A Review of Personality in Voice-Based Man Machine Interaction

Abstract

Abstract. In this paper, we will discuss state-of-the-art techniques for personality-aware user interfaces, and summarize our own recent work in automatically recognizing and synthesizing speech with "personality". We present an overview of personality "metrics", and show how they can be applied to the perception of voices, not only the description of personally known individuals. We present use cases for personality-aware speech input and/ or output, and discuss approaches at defining "personality" in this context. We cover the state of the art in recognizing personality from speech, using an approach, which is very similar to other meta-data extraction schemes. We also discuss the difficulty of generating appropriate responses in an automatic dialog system, including work on generating consistency of what is being said, and how it is being said. To conclude, we present speculations about the future of this exciting area of ongoing research.

The INTERSPEECH 2012 speaker trait challenge

“…As evaluation measure, we stick with unweighted average (UA) recall as used since the first Challenge held in 2009 [10]. In the 1 : arithmetic mean of LLD / positive ∆ LLD. 2 : only applied to voice related LLD.…”

“…By that -as opposed to the INTERSPEECH 2010 Paralinguistic Challenge -we now deal with perceived speaker traits. Apart from intelligent and socially competent future agents and robots [1], main applications are found in the medical domain and surveillance.…”

Which One? Choosing Favorite Robot After Different Styles of Storytelling and Robots’ Conversation

“…The first experiment presented a comparison between the cheerful and the serious personalities, which were reflected in the robots’ voices ( Nass and Lee, 2000 ) by asserting the same speed and different voice shaping. The serious robot had a voice with 20% more depth, based on voice personality stereotypes ( Metze, Black, and Polzehl 2011 ). In the second experiment, where all robots had a cheerful personality, the voice shaping and speed varied by 2–5% for each robot, so as to have an almost unnoticeable difference among the robots’ voices, especially during the robot-conversation session ( Polzehl, Moller, and Metze 2011 ).…”