JRM Vol.29 No.1 pp. 236-246
doi: 10.20965/jrm.2017.p0236


Bird Song Scene Analysis Using a Spatial-Cue-Based Probabilistic Model

Ryosuke Kojima*1, Osamu Sugiyama*1, Kotaro Hoshiba*2, Kazuhiro Nakadai*2,*3, Reiji Suzuki*4, and Charles E. Taylor*5

*1Graduate School of Information Science and Engineering, Tokyo Institute of Technology
2-12-1 Ookayama, Meguro-ku, Tokyo 152-8552, Japan

*2Department of Systems and Control Engineering, School of Engineering, Tokyo Institute of Technology
2-12-1 Ookayama, Meguro-ku, Tokyo 152-8552, Japan

*3Honda Research Institute Japan Co., Ltd.
8-1 Honcho, Wako, Saitama 351-0114, Japan

*4Graduate School of Information Science, Nagoya University
Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8601, Japan

*5Department of Ecology and Evolutionary Biology, University of California, Los Angeles (UCLA)
Los Angeles, CA 90095, USA

July 23, 2016
December 1, 2016
February 20, 2017
bird song identification, robot audition, scene understanding, probabilistic model

Bird Song Scene Analysis Using a Spatial-Cue-Based Probabilistic Model

Spatial-cue-based probabilistic model

This paper addresses bird song scene analysis based on semi-automatic annotation. Research in animal behavior, especially in birds, would be aided by automated or semi-automated systems that can localize sounds, measure their timing, and identify their sources. This is difficult to achieve in real environments, in which several birds at different locations may be singing at the same time. Analysis of recordings from the wild has usually required manual annotation. These annotations may be inaccurate or inconsistent, as they may vary within and between observers. Here we suggest a system that uses automated methods from robot audition, including sound source detection, localization, separation and identification. In robot audition, these technologies are assessed separately, but combining them has often led to poor performance in natural setting. We propose a new Spatial-Cue-Based Probabilistic Model (SCBPM) for their integration focusing on spatial information. A second problem has been that supervised machine learning methods usually require a pre-trained model, which may need a large training set of annotated labels. We have employed a semi-automatic annotation approach, in which a semi-supervised training method is deduced for a new model. This method requires much less pre-annotation. Preliminary experiments with recordings of bird songs from the wild revealed that our system outperformed the identification accuracy of a method based on conventional robot audition.*
* This paper is an extension of a proceeding of IROS2015.

Cite this article as:
R. Kojima, O. Sugiyama, K. Hoshiba, K. Nakadai, R. Suzuki, and C. Taylor, “Bird Song Scene Analysis Using a Spatial-Cue-Based Probabilistic Model,” J. Robot. Mechatron., Vol.29, No.1, pp. 236-246, 2017.
Data files:
  1. [1] T. Otsuka, K. Ishiguro, H. Sawada, and H. G. Okuno, “Bayesian nonparametrics for microphone array processing,” T-ASLP, Vol.22, No.2, pp. 493-504, 2014.
  2. [2] X. Anguera, S. Bozonnet, N. Evans, C. Fredouille, G. Friedland, and O. Vinyals, “Speaker diarization: A review of recent research,” IEEE Trans. on Audio, Speech, and Language Processing, Vol.20, No.2, pp. 356-370, 2012.
  3. [3] J. M. Pardo, X. Anguera, and C. Wooters, “Speaker diarization for multiple distant microphone meetings: mixing acoustic features and inter-channel time differences,” Proc. of the Ninth Int. Conf. on Spoken Language Processing, pp. 2194-2197, 2006.
  4. [4] C. K. Catchpole and P. J. Slater, “Bird song: biological themes and variations,” Cambridge University Press, 2003.
  5. [5] F. Briggs, R. Raich, K. Eftaxias, Z. Lei, and Y. Huang, “The ninth annual MLSP competition: overview,” IEEE Int. workshop on machine learning for signal processing, pp. 22-25, Sept. 2013.
  6. [6] H. Goëau, H. Glotin, W. P. Vellinga, R. Planqué, and A. Joly, “Life-CLEF Bird Identification Task 2016,” CLEF working notes 2016, 2016.
  7. [7] K. N. R. Suzuki, S. Matsubayashi, and H. G. Okuno, “Localizing bird songs using an open source robot audition system with a microphone array,” Proc. of Interspeech 2016, pp. 2026-2030, 2016.
  8. [8] K. Ryosuke, S. Osamu, S. Reij, N. Kazuhiro, and C. E. Taylor, “Semi-automatic bird song analysis by spatial-cue-based integration of sound source detection, localization, separation, and identification,” IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, 2016.
  9. [9] R. O. Schmidt, “Multiple emitter location and signal parameter estimation,” IEEE Trans. on Antennas and Propagation, Vol.34, No.3, pp. 276-280, 1986.
  10. [10] P. Aarabi, “The fusion of distributed microphone arrays for sound localization,” Eurasip J. on Applied Signal Processing, Vol.2003, No.4, pp. 338-347, 2003.
  11. [11] J. M. Valin, F. Michaud, and J. Rouat, “Robust 3D localization and tracking of sound sources using beamforming and particle filtering,” 2006 IEEE Int. Conf. on Acoustics Speech and Signal Processing Proc., Vol.4, pp. IV-IV, 2006.
  12. [12] C. V. Cotton and D. P. Ellis, “Spectral vs. spectro-temporal features for acoustic event detection,” WASPAA-2011, pp. 69-72, 2011.
  13. [13] Y. Ohishi, D. Mochihashi, T. Matsui, M. Nakano, H. Kameoka, T. Izumitani, and K. Kashino, “Bayesian semi-supervised audio event transcription based on Markov indian buffet process,” ICASSP-2013, pp. 3163-3167, 2013.
  14. [14] M. L. Chin and J. J. Burred, “Audio event detection based on layered symbolic sequence representations,” 2012 IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP), pp. 1953-1956, 2012.
  15. [15] D. Rybach, R. Schlüter, and H. Ney, “Silence is golden: Modeling non-speech events in WFST-based dynamic network decoders,” ICASSP-2012, pp. 4205-4208, 2012.
  16. [16] Y. Sasaki, M. Kaneyoshi, S. Kagami, H. Mizoguchi, and T. Enomoto, “Daily sound recognition using pitch-cluster-maps for mobile robot audition,” IROS-2009, pp. 2724-2729, 2009.
  17. [17] C. Baugé, M. Lagrange, J. Andén, and S. Mallat, “Representing environmental sounds using the separable scattering transform,” ICASSP-2013, pp. 8667-8671, 2013.
  18. [18] V. Ramasubramanian, R. Karthik, S. Thiyagarajan, and S. Cherla, “Continuous audio analytics by HMM and viterbi decoding,” ICASSP-2011, pp. 2396-2399, 2011.
  19. [19] K. Nakamura and K. Nakadai, “Robot audition based acoustic event identification using a Bayesian model considering spectral and temporal uncertainties,” 2015 IEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), pp. 4840-4845, 2015.
  20. [20] P. W. Holland, K. B. Laskey, and S. Leinhardt, “Stochastic block-models: First steps,” Social networks, Vol.5, No.2, pp. 109-137, 1983.
  21. [21] K. Nigam, A. K. McCallum, S. Thrun, and T. Mitchell, “Text classification from labeled and unlabeled documents using EM,” Machine learning, Vol.39, No.2-3, pp. 103-134, 2000.
  22. [22] S. Uemura, O. Sugiyama, R. Kojima, and K. Nakadai, “Outdoor Acoustic Event Identification using Sound Source Separation and Deep Learning with a Quadrotor-Embedded Microphone Array,” ICAM2015, pp. 329-330, 2015.
  23. [23] H. Nakajima, K. Nakadai, Y. Hasegawa, and H. Tsujino, “Correlation matrix estimation by an optimally controlled recursive average method and its application to blind source separation,” Acoustical Science and Technology, Vol.31, No.3, pp. 205-212, 2010.
  24. [24] A. Banerjee, I. S. Dhillon, J. Ghosh, and S. Sra, “Clustering on the unit hypersphere using von Mises-Fisher distributions,” J. of Machine Learning Research, Vol.6, pp. 1345-1382, 2005.
  25. [25] S. Sra, “A short note on parameter approximation for von Mises-Fisher distributions: and a fast implementation of Is(x),” Computational Statistics, Vol.27, No.1, pp. 177-190, 2012.
  26. [26] K. Nakadai, H. G. Okuno, H. Nakajima, Y. Hasegawa, and H. Tsujino, “An open source software system for robot audition HARK and its evaluation,” Humanoid Robots, 2008. Humanoids 2008. 8th IEEE-RAS Int. Conf. on, pp. 561-566, 2008.
  27. [27] G. Schwarz et al., “Estimating the dimension of a model,” The annals of statistics, Vol.6, No.2, pp. 461-464, 1978.

*This site is desgined based on HTML5 and CSS3 for modern browsers, e.g. Chrome, Firefox, Safari, Edge, IE9,10,11, Opera.

Last updated on Nov. 12, 2018