Words only contribute 7% of information in human communications whereas paralinguistic information (pitch, volume, emotions, etc.) contributes 40%. Automatic recognition of speaker state aims to achieve a more natural interaction between humans and machines by making computers react appropriately, and for instance detect potentially dangerous situations such as high levels of alcohol intoxication, somnolence or fatigue. These technologies can be applied to: security applications using detection of fear in abnormal situations, systems for real-life emotion detection, automatic diagnosis of psychiatric diseases, video-conferencing and assistance for diagnosis in psychology, among others.
Contact: Marcelo Albornoz , Leandro Vignolo