Determinants of Audio-Visual effects in degraded and non-degraded speech. Seeing a speaker's face can affect the perception of their speech in a number of ways. This project proposes a detailed comparison of factors that affect Audio-Visual (AV) facilitation of degraded speech detection and identification. Detection-based tasks should be more sensitive to signal based correlations whereas identification-based effects more sensitive to complementary information. The significance of the current pr ....Determinants of Audio-Visual effects in degraded and non-degraded speech. Seeing a speaker's face can affect the perception of their speech in a number of ways. This project proposes a detailed comparison of factors that affect Audio-Visual (AV) facilitation of degraded speech detection and identification. Detection-based tasks should be more sensitive to signal based correlations whereas identification-based effects more sensitive to complementary information. The significance of the current proposal is that it offers both a strategy and a connected series of experiments for determining key behavioural constraints on AV speech integration. Understanding AV interactions will build links between neurophysiological processes and coherent perception and have important implications for AV application.Read moreRead less
Using AI to reveal the true extent & context of alcohol exposure in videos. This project aims to extend an artificial intelligence algorithm to automatically identify and quantify alcohol prevalence in videos. The project is expected to generate significant new knowledge about alcohol’s exposure in these videos’ social, emotional, and environmental contexts. The expected outcomes include a more efficient and automated method of revealing alcohol pervasiveness and its context in the 1000 most wat ....Using AI to reveal the true extent & context of alcohol exposure in videos. This project aims to extend an artificial intelligence algorithm to automatically identify and quantify alcohol prevalence in videos. The project is expected to generate significant new knowledge about alcohol’s exposure in these videos’ social, emotional, and environmental contexts. The expected outcomes include a more efficient and automated method of revealing alcohol pervasiveness and its context in the 1000 most watched videos in Australia, making costly manual coding redundant. Anticipated benefits include enabling governments to better monitor compliance to alcohol product placement guidelines and increased public awareness of the frequency and harmful effects of being exposed to alcohol in videos.Read moreRead less
Establishing how head and face movement properties contribute to the perception of speech and identity. The proposed studies provide an extensive research program into audio-visual speech processing and person identification: key components of face-to-face communication. The likely impact and benefits of the project concern its contribution to perceptual theory (linking signal, brain and behaviour) and its practical implications: For determining the viability of multimodal biometric identificati ....Establishing how head and face movement properties contribute to the perception of speech and identity. The proposed studies provide an extensive research program into audio-visual speech processing and person identification: key components of face-to-face communication. The likely impact and benefits of the project concern its contribution to perceptual theory (linking signal, brain and behaviour) and its practical implications: For determining the viability of multimodal biometric identification procedures by assessing the distinctiveness and permanence of AV speech characteristics. In the development of a visual dubbing technique that has potential for communication in noisy environments (and for the deaf) and for the development of a morphable model for AV presentation that has application for both first and second language learning.Read moreRead less