Pedestrian navigation through multi-track spatial music

Abstract

Spatial audio exploits the human's natural ability to localize the origin of a sound by applying special filters to recorded audio. An interesting application of this is providing strong navigational cues to pedestrians without visual aids, creating a hand-free experience. Current systems require users to listen for artificial beacon sources (which blocks the use of headphones for other sources) or re-position an entire song in space, severely reducing the listening experience.

In this Master's thesis, we present a system that utilizes multi-track recordings to provide navigational cues by shifting only a single track in auditory space. In a real-world field study with 17 participants, who walked a route in a small village centre, we used varied popular music to guide them from waypoint to waypoint. The impact of different types of beacon tracks (Vocals, Drums and 'Mix of Lead Instruments/Vocals') were studied, finding the Vocals to be the most effective and preferred by the participants. The users managed to complete the test successfully and rated the experiment fun with a listening experience close to common music listening.