Right: Stream networks and data collection sites in H.J.A.

Oregon State Univeristy Bioacoustics Group

This summer, Matthew Betts, Sarah Frey and Adam Hadley deployed 15 Songmeters to record bird sounds in H.J. Andrews. These devices collected over 1 terabyte of audio data, 20 minutes per hour, 24 hours a day, at sites with varying altitutes and vegetation.

The challenge now is to develop algorithms to automatically identify which species of birds are present in an audio recording. Solving this problem will allow us to create maps of bird activity at an unprecedented temporal resolution. Raviv Raich, Xiaoli Fern, Forrest Briggs, and Balaji Lakshminarayanan have been working on the problem of automatic species recognition from audio data. We developed accurate and efficient algorithms for Bayesian classification of audio (see publications below). Audio data from H.J.A. poses many challenges for machine learning, such as noise from streams, wind and vehicles, and multiple species of birds vocalizing simultaneously. Future work will include statistical models for audio classification, classification in noisy environments with multiple sound sources, scaling machine learning algorithms to large data sets, and active learning (to minimize human labeling effort).

• Faculty

  • Matthew Betts, Assistant Professor, Forest Wildlife Landscape Ecology, OSU
  • Raviv Raich, Assistant Professor, School of Electrical Engineering and Computer Science, OSU
  • Xiaoli Fern, Assistant Professor, School of Electrical Engineering and Computer Science, OSU

• Students

  • Sarah Frey, Ph.D. student in Ecology, OSU
  • Adam Hadley, Ph.D. student in Ecology, OSU
  • Forrest Briggs, Ph.D. student in CS, OSU
  • Balaji Lakshminarayanan, M.S. student in EECS, OSU

• Collaborators

  • Dave Mellinger, Associate Professor, Senior Research Cooperative Institute for Marine Resources Studies, Hatfield Marine Science Center, OSU
  • Jed Irvine, Faculty Research Assistant, OSU

• Forrest Briggs, Raviv Raich, and Xiaoli Z. Fern, "Audio Classification of Bird Species: a Statistical Manifold Approach", to appear in Proc. International Conference on Data Mining, ICDM 2009, (pdf)
  
  
• B. Lakshminarayanan, R. Raich, and X. Fern, "A syllable-level probabilistic framework for bird species identification," to appear in Proc. IEEE International Conference on Machine Learning and Applications, 2009. (pdf)
  
  

• Forrest Briggs, Xiaoli Z. Fern, and Raviv Raich, "Acoustic Classification of Bird Species from Syllables: an Empirical Study," 2009, (pdf)