May 2019 - September 2019

GAME AUDIO: EXPLORING BINAURAL VS STEREO REPRODUCTION OVER HEADPHONES

Abstract

This paper seeks to explore the differences in sound quality between binaural and stereo audio reproduction over headphones. A gap in research is revealed through the analysis of previous literature, where audio has been spatially enhanced for headphones from pre-exisiting video games in order to explore the differences in reproduction formats. This paper introduced a method of critically analysing audio over headphones to understand the perceived differences between both forms of audio in a virtual testing environment. This environment tasked participants to locate hidden waypoints through sound; providing both quantitative and qualitative information on the interaction between the participant and the audio format. A video game was used due to the dynamic nature it offers; allowing participants to have more degrees of interaction with stimulus as opposed to film or music (which are static mediums). The game was developed in the Unity Editor, where audio was implemented through FMOD Studio. The Googles Resonance Audio package was used to binaurally render audio sources within the game environment. Based on prior literature, this paper predicted that binaural audio would be more fitting for a dynamic medium that would accurately reflect real-world environmental effects. This test was conducted with sixteen experienced listeners as defined by Zacharov and Koivuniemi (Koivuniemi and Zacharov, 2005). The results found statistical significance for better perceived Enjoyment and Clarity as well as more pronounced Comb Filtering over the binaural format. However, the results of this paper highlight the need for additional testing of binaural spatialisers within video game environments, developed for the sole purpose of exploring the various sub-sections of sound quality.

​

February 2018- May 2018

INVESTIGATING THE PERCEIVED TIMBRE AND LOCATEDNESS OF DIFFERENT SURROUND SOUND DECODERS

Abstract

Current Ambisonic technologies allow users to experience three-dimensional sound fields over various loudspeaker arrays. Directional information is provided over these arrays through the use of surround sound decoders. Using Cycling MAX and IRCAM SPAT, the perceived timbre and locatedness of various Ambisonic decoders over a first order cuboid loudspeaker array were tested at the University of York Audio Lab. The listening tests conducted involved panning a moveable pink noise burst using a 3D mouse to a static reference burst within the loudspeaker array. Moreover, participants were asked to listen and rate various stimuli using a list of timbral attributes. Word anchors and definitions were provided for each attribute supplied. These listening tests have provided insight into what decoder to choose for audio reproduction over first-order Ambisonic speaker arrays. The following decoders where used during the listening tests: Basic, Inphase, BasicMaxre and InphaseMaxre. When assessing timbre, a definite difference in decoder performance was observed. From the results one could argued that both the Basic and Inphase produced the most consistent results thought-out all samples. Each decoder present favourable and unfavourable characteristics with many attributes being rated similarly on average. Overall it can be argued that BasicMaxre produced the most preferred characteristics while InphaseMaxre depicted the most unfavourable characteristics. The localisation tests showed BasicMaxre allowed participants to most accurately locate sources across the azimuth while elevation was most accurate with the Maxre decoder. These tests demonstrated the multi-dimesnsional properties of assessing timbre in the context of spatial audio. and the need for additional experienced listeners to further validate the use of each decoder for various mediums.