07.05.2019: Volker Dellwo

Speaker-idiosyncratic information in voice has typically been viewed as a static by-product of speech production that needs to be cancelled out to obtain the abstract underlying linguistic forms. Here we argue that idiosyncratic information is dynamic and can to some degree be controlled by the speaker, for example by applying different speaking styles which reveal more or less information about a speaker's identity. We will show examples from infant-directed, deceptive and clear speaking styles, which reveal differnet degrees of acoustic individuality information which in return leads to measurable differences in recognition performance by humans and machines. We argue that speaking styles have evolved in parts to facilitate individuality recognition.