Impact of Microphone Orientation and Distance on BSS Quality within Interaction Devices

摘要

Smartphones have transformed the technology landscape and cultural aspects of modern society. Smartphones are equipped with multiple sensors designated for different purposes expanding their application into a wider area. Tasks, which earlier needed a computer are now performed on a smartphone and are used for health care and assisted living applications. Our research focusses on development of a smartphone application which enables the recovery of spatial hearing experience of persons with hearing disabilities. This application provides the user, the choices to select the sound signal or signals of their interest and enhance it interactively. This app is based on a Blind Source Separation (BSS) algorithm, which separates the audio signals captured by the device into their individual sound sources. The microphones (mic)s of the smartphones are the main sensors serving this purpose. This paper estimates the impact of mic sensitivity, orientation of the phone and the distance between the mics on the quality of the separation process. This is quantified by calculating the Channel Differences (CD) and the Signal to Interference Ratio (SIR) of the captured signals. Both internal and external mics are tested with android smartphones. The captured signals are separated using PARallel FACtor analysis (PARA-FAC) based BSS algorithm and the quality of separation is tested subjectively based on listening experience.