As discussed in Sections 3.1–3.9, computational bioacoustic research and technology development activities depend on the availability of manually selected and, in some cases, duly annotated audio data sets. These data sets serve different purposes, among which are the
1. creation of statistical models,
2. fine-tuning of various adjustable parameters of the automated technology,
3. evaluation of the overall system performance, and so on.
In Sections 4.1–4.4, we discuss in detail the content, design, and specifics of these audio data sets, which we refer to as acoustic libraries. In order to facilitate the exposition, we briefly mention the purpose and practical use of each of these acoustic libraries. Specifically, ...