Chirplet Transform Applied to Simulated and Real Blue Whale (Balaenoptera musculus) Calls
Chirplet transform performance to identify low-frequency blue whale calls is tested with simulations and observations from North-West Atlantic. The three different calls are simulated using linear or quadratic frequency sweeping chirps and a hanning window. The performance of Chirplet transform to accurately estimate the chirp parameters with or without noise is first assessed. Then the performance to classify the real vocalizations from the test dataset using the three features best estimated from the simulations is then assessed. The method has a high classification rate and appears promising to efficiently identify these blue whale signature vocalizations with a reduced number of parameters, even under low signal to noise ratios.
KeywordsBlue whale vocalizations chirplet transform feature extraction vector quantization noise
- 5.Mellinger, D.K., Stafford, K.M., Moore, S.E., Dziak, R.P., Matsumoto, H.: An overview of fixed passive acoustic observation methods for cetaceans. Oceanography 20(4), 36–45 (2007)Google Scholar
- 7.Mann, S., Haykin, S.: Adaptive Chirplet Transform: an adaptive generalization of the wavelet transform. Optical Engineering 31(6), 1243–1256 Google Scholar
- 9.J.C. O’Neill, DiscreteTFDs: Time-Frequency Analysis Software, http://tfd.sourceforge.net/