Speech perception in virtual environments

摘要: Many virtual environments like interactive computer games, educational software or training simulations make use of speech to convey important information the user. These applications typically present a combination background music, sound effects, ambient sounds and dialog simultaneously create rich auditory environment. Since allow users roam freely among different producing objects, designers do not always have exact control over what user will perceive at any given time. This dissertation investigates factors that influence perception in under adverse listening conditions. A environment was created study hearing performance audio-visual conditions. The two main areas investigation were contribution "spatial unmasking" lip animation perception. Spatial unmasking refers benefit achieved when target masking are presented from locations. Both visual influencing considered. The capability modern hardware produce spatial release using real-time 3D spatialization compared with pre-computed method creating spatialized sound. It found could be consumer card either headphone surround speaker display. Surprisingly, less effective subjects better than used. Most research on has been conducted pure environments. an additional cue first investigated determine whether this provided benefit. No difference observed visible objects same location as stimuli. Because inherent limitations display devices, often perfectly aligned, causing sound-producing object seen where it is heard. integration between conflicting see had noise. significant regardless stimuli correct matching position, spatially disparate source. Lastly rudimentary investigated. results showed animations significantly contribute also incorrect result worse no used all. The conclusions are: That capabilities can should speech; Perfectly align very for perception; Even enhance