Affective speech design: Emotional I/O

Abstract

Emotional states are in many cases crucial to the usability, likability, and effectiveness of an interactive system. An ideal system must be able to both efficiently analyze emotional states of its users and believably convey emotions back to its users, as well as altering its behaviors to best fit the needs of its users. Noninvasive, natural means of ascertaining emotion are essential for users to accept emotion-detection in a system. Analyzing speech patterns is one such means. Speech signals can be collected using a microphone with little to no physical contact with the user and contain a plethora of information for determining emotional baselines and temporary states. A smart, adaptive system could use this emotional information to maximize positive sentiment and affect of the user during and after its use, with system manipulation and the use of emotionally expressive agents.