The Journal of the Korea institute of electronic communication sciences (한국전자통신학회논문지)

Korea Institute of Electronic Communication Science (한국전자통신학회)
권호 표지
DOI QR코드

DOI QR Code

Improvement of Background Sound Reduction Performance by Non-negative matrix Factorization Method by Wiener Filter Post-processing

위너필터 후처리를 통한 비음수행렬분해 기법의 배경음 저감 성능 향상

  • 이상협 (경성대학교 대학원 전자공학과) ;
  • 김현태 (동의대학교 응용소프트웨어공학전공)
  • Received : 2019.07.02
  • Accepted : 2019.08.15
  • Published : 2019.08.31
Download PDF
⟨ Previous Next ⟩

Abstract

In this paper, we propose a method to improve the background sound separation performance by adding a Wiener filter to the end of the non - negative matrix factorization method. In the case of a mixed voice signal with background sound, a part that has not yet been completely separated may remain in the signal that separated first by the non-negative matrix factorization method. In this case, it can be reduced in proportion to the size of the residual signal due to the Wiener filter, so that the background sound separation or reduction effect can be expected. Experimental results show that the addition of the Wiener filter is more effective than the case of applying the non-negative matrix factorization method.

본 논문에서는 비음수 행렬 분해 필터 뒷단에 위너필터를 추가하여 배경음 분리 성능을 향상하는 방법을 제안한다. 배경음이 혼재된 음성 신호의 경우 비음수 행렬 분해 기법으로 1차 분리된 신호에는 아직 완전히 분리되지 못한 부분이 잔류할 수 있다. 이러한 경우 위너필터에 의해 잔류하는 신호의 크기에 비례하여 줄여줄 수 있어 배경음 분리 또는 저감 효과를 기대할 수 있다. 실험을 통해 위너필터를 추가한 경우가 비음수행렬 분해 기법만 적용한 경우에 비해 저감 효과가 높은 것을 확인할 수 있었다.

Keywords

KCTSAD_2019_v14n4_729_f0001.png 이미지

그림 1. 제안하는 방법의 블록도 Fig. 1. Block diagram for the proposed method

KCTSAD_2019_v14n4_729_f0002.png 이미지

그림 2. SNR 테스트를 위한 블록도 Fig. 2. Block diagram for SNR test

표 1. 기존 대표 방법과 실시간 NMF간 성능 평가 결과 Table 1. Performance evaluation results between the proposed method and conventional NMF

KCTSAD_2019_v14n4_729_t0001.png 이미지

표 2. 후처리 연동 방법과 기존 NMF간 성능 평가 결과 Table 2. Performance evaluation results between the proposed method and conventional NMF

KCTSAD_2019_v14n4_729_t0002.png 이미지

References

  1. H. Lee, J. Seo, Y. Lee, J. Kim, and S. Lee, "Study on optimal number of latent source in speech enhancement based Bayesian nonnegative matrix factorization," In Proceeding of 2015 The Korean Institute of Broadcast and Media Engineering Conference, Jeju island, South Korea, July 2015, pp. 418-420.
  2. J. Yoo, M. Kim, K. Kang, and S. Choi, "Nonnegative Matrix Partial Co-factorization for Drum Source Separation," 2010 IEEE International Conference on ASSP(Acoustics, Speech, and Signal Processing), Dallas, Texas, USA, Mar. 2010.
  3. S. Lee, "Online Monaural Ambient Sound Extraction based on Nonnegative Matrix Factorization Method for Audio Contents," Journal of The Korean Institute of Broadcast and Media Engineering, vol. 19, no. 6, Nov. 2014, pp. 819-825. https://doi.org/10.5909/JBE.2014.19.6.819
  4. S. Lee, "Research on Reducing Background Sounds in Mixed Speech Signals," Master Thesis, Dongeui University, 2018.
  5. R. Parekh, J. Yang, and V. Honavar, "Constructive neural network learning algorithm for pattern classification," J. of IEEE Transactions on Neural Networks, vol. 11, no. 2, 2000, pp. 436-451. https://doi.org/10.1109/72.839013
  6. D. Liu, T. Chang, and Y. Zhang "A new learning algorithm for feed forward neural networks," In Proceedings of IEEE International Symposium on Intelligent Control, Mexico City, Mexico, 2001, pp. 39-44.
  7. G. Son and J. Park, "A Study of Kernel Characterisitics of CNN Deep Learning for Effective Fire Detection Based on Video," J. of the Korea Institute of Electronic Communication Science, vol. 13, no. 6, 2018, pp. 1257-1262. https://doi.org/10.13067/JKIECS.2018.13.6.1257
  8. J. Park, M. Wiranegara, and G. Son, "Multi-channel Video Analysis Based on Deep Learning for Video Surveillance," J. of the Korea Institute of Electronic Communication Science, vol. 13, no. 6, 2018, pp. 1263-1268. https://doi.org/10.13067/JKIECS.2018.13.6.1263
  9. V. Vapnik, "An overview of statistical learning theory," J. of IEEE Transactions on Neural Networks, vol.10, issue 5, 1999, pp. 988-999. https://doi.org/10.1109/72.788640
  10. M. M. Adankon, M. Cheriet, and A. Biem, "Semisupervised Learning Using Bayesian Interpretation : Application to LS-SVM," J. of IEEE Transactions on Neural Networks, vol. 22, issue. 4, 2011, pp. 513-524. https://doi.org/10.1109/TNN.2011.2105888
  11. S. Lee, J. Park, and H. Kim, "Reduction of background sound on non-negative matrix factorization combined with Wiener filter post-processing," Proceedings of 18th International Conference on Control, Auto-mation and Systems(ICCAS2018), Pyeongchang, South Korea, Oct. 2018.