Automatic segmentation of traffic noise

Abstract:

The state of the art in audio source separation reflects a growing interest towards developing tools oriented to speech and music applications. However, its application to the study of noise, and more specifically to the study of traffic noise, is scarce. The source separation in this work environment is a key process for subsequent classification and analysis, and has motivated the present research carried out. This paper is intended to present, therefore, the different strategies followed in the implementation of an automatic segmentator for traffic noise, focusing the research on two main aspects: The removal of components of noise unrelated with the sources of study and, by other side, the detection and separation of audio sources acquired. For this purpose, measures of traffic noise are made and results are analyzed for the running system, achieving a high level of performance in real work conditions.

Año de publicación:

2008

Keywords:

Fuente:

scopus