FinchCatcher is a new software package for the acoustic analysis of natural sounds, specifically designed for pure-tone bird songs. It is especially useful for field recordings with ambient noise, and is convenient for large sample sizes. FinchCatcher was created and is maintained by Dr. Chenghui Ju. FinchCatcher is a copyrighted product of the Lahti Lab at the Department of Biology, Queens College, City University of New York.
- Separates acoustic signals from background noise, minimizing the influence of varied acoustic backgrounds on measurements and facilitating unbiased measurements of acoustic features across recordings
- Uses feature extraction to describe signals, emphasizing shape-related features; in addition to using commonly used features pertaining to simple aspects of shapes such as syllable duration and frequency bandwidth, FinchCatcher also introduces more complex derivations like the number of frequency-time inflection points and frequency-time excursion length
- Employs dynamic time warping (DTW) to derive a general similarity score between signals; DTW searches for the optimal alignment between two time series regardless of how they may have been warped in the time domain
- Introduces a dynamic tree-cut procedure for classification of signals or signal elements
- Permits a range of user customization (and choices will continue to increase with following versions)
The current version of FinchCatcher is Beta 5, introduced in December 2016.
Ju, C. (2016). FinchCatcher (Version Beta 5) [Computer software]. New York, NY: Queens College Department of Biology. Available from http://finchcatcher.net.
Ju, C. and D. C. Lahti (2016). FinchCatcher Manual: Version Beta 5. New York, NY: Queens College Department of Biology.