We have been developing an automatic classification system for bird vocalisations. Many biologists have been using the early one-dimensional version of our system and we have been working on a two-dimensional method. The software extracts a spectrogram from the bird vocalisation using the LPC spectrum analysis and classifies the images of spectrogram using a similarity scale and cluster analysis. We use the new similarity scale called the "Two-dimensional Geometric Distance" that has been developed by Jinnai and Boucher. In this paper, we introduce the principles of the Two-dimensional Geometric Distance, demonstrate the two-dimensional pattern matching software, and describe design considerations in a new automatic classification system for bird vocalisations. Testing has shown an order of magnitude improvement in accuracy over the one-dimensional method.
|Title of host publication||Proceedings of 20th International Congress on Acoustics, ICA 2010|
|Number of pages||7|
|Publication status||Published - 1 Dec 2010|
|Event||20th International Congress on Acoustics 2010, ICA 2010 - Incorporating the 2010 Annual Conference of the Australian Acoustical Society - Sydney, NSW, Australia|
Duration: 23 Aug 2010 → 27 Aug 2010
|Conference||20th International Congress on Acoustics 2010, ICA 2010 - Incorporating the 2010 Annual Conference of the Australian Acoustical Society|
|Period||23/08/10 → 27/08/10|