Artificial neural network discrimination of black-capped chickadee (Poecile atricapillus) call notes

Carly M. Nickerson, Laurie L. Bloomfield, Michael R. W. Dawson, Christopher B. Sturdy
2006 Journal of the Acoustical Society of America  
Artificial neural networks were trained to discriminate between two different notes from the "chick-a-dee" call of the black-capped chickadee ͑Poecile atricapillus͒. An individual note was represented as a vector of nine summary features taken from note spectrograms. A network was trained to respond to exemplar notes of one type ͑e.g., A notes͒ and to fail to respond to exemplar notes of another type ͑e.g., B notes͒. After this training, the network was presented novel notes of the two
more » ... of the two different types, as well as notes of the same two types that had been shifted upwards or downwards in frequency. The strength of the response of the network to each novel and shifted note was recorded. When network responses were plotted as a function of the degree of frequency shift, the results were very similar to those observed in birds that were trained in an analogous task ͓Charrier et al., J. Comp. Psychol. 119͑4͒, 371-380 ͑2005͔͒. The implications of these results to simulating behavioral studies of animal communication are discussed.
doi:10.1121/1.2211509 pmid:16938997 fatcat:67776tztwzgd7c6wgjyhtvnlxq