In virtual audio, sound sources need to be modeled as realistically as possible to achieve a sense of immersion in the scene. The human auditory system performs an analysis of sound, which enables the extraction of different information to recognize and localize sound sources. There are various aspects known to result in the differences of sound perception of the listener. Different loudness levels and source locations are two of the reasons why the difference in the perception of sound occurs. However, some cases suggest that sound can still be perceived as “different” even with identical loudness and incidence direction. One example of this phenomenon can be experienced in music, as the sound of a guitar can be differentiated with the sound of a violin even when the same tone is played on both instruments with the same loudness level and when the instruments are positioned at the same location relative to the listener. This phenomenon is summed up by the term “sound coloration” or “timbre” which describes different sound characteristics, such as harmonicity, sharpness and roughness, which can’t be described using the sound descriptors of loudness and sound incidence location. Many attempts have been made to represent audio signals in a way that timbral descriptors can be extracted from them, but most have only been focusing on music applications and furthermore restricted to magnitude information. This thesis examines the perception of timbre in the context of the head-related transfer function (HRTF). Pre-existing metrics from literature are implemented in this thesis and also new metrics with assumed perceptual relevance are developed with the goal of presenting a quantitative difference scale in coloration. Relevant conclusions for the paradigm design of listening experiments are drawn.

Melden Sie sich hier an um die Einladung zu dem Vortrag per E-Mail zu erhalten.
Register here to receive the invitation to this talk via E-Mail.

Zoom-Meeting-ID: 942 9566 5525
Passwort: 247337