Unit:
Κατεύθυνση Μουσική ΤεχνολογίαLibrary of the School of Philosophy
Author:
Koutsogiannis Georgios
Supervisors info:
Αναστασία Γεωργάκη, Αναπληρώτρια Καθηγήτρια, τμήμα Μουσικών Σπουδών - Ε.Κ.Π.Α.
Αρετή Ανδρεοπούλου, Επίκουρη Καθηγήτρια, τμήμα Μουσικών Σπουδών - Ε.Κ.Π.Α.
Βασίλης Κατσούρος, Ερευνητής Α', Ινστιτούτο Επεξεργασίας του Λόγου
Original Title:
birdMelos: Δημιουργία εφαρμογής για την αυτόματη αναγνώριση των φωνημάτων των πτηνών
Translated title:
birdMelos: Development of an application for automatic birdsong recognition
Summary:
Throughout the ages birdsong has influenced composers and musicians in various ways, while at the same time have been observed analogies between birds' “music” and human music. A major issue is whether in birdsong there is some other use than the biological one. The science of musicology seeks answers through its various fields: biomusicology, zoomusicology and ornithomusicology. The attempt to decode bird vocalizations is done through various scientific fields, whereas the contribution of technology to this effort constitutes a subject of discussion. During the last few years, companies in association with bioacoustics laboratories have introduced particular acoustical analysis software for bird calls, via the so-called automatic pattern recognition method, which identifies with both the rythmic and the melodic structure of birdsong, always in accordance with the species' biological classification. Although such applications are based on an entirely different methodology, they appear to have some common limitations. For the purposes of this work, the birdMelos novel application was developed to explore the musical parameters of the vocal communication in birds and give a tool for identifying the structure of the birdsong related to the specie. Its design was based on the MATLAB platform, using the MIRtoolbox 1.7.1 library, and focuses on three types whose calls have clear and distinct structures: otus scops, cuculus canorus and streptopelia decaocto. BirdMelos is capable of performing sound analysis on any given sample - stereo or mono - and identifying the species, while also implementing custom noise reduction filters to improve the classification rate performance.
Main subject category:
Technology - Computer science
Keywords:
Acoustic ecology, biomusicology, bird vocalization, birdsong recognition, machine learning, music technology, sound analysis
