Research to identify birds by their songs
The goal of this research being performed by UN engineers is to improve bird classification and recognition of new species by means of bird song computer analysis.Manizales, 11 de abril de 2012 — Agencia de Noticias UN-
One of the largest natural riches of Colombia may be seen soaring the skies, particularly in the coffee growing region of the country. This area harbors almost 10% of the bird species of the world.
Keeping this ecological diversity in mind and following processes such as those developed by the Automated Remote Biodiversity Monitoring Network - Arbimon, the UN Processing and Signal Recognition Group in Manizales has carried out a bird acoustic characterization study to create software adaptable to the biodiversity of this region.
“ The Arbimon was formed at the Universidad de Puerto Rico. Based on their model, and in cooperation with Computer Science, Biology and Ecology professionals, we wanted to implement new automated signal recognition and processing techniques”, says Mauricio Orozco Alzate, Associate Professor and research coordinator in Manizales.
This is a new technology which attempts to mimic human reasoning. It consists of directly comparing two input sources, without measuring any of their particular properties. For example, you might recognize a person by a certain physical trait, such as the face, without stopping to look at the color of the skin, eyes or hair.
Therefore, when the system “hears” a particular bird song, it classifies the acoustic signal based on previous database inputs. The database has 538 bird songs from 11 different species, all taken from the Natural Rio Blanco National Reserve in Manizales. The system has a success rate of 97.87% at identifying species.
This figure shows the potential of the process, “as it has only failed at correctly identifying two species which have very similar sounding songs”, says José Francisco Ruiz Muñoz, who developed the project as part of his graduate work for his Masters degree in industrial automation engineering.
Implementation of this method is only a starting point for a larger scale project. A larger future project might be continuous automated field monitoring of species.
Read the complete article at http://www.unperiodico.unal.edu.co/dper/article/nueva-tecnica-para-cazar-el-trino-de-las-aves/index.html(Por: Fin/amej/sup/vbr