Benjamin Weiss - Talker Quality in Human and Machine Interaction: Modeling the Listener’s Perspective in Passive and Interactive Scenarios

More information about this series at http://www.springer.com/series/10013

This Springer imprint is published by the registered company Springer Nature Switzerland AG.

The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

When humans engage in spoken interaction, they assess the obvious features of the interaction partner, such as the verbal and nonverbal signals of the voice, as these contribute to the subjective evaluation of the interlocutor. In this book, the back- ground, state of research, and own contributions to the assessment and prediction of talker quality that is constituted in voice perception and in dialog are presented. Starting from theories and empirical findings from human interaction, major results and approaches are transferred to the domain of human-computer interaction. The main subject of this book is to contribute to the evaluation of spoken interaction in both humans and between human and computer, and in particular to the quality subsequently attributed to the speaking system or person, based on the listening and interactive experience.

The theories, methods, and results presented are focused on the first impression of the people engaged in such a vocal conversation. This means hearing a voice for the first time in either a passive scenario (listening only) or interacting with a person or computer for the first time (interactive scenario). By using the term first impression, the research focus is set to the human perception and evaluation of voices and conversations experienced, which represents the beginning attitude formation of the participating human towards the (other) speaker, may it be a real person or a computer agent.

The main scientific contribution is not to psychological theory development, but it is an informed engineering approach for describing subjective quality as experienced by users. As mentioned in the subtitle, the major results of this book are the development of quantitative models of user ratings to represent subjective quality. The most important part of this modeling is the identification of relevant parameters as predictors. Such predictors of talker quality can be acoustic features, for the case of voices quality that is typically assessed in passive scenarios, or data describing interaction behavior for the course of conversation. By modeling voice-stimulated user ratings, this book also contributes to the identification of the most important perceptual dimensions as these factors provide the valuable insight in search for quantitative predictors, and derives basic insights and principles in designing and evaluating modern spoken conversational systems , with the aim of increasing and ensuring quality.

This book is intended for advanced readers, who have already a background in speech signal processing and some basic knowledge in social psychology. Results on the topic of talker appraisal are presented mostly in a rather condensed form, and not all basic terms are defined or explained. Several of my own contributions, in addition to other work and results presented here, are of course published elsewhere. Therefore, please use the references to obtain more details and additional information if the respective summarized section does not answer all your questions.

I want to thank first and foremost my mentor, Sebastian Mller, who supported me for over 10 years, in which I have worked on my favorite topics. He has built up a great team at the Quality and Usability Lab. There are so many current and former colleagues, too many to name here, without whom this book would not exist. Many are listed in the references, so I mention just Christine, Klaus, Ina, Felix, Irene, Yasmin, Stefan, Tilo, Thilo, Matthias, and Laura. On many conferences and on other occasions, I received valuable comments from colleagues in the field, in anonymous reviews but also in person while enjoying lively discussions. In particular, I want to thank Jrgen and Timo. My gratitude also goes to Petra Wagner and Elmar Nth for writing reports on this manuscript for the habilitation committee. And lastly, I want to thank my family for being there for me.

I would like to acknowledge the financial support of the Deutsche Forschungsgemeinschaft throughout my academic career, lastly by the projects Sympathie von Stimme und Sprechweise Analyse und Modellierung auditiver und akustischer Merkmale and Human Perception and Automatic Detection of Speaker Personality and Likability: Influence of Modern Telecommunication Channels.