Abstract
This paper presents a method for automatic classification of birdsong notes. The elaborate method performs correlation calculationsapplied to spectrographic images to determine the similarity between the notes of a vocal repertoire, so the method was called SITM (Spectrographic Image Template Matching). The notes Nj of each phrase Pi which makes up the song of an individual of aspecie, are generally classified with letters of the alphabet according to the sequence emitted that is viewed in a spectrogram.By setting a minimum degree of similarity between the notes of the repertoire of each individual, it is possible to automate thistype of visual classification. Performance evaluation of automatic classification was performed with the vocal repertoire of threeindividuals of T. ruviventris recorded in to the wild. With 96% confidence, according to statistical inference, the rate of correctclassification via SITM was 77% to 97% of the notes in recording FNJV5915, 64% to 82% of the notes in recording FNJV5932and 74% to 97% of notes in recording FNJV5955.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
