An application of lossy compression of the echolocation signals of toothed whale

More and more attentions have been paid to the research of echolocation signals of tooth whales these years. However, due to the high-frequency characteristics of the signals, the sampling rate of many of the experimental data can be up to 500 kHz or even 2000 kHz, which lead to a huge amount of experimental data, this bring some difficulties to the follow-up storage and analysis work. However, due to the difficulty of the audio signals itself, there are few studies on the data compression of the echolocation signals. In this paper, an F-shift lossy compression algorithm based on L∞ criterion, which is developed from Haar wavelet transform, is used to compress and reconstruct the measured echolocation signals. By setting different Δ values, we can get different compression ratios. The signal to noise ratio (SNR) and percent error are used to calculate the reconstruction error. Different click parameters, such as peak frequency, center frequency, 3dB bandwidth, are also calculated to evaluate the performance of the F-shift algorithm. The experiments suggest that the performance of this algorithm is related to the Δ values and the peak-to-peak values of click signals. In addition, data compression has the least effect on the center frequency parameter of the echolocation signals. When the Δ value is set to be 3 dB above the root mean square (RMS) value of noise, the compression ratio is about 100, and the reconstruction error of the parameters of the echolocation signals can be bounded within 10%. This experiment suggests that the F-shift algorithm does have a certain application prospect.