Towards a Speaking Style-Adaptive Assistant for Task-Oriented Applications

Abstract:

This work presents the results of two user studies that help in modeling a speaking style-adaptive assistant for task-oriented applications, such as route guidance or setting the temperature inside a car. We have a look at the linguistic cues in the user’s speaking style and at the desired adaptivity features in the system’s speaking style. We investigate politeness, vocabulary, utterance length, and style of addressing. The results show that vocabulary and style of addressing are suitable to be modeled adaptively on the output side, and can also be deliberately used as triggers for the adaptive system behavior.


Year: 2018
In session: Poster
Pages: 143 to 150