Blind Source Separation Performance based on Microphone Sensitivity and Orientation within Interaction Devices

Abstract

The use of smartphones has tremendously increased in the last decade. The coupling of a mobile phone with multiple sensors has expanded the usage of smartphones into a wider area. A basic research on pre-requisites of Blind Source Separation (BSS) to be used in interaction devices is performed. Our focus of research is the application of signal processing techniques for BSS of audio signals captured using interaction devices like the smartphones with the final aim of recovering the spatial listening experience of persons with reduced hearing abilities using the recently proposed app-based system. With this app, the user is able to interactively enhance speech or sound sources in this surroundings using the interactive interface on his smartphone. The sensitivity and orientation of the smartphone microphones (mics) play a very important role to achieve a better quality of BSS and hence is one of the prime pre-requisites for BSS. Therefore, a benchmarking regarding the interdependence of the different internal and external mics as well as the orientation of the smartphones relative to the sources on the separation performance is discussed in this paper. Four smartphones are used to examine the influence of mic sensitivity on the BSS quality of captured signals. Further external mics with smartphones are tested for their influence on BSS. The captured audio signals are separated using PARAllel FACtor analysis (PARAFAC) based BSS algorithm.