This project will continue the development of computerized analysis of bioacoustic signals. Many animals use sound for communicating, and research on vocal learning in some non-human species provides models for understanding speech development in humans. Despite this importance, there are few rigorous, standardized methods for describing bioacoustic signals, or for quantitatively assessing and comparing animal vocalizations. A wide variety of techniques are used now, and many are very laborious, difficult to repeat, and often depend on subjective interpretation. This work will further develop computer-aided methods for quantitatively analyzing animal vocalizations. The research is designed on two related fronts. 1) Advanced, standardized signal processing techniques will be constructed, to serve as tools for the description, storage and comparison of such sounds. These techniques will provide a quantitative base for laboratory analysis of auditory perception of communication sounds, or for field data on vocal communication in a wide range of animals. 2) These techniques will be used to examine vocal development in a well-studied songbird that learns its specific song. Vocal learning in songbirds has become better understood recently because of new anatomical and neurochemical information. The specific pathways have been outlined between parts of the brain involved in song production and hearing, and specific hormones and neurochemicals have been found to affect the production and perceptions of the song. These techniques will allow the complex vocal sounds of the song to be analyzed as the individual components of the song develop during learning. This approach promises to provide a novel productive way of investigating complex vocal behavior during learning, and will have impact on the fields of speech and communication as well as auditory perception and bioacoustics.
