Separation of Single Channel Mixture Using Time-domain Basis Functions

  • 장길진 (한국과학기술원 Department of Computer Science) ;
  • 오영환 (한국과학기술원 Department of Computer Science)
  • Published : 2002.05.01

Abstract

We present a new technique for achieving source separation when given only a single channel recording. The main idea is based on exploiting the inherent time structure of sound sources by learning a priori sets of time-domain basis functions that encode the sources in a statistically efficient manner. We derive a learning algorithm using a maximum likelihood approach given the observed single channel data and sets of basis functions. For each time point we infer the source parameters and their contribution factors. This inference is possible due to the prior knowledge of the basis functions and the associated coefficient densities. A flexible model for density estimation allows accurate modeling of the observation, and our experimental results exhibit a high level of separation performance for simulated mixtures as well as real environment recordings employing mixtures of two different sources. We show separation results of two music signals as well as the separation of two voice signals.

Keywords

References

  1. Computer Speech and Language v.8 no.4 Computational auditory scene analysis G.J.Brown;M.Cooke https://doi.org/10.1006/csla.1994.1016
  2. Signal Processing v.36 Independent component analysis, A new concept? P.Comon https://doi.org/10.1016/0165-1684(94)90029-9
  3. Speech Communications v.27 Listening to two simultaneous speeches H.G.Okuno;T.Nakatani;T.Kawabata https://doi.org/10.1016/S0167-6393(98)00080-6
  4. IEEE Trans. on Neural Networks v.10 Seperation of speech from interfering sounds based on oscillatory correlation D.L.Wang;G.J.Brown https://doi.org/10.1109/72.761727
  5. Advanced in Neural Information Processing Systems v.13 One microphone source seperation S.T.Roweis
  6. in proc. ICASSP(Salt Lake City, Utah) The statiscal structures of male and female speech signals T.W.Lee;G.J.Jang
  7. Network: Compultation in Neural Systems Learning the higher-order structures of a natural sound A.J.Bell;T.J.Sejnowski
  8. Journal of VLSI Signal Processing v.26 no.1-2 Flexible independent component analysis S.Choi;A.Cichocki;S.Amari https://doi.org/10.1023/A:1008135131269
  9. IEEE Signal Processing Letters v.4 Infomax and maximum likelihood for blind source separation J.F.Cardoso https://doi.org/10.1109/97.566704
  10. in Proc. ICONIP(Hong Kong) A context-sensitive generalization of ICA B.Pearlmutter;L.Parra
  11. in Proc. International Workshop on Independent Component Analysis(ICA'00), (Helsinki) The generalized gaussian mixture model using ICA T.W.Lee;M.S.Lewicki
  12. IEEE Trans. on Signal Proc. v.45 no.11 Blind source separation semiparametric statistical approach S.Amari;J.F.Cardoso https://doi.org/10.1109/78.650095
  13. In Proc. EUSIPCO Separation of a mixture of independent sources through a maximum likelihood approach D.T.Pham;P.Garrat;C.Jutten
  14. Neural Computation v.7 An information maximization approach to blind separation and blind deconvolution A.J.Bell;T.J.Sejnowski https://doi.org/10.1162/neco.1995.7.6.1129