chirp is a set of related tools for analyzing and comparing bioacoustic recordings. It includes the following components:
A pitch-detection algorithm that tracks the pitch of a signal as a function of time. A number of features enable the algorithm to make stable estimates even from very noisy signals. Performance is greatly improved by masking out noise in the spectrotemporal domain.
Algorithms for making pairwise comparisons of signals, including spectrographic cross-correlation, spectrographic dynamic time warping (DTW), and pitch-based DTW.
Command-line scripts for batch, multithreaded analyses.
A graphical user interface for examining recordings and creating masks. This program can run batch pitch analyses and perform pairwise comparisons using pitch and other methods for large sets of recordings.
For more information, visit the project website
Installation and use instructions can be found on the wiki