The quantitative analysis of animal sounds is important both in basic studies of animal behavior and in applied contexts such as acoustic censusing of endangered animal populations. In recent years, the primary tools for such analyses have been computer programs that allow researchers to visualize signals and to make measurements of time, frequency, and amplitude parameters of individual sounds. Two limitations of these tools are that (1) they require the operator to manually select the sounds to be measured, and (2) they are not easily modified to incorporate new measurements or complex analysis algorithms. These restrictions have emerged as key impediments to the rapid analysis of large quantities of acoustic data. We propose to develop a software instrument for the analysis of animal acoustic signals that will overcome these limitations of existing packages. The proposed program - herein identified as "Raven" - will combine a user-friendly interface with advanced tools that enable users to (1) automate the processing of large quantities of acoustic data, (2) visually evaluate the performance of automatic acoustic data processing algorithms, and (3) extend and customize the capabilities of the program to incorporate new types of measurements and analyses. The key to automating and extending Raven's capabilities is the program's ability to communicate with external programs via a publicly defined interface. An external program, which can be written in a general-purpose programming language, can perform arbitrarily complex analyses of signals (e.g., selecting signals from background noise, classifying signal types). The results of this processing can then be used to direct measurements or interactive displays in Raven. Raven will be an improved data exploration and analysis utility, and will also catalyze the development of more sophisticated acoustic processing tools by providing biologists with insights into the operation of signal processing algorithms.
