A Dictionary Learning Method Based On Emd For Audio Sparse Representation

Yueming Wang, Zenghui Zhang, Rendong Ying, Peilin LiuZenodo 2013 · DOI

Abstract

Sparse representation has long been studied and several<br> dictionary learning methods have been proposed. The dictionary<br> learning methods are widely used because they are adaptive. In this<br> paper, a new dictionary learning method for audio is proposed. Signals<br> are at first decomposed into different degrees of Intrinsic Mode<br> Functions (IMF) using Empirical Mode Decomposition (EMD)<br> technique. Then these IMFs form a learned dictionary. To reduce the<br> size of the dictionary, the K-means method is applied to the dictionary<br> to generate a K-EMD dictionary. Compared to K-SVD algorithm, the<br> K-EMD dictionary decomposes audio signals into structured<br> components, thus the sparsity of the representation is increased by<br> 34.4% and the SNR of the recovered audio signals is increased by<br> 20.9%.
BibTeX
@inproceedings{wang2013dictionary,
  author = {Yueming Wang and Zenghui Zhang and Rendong Ying and Peilin Liu},
  title = {{A Dictionary Learning Method Based On Emd For Audio Sparse Representation}},
  booktitle = {Zenodo},
  year = {2013},
  doi = {10.5281/zenodo.1087631},
}

Loading related content...

Figure