Auditory Scene Analysis Using Independent Component Analysis

The human auditory system can perceive and locate the sound he focuses mixed with complex environment sounds. Auditory scene analysis tries to model auditory processing by computer technology.In this thesis the combination of Independent Component Analysis (ICA) method and Time-Frequency (TF) analysis method are proposed in the research of auditory scene analysis. The history and current state of development and research of the two methods are introduced firstly. Then a new model of auditory scene analysis is constructed. It uses ICA to separate mixed speech signals received in the scene, and the identification of speech signals is accomplished based on the combination of TF and distance measures. At the end, the sampling norms of experimental data and assessment of results are showed to prove the theoretical importance of the model.