8 Pages

AugECE – A framework for audio signal extraction, classification, and enhancement

WithH. Wang

Classification, and Enhancement) that accomplishes the extraction, classification, and enhancement of audio signals. Specifically, AugECE first extracts the signals of interests from a mixture of audio signals via independent component analysis (ICA); AugECE then classifies them by k-nearest neighbor (kNN) to match their corresponding labels; last, wavelet-based hidden Markov models (HMM) and principal component analysis (PCA) are further employed to enhance the quality of the extracted signals by reducing background noise and/or residual signals originated from other sources as well. A high-level schematic diagram is illustrated in Fig. 1.