A Maze with Spatial Audio Cues for the Oculus Rift.

Download .zip Download .tar.gz View on GitHub

Motivation & Background

This is being developed as part of the final project for CS 498sl - Virtual Reality at the University of Illinois at Urbana - Champaign in the Spring 2015 Semester. This is the pilot offering of this class, and is instructed by Steve LaValle.

We have two main motivations behind this project. The first of these is the technical problem of spacial audio. The implementation of audio in virtual reality has been attempted but not polished. Current audio engines are built for games with short, nearby sound effects (e.g, gunfire or splashing) and background music. To have a richer virtual reality experience, it must be possible to create richer audio that mimics real-world audio effects in real time. Publicly available research in spatial stereo audio is easy to find, however the publicly available implementations of spacial audio are unremarkable. The best available implementation we’ve found so far is RealSpace3D Audio, a plug in for Unity 3D; however, we did not find this sufficiently robust for our project.

We have therefore implemented our own spatial audio algorithm for this reason. Our algorithm is based off work from Microsoft Research and the University of North Carolina as presented in the paper - Precomputed Wave Simulation for Real-Time Sound Propagation of Dynamic Sources in Complex Scenes

The second is that of lingering horror. We often find people trying to convey horrific memories, such as those of wars, and similar tragic events. However this sentiment is rarely felt by the person receiving this communication as the magnitude of horror involved is often far greater than what the revier imagines. Usually, the feeling of horror only sets in when a more power stimulus is offered. This is usually the case with well crafted books, movies, and video games.

In this regard, the maze has Jessie Pope's The Call Permiating throughout the scene. This poem was used to recruit rally and recruit young men for the British war efforts during the First World War. Furthermore, we have scenes from BBC's Our World War playing on the walls in an attempt to juxtapose the attitudes of men as they were being recruited, from when they were at war.


Users start off at one end of a large maze. Their goal is to find the source of the poem (Jessie Pope's The Call"). They must use the spatial and directional cues being offered to them by this sound. A successful solution of this maze by a user is proof of concept that our algorithm is at least accurate enough to help users take decisions in a grid allowing 4 forking decisions.

What we found

Our implementation takes into account the occlusion of sound by items placed between the user and the sound source. This often throws users off from the direction of the source. This is because we have significantly decreased the sound's amplitude falloff with distance, and this leads to rather odd sounding reflections off walls. We decreased the falloff in this manner so as to allow a single source of audio to be heard from everywhere inside the maze


Name Github Website
Abhishek Modi modi95
Eric Ahn wchill
Mark Miller MrMallIronMaker