Dynamic Time Warping for spectrogram Alignment and AVErage sequence computation
DTWave is a tool for analysing sound sequences, it allows to
– measure a pairwise distance between two or more sounds,
– compute an average sequence from a pair of sounds.
Currently, DTWave reads and writes HTK file format (http://htk.eng.cam.ac.uk/) but future releases will be format independent
Primarily designed for bioacoustics, DTWave can also be used for any kind of sequence data.
DTWave is entirely written in C and is therefore relatively cross-platform independent.
DTWave is provided as is, with no warranty. The developers, the Bioinformatics Institute, The University of Auckland, and any other associated parties, provide no warranty for the use of DTWave. While every effort is taken to provide a reliable and trustworthy application, the authors, and associated parties, accept no liability for any damages, loss, or inconvenience caused by the use of DTWave. Comments, suggestions, queries, and bug reports are encouraged though – please feel free to email the author (firstname.lastname@example.org).
Download compiled program for windows, linux and mac: dtwave.tar.gz
To cite the program, please refer to:
- Louis Ranjard, Michael G. Anderson, Matt J. Rayner, Robert B. Payne, Ian McLean, James V. Briskie, Howard A. Ross, Dianne H. Brunton, Sarah M. N. Woolley, Mark E. Hauber, Bioacoustic distances between the begging calls of brood parasites and their host species: a comparison of metrics and techniques, Behavioral Ecology and Sociobiology, 2010. http://dx.doi.org/10.1007/s00265-010-1065-2
A technical description of the method is available here:
- Louis Ranjard, Howard A. Ross, Unsupervised bird song syllable classification using evolving neural networks, Journal of the Acoustical Society of America 123(6):4358-4368, 2008