SoundSpace: What and Where Through Sound

Accessibility technologies for visually impaired users often convey visual scenes through discrete verbal descriptions, which can interrupt natural hearing and increase cognitive load. Other technologies that integrate spatialized audio often focus on orientation, providing spatial cues without conveying what is present. We present SoundSpace, a real-time system that represents object identity and spatial layout through structured auditory cues. SoundSpace builds on prior sensory substitution approaches to combine brief spoken object naming with continuous mappings of distance, vertical position, and horizontal location to loudness, pitch, and stereo panning. To balance spatial awareness against cognitive load, the system separates scene sensing from audio output, providing periodic spatial sweeps and immediate updates when objects move. Open-vocabulary detection and environment profiles allow users to restrict feedback to task-relevant objects, reducing auditory clutter. We describe the design and implementation of SoundSpace and discuss its implications for non-visual spatial perception and active exploration.