Auditory Immersion in 3D Multichannel Audio Reproduction

results showed however, that the perception of factors relating to vertical sound perception of both envelopment and localisation were rated lower on average compared with the comparable question relating to horizontal sound perception.

An initial experiment was designed to create an optimal speaker layer balance for the four ‘22.2’ (not all recordings were presented natively in 22.2 and no subwoofers were utalised) recordings being utilised as stimuli for remaining experiments. This test showed that height layers were mixed louder than floor layers for three out of the four content stimuli, though all height and floor layer were on average mixed at a lower average loudness compared to the main layer level which was fixed throughout. The final experiment compared different speaker formats (mono, 2.0, 5.1, 9.1 and 22.2) with perceptual features highlighted throughout as potentially the most influential to the perception of immersion. Results concluded that 5.1, 9.1 and 22.2 formats were found to be significantly similar in the majority of test cases. 2.0 and mono formats found sporadic statistical similarity but were consistently rated lower than the other formats. When main effect of the format was considered with dependency on the perceptual attributes utalised, it was discovered that all stimuli results were found to be statistically significant when a Friedman test was carried out for the factors of Listener Envelopment (LEV) and Presence (Pres), but were not for Overall Tonal Quality (OTQ) and Quality of Experience (QoE). The result suggesting that the perceptual factors of OTQ and QoE are highly content specific in terms of user perception. LEV and Presence on the other hand are perhaps not as closely linked to content overall. Significant changes in perception are more clearly identified for these perceptual factors when considering different format reproductions, with no significant differences being found between 22.2, 9.1 and 5.1 across most stimuli tested, and mono being the lowest rated format for all stimuli and perceptual attributes.

MRes project: Oct 2018 – Jun 2020

Researcher: Callum Eaton

Supervisor: Dr Hyunkook Lee

Publications:

Next Post
Towards a Perceptual Model of Clarity in Music Mixes
Previous Post
ASPEN (APL Spatial Audio Engine)