1 / 20

Dynamic time warping for automatic classification of killer whale vocalizations

Dynamic time warping for automatic classification of killer whale vocalizations. Judy Brown Patrick Miller. Physics Dept, Wellesley College Media-Lab MIT Sea Mammal Research Unit, University of St. Andrews. the Plan :.

leoncraig
Download Presentation

Dynamic time warping for automatic classification of killer whale vocalizations

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Dynamic time warping for automatic classification of killer whale vocalizations Judy Brown Patrick Miller Physics Dept, Wellesley College Media-Lab MIT Sea Mammal Research Unit, University of St. Andrews

  2. the Plan : • Data: Pitch contours from two sets of vocalizations previously classified perceptually • Marineland group – 9 call types • Northern resident whales - 7 call types • Digression: two-source calls • Dynamic time warping for automatic classification of pitch contours • Results

  3. Pitch contours of Marineland set

  4. Pitch contours of Northern resident set

  5. Example with low freq and high freq sources (slowed by 6)

  6. Digression: Mechanism for Two SourcesCentral portion of Call 2.1ms -> 475 Hz (slowed by 6) .16 ms -> 6.2 kHz .113 s

  7. SIMULATION (slowed by 6) LOW FREQUENCY COMPONENT 450 - 900 Hz HIGH FREQUENCY COMPONENT 5800 - 6500 Hz

  8. SIMULATION LOW FREQUENCY COMPONENT 450 - 900 Hz (slowed by 6) HIGH FREQUENCY COMPONENT 5800 - 6500 Hz (slowed by 6)

  9. Dynamic Time Warping : the GoalCompare contours and quantify the difference.

  10. Difference Matrix D[i,j] = |T(j) - Q(i)|

  11. Difference Matrix and Cost Matrix D[i,j] = |T(j) - Q(i)| Example: insertion=1 deletion = 0 M =

  12. Cost Matrix with Query and Target Dissimilarity is M(qmax, tmax)

  13. Warping Function=Minimum Path

  14. Dissimilarity Matrix n1 n2 n32 n33 n4 n5 n9

  15. Clustering DISTANCES MULTIDIMENSIONAL SCALING (mds) Distances Positions in coordinate-like Space CLUSTERING – Form groups with minimum distanceswithin a group

  16. Clustering Results n32 n4 n5 n2 n33 n9 n1

  17. Another Look at n2, n4, n5, n9

  18. Results NORTHERN RESIDENT CALLS MARINELAND CALLS

  19. CONCLUSIONS • UPSIDE • HIGH ACCURACY for SEPARATED CONTOURS • VALIDITY of COMPARISON to HUMAN CLASSIFICATION ? • DOWNSIDE • TIME-CONSUMING MEASUREMENT of the PITCH CONTOURS

  20. Brown, J. C., A. Hodgins-Davis, and P. J. O. Miller (2006). “Classification of vocalizations of killer whales using dynamic time warping’’ J. Acoust. Soc. Am. 119, EL34-EL40. Acknowledgements: JCB thanks Dr. Eamonn Keogh for his MATLAB code which was adapted for some of the Figures. PJOM is grateful for funding from WHOI’s Ocean Life Institute and a Royal Society fellowship.

More Related