A Study of Audio Watermarking Method Using Non-negative Matrix Factorization for a Duet of Different Instruments

  • Harumi MurataEmail author
  • Akio Ogihara
Conference paper
Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 82)


In this paper, we propose an audio watermarking method using non-negative matrix factorization (NMF) for a duet of different instruments. NMF is applied to host signal and the amplitude spectrogram is factorized into the basis matrix and the activation matrix. In the proposed method, the activation coefficients of each musical instrument can be obtained by NMF, and watermarks can be embedded into a specific instrument or multiple instruments. Onset and offset times are estimated from the activation matrix, and its duration is defined as one note. However, it is possibility that onset and offset times are changed due to embedding process, and watermarks cannot be extracted correctly unless note information is known. Therefore, in this paper, onset and offset times are estimated after smoothing activation coefficients, and errors of note estimation are reduced. From this, we want to extract watermarks even if note information is unknown.


Audio watermarking Nonnegative matrix factorization Duet 


  1. 1.
    Lie, W.N., Chang, L.C.: Robust and high-quality time-domain audio watermarking based on low-frequency amplitude modification. IEEE Trans. Multimedia 8(1), 46–59 (2006)CrossRefGoogle Scholar
  2. 2.
    Murata, H., Ogihara, A.: Digital watermark for real musical instrument sounds using non-negative matrix factorization. In: International Technical Conference on Circuits/Systems, Computers and Communications, pp. 677–680 (2016)Google Scholar
  3. 3.
    Lee, D.D., Seung, H.S.: Algorithms for non-negative matrix factorization. Neural Inf. Process. Syst. 13, 556–562 (2001)Google Scholar
  4. 4.
    IHC Evaluation Criteria and Competition,, Accessed March 30 2017
  5. 5.
    Blancou, V., Fujii, K., Okada, T.: 40 Etudes for Clarinet and Piano. UNIVERSITY EDUCATION PRESS (2006)Google Scholar

Copyright information

© Springer International Publishing AG 2018

Authors and Affiliations

  1. 1.Department of Information Engineering, School of EngineeringChukyo UniversityToyotaJapan
  2. 2.Department of Informatics, Faculty of EngineeringKindai UniversityHigashi-hiroshimaJapan

Personalised recommendations