8th Speech in Noise Workshop, 7-8 January 2016, Groningen

Investigating the effects of noise on arousal during a communication task

Nemanja Cvijanovic(a)
Philips Research Laboratories Eindhoven, NL

Patrick Kechichian
Philips Research Laboratories Eindhoven, NL

Kees Janse
Philips Research Laboratories Eindhoven, NL

Armin Kohlrausch
Philips Research Laboratories Eindhoven; TU Eindhoven, NL

(a) Presenting
(b) Attending

The performance of speech communication systems depends highly on the acoustic characteristics of the environment in which they are being used. In a natural environment, noise is inevitable and ubiquitous and has detrimental effects on communication. During a conversation, it can not only reduce speech intelligibility and lead to an increase in listening effort but also cause annoyance and stress. A lot of work has been invested in the past in methods reducing ambient noise without degrading the speech signal. Various solutions exist, and the benefit of such noise reduction algorithms in communication systems is usually quantified using metrics like the signal-to-noise ratio (SNR) or the perceptual evaluation of speech quality (PESQ). Through our study we want to examine the possibility of evaluations based on physiological changes in users of a communications system.
The aim of our study is to investigate the effects of noise on arousal. For this, we examine physiological changes in speakers in a communication setting for different levels of background noise. In our communication setting, two speakers are seated in separate rooms. They communicate with each other using headsets and microphones on the tables in front of them and try to solve collaborative tasks in different noise conditions. The goal of this setup is to emulate a hands-free communication scenario (e.g., a Skype conversation). Each pair of speakers solves the same types of tasks in low, moderate and high noise conditions with relaxation periods in between. Throughout the conversation, we measure and record the heart rate and skin conductance of each speaker. These physiological signals are later analyzed and extracted features are used to determine arousal states throughout the experiment.
Here, we explain our design, discuss the relevant physiological signals and their characteristics and show some preliminary results and insights. On the long run, quantifying the effects of noise on arousal could enable us to determine the benefits of a noise reduction algorithm on arousal and thus well-being of the users of speech communication systems.

Last modified 2016-05-12 14:22:09