Single channel source separation using sparse NMF and graph regularization

Tuan Pham, Yuan Shan Lee, Yan Bo Lin, Tzu Chiang Tai, And Jia Ching Wang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Scopus citations


The aim of single channel source separation is to accurately recover signals from mixtures. In supervised case, non-negative matrix factorization (NMF) is a popular method to separate mixed signals from learned dictionaries. These dictionaries can be produced efficiently by sparse NMF to approximate the input signal as closely as possible. However, previous methods neither consider the structure of the data in terms of the similarity between vertices of the input signal nor use state-of-art variants of NMF that are more efficient than conventional ones. This paper presents a method that incorporate graph regularization constraint into a group sparsity NMF to improve the performance of source separation. Experimental results demonstrate that our method is outstandingly effective for speech separation in two representative scenarios.

Original languageEnglish
Title of host publicationProceedings of the ASE BigData and SocialInformatics 2015, ASE BD and SI 2015
PublisherAssociation for Computing Machinery
ISBN (Electronic)9781450337359
StatePublished - 7 Oct 2015
EventASE BigData and SocialInformatics, ASE BD and SI 2015 - Kaohsiung, Taiwan
Duration: 7 Oct 20159 Oct 2015

Publication series

NameACM International Conference Proceeding Series


ConferenceASE BigData and SocialInformatics, ASE BD and SI 2015


  • Graph regularization
  • Non-negative matrix factorization
  • Source separation
  • Sparse coding


Dive into the research topics of 'Single channel source separation using sparse NMF and graph regularization'. Together they form a unique fingerprint.

Cite this