Recognising Biological Sounds Using Machine Learning

The development of a software system which can detect and identify the ight calls of migrating birds is reported. The system rst produces a spectrogram using a Discrete Fourier transform. Calls are detected in the spectrogram using an ad-hoc combination of local peak-nding and a connectedness measure. Attributes are extracted both globally from the call and from a window moved incrementally through the call. Quinlan's C4.5 machine learning system is used to induce a decision tree-based classiier. The system has been tested on a set of 138 ight calls from 9 species of birds. Some calls are faint and interfering insect noise is present in others. Eight-fold re-sampling was used to classify the calls unseen. 78% of calls were identiied correctly, 4% incorrectly and 18% were left unclassiied.