Design of a sound database for stress assessment of indoor noises in a TV-watching environment

Abstract

When watching television (TV) at home, indoor noise, such as door slam, water tap, and washing machine sound, could overlap TV audio, thus it causes to reduce TV listeners' capability to understand the TV content. This makes them stressed. Accurate evaluation of the stress created by indoor noise is a first step to soften the stress and it can be realized by investigating the relationship between the stress and indoor noise. However, the types of indoor noise are numerous and uncertain when such noises occur in real life. Thus, an indoor sound database (DB) that includes both TV contents and noises is in high demand for further research on such stress assessment. Throughout further investigation of the stress evaluation by using the proper sound DB, stress arisen by indoor noises would be alleviated, resulting improvement of the overall quality of daily life. In this paper, we propose a design method of a DB for the assessment of stress created by indoor noise in a TV-watching environment. To collect the DB, both binaural and multi-channel recording configurations are considered for recording TV audio contents and the various types of indoor noise. In particular, a dummy head and a linear microphone array are used to record sound signals for binaural and multi-channel configuration, respectively. The proposed sound DB consists of five. different TV audio contents and 28 different types of indoor noise, whose total length is ten and two hours, respectively. Finally, an experiment regarding the sound localization of indoor noise is carried out to demonstrate the usefulness of the proposed sound DB.