detections into pitch coordinates; (ii) a dense action spotting module that temporally localizes ball-related events with strict 1-second precision; and (iii) a Unity-based logic engine that processes a structured game-state stream and generates procedurally created 3D spatial audio using head-related transfer functions (HRTFs) and head tracking. The audio engine renders in real time, but the current prototype is not yet a live system: the computer-vision stage processes pre-recorded broadcast footage sequentially to preserve identity consistency, and end-to-end latency has not yet been measured. The design supports robust post-processing under occlusions and viewpoint changes, and an event-to-sound audio lexicon that reduces listener cognitive load by mapping football semantics to distinct acoustic cues.
No takes yet. Share an insight, caveat, or question.
Marques et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: