Contents lists available at ScienceDirect
Applied Acoustics
journal homepage: www.elsevier.com/locate/apacoust
Clicks classification of sperm whale and long-finned pilot whale based on
continuous wavelet transform and artificial neural network
Jia-jia Jiang
a,b,1
, Ling-ran Bu
a,b,1,
⁎
, Xian-quan Wang
a,b
, Chun-yue Li
a,b
, Zhong-bo Sun
a,b
,
Han Yan
c
, Bo Hua
c
, Fa-jie Duan
a,b
, Jian Yang
d
a
State Key Lab of Precision Measuring Technology and Instruments, Tianjin University, 92 Weijin Road, Nankai District, Tianjin, China
b
The Key Laboratory of Micro Opto-electro Mechanical System Technology, Tianjin University, 92 Weijin Road, Nankai District, Tianjin, China
c
Systems Engineering Research Institute, China State Shipbuilding Corporation (CSSC), 5 Yuetanbei Street Xicheng District, Beijing, China
d
School of Electronic Engineering, Xidian University, Xi’an, China
ABSTRACT
Passive acoustic observation of whales is an increasingly important tool for whale research. Clicks are the
predominant vocalizations of toothed whales, such as sperm whales and long-finned pilot whales. Classifying
clicks of sperm whales and long-finned pilot whales is an essential task for the passive acoustic observation of the
two whale species, especially in the case that both whale species vocalize in the same observed area. In this
paper, we proposed a method performing the automated classification of clicks produced by sperm whales and
long-finned pilot whales. First, the two types of whales’ original sounds were denoised using a wavelet denoising
method. Then, a dual-threshold endpoint detection algorithm was utilized to detect and pick out all clicks from
the denoised sounds. The continuous wavelet transform was applied to decompose the picked clicks, and a
wavelet coefficient matrix can be obtained for each picked click. Focusing on the energy distribution and
duration difference between the two types of whales’ clicks, we proposed a feature-vector extraction algorithm
based on the wavelet coefficient matrix. For each picked click, scale (frequency) features and time feature were
obtained respectively and they were used to form the feature vector. Finally, a back propagation (BP) neural
network was designed as a classifier of feature-vector to output final classification result. The experiment results
show the proposed method can obtain high classification performances. The effect of training dataset size, and
the number of training features on the classification performance was also examined in the experiments.
1. Introduction
In recent years, passive acoustic observation has gained more and
more attention in the field of whale species research [1]. Passive
acoustic observation, which only captures sounds from the surrounding
environment of the observation instrument, can be used to monitor
whales in a non-invasive manner [1]. Compared with visual observa-
tion methods, passive acoustic observation has a better monitoring
performance. In addition, it can continue at night, in poor weather, and
under other conditions in which visual observation cannot. The method
can be used to measure the range and seasonal occurrence of whale
species [2], to estimate the abundance of a species in a given area [3],
to determine the population structure [4] and so on. For the above
applications, a necessary precondition is to identify which species
produce a given sound, particularly under the conditions that the target
species are difficult to identify visually [5]. Correctly classifying various
whale sounds into corresponding whale species, which is an essential
and primary task for passive acoustic observation applications, can
assist observers to determine the species composition of the observed
whale and further to determine whether or not the observation in-
strument turns towards the target species or continues to observe the
target whale [6].
Generally, whale sounds can be classified into many sub-categories,
such as whistles, pulses, and clicks [7]. Whale clicks are important
vocalizations that are widespread in a variety of whales [7,8]. More-
over, whale clicks are considered as the predominant vocalizations of
toothed whales [7,9]. Sperm whales (Physeter macrocephalus) and
long-finned pilot whales (Globicephala melas) are two typical toothed
whale species that can produce multiple clicks which can be applied for
navigation, prey detection, and communication. [7,8,10–12]. Due to
the unique physical structures of vocal organs, sperm whales are
thought to produce only clicks [7,13] Moreover, according to the
https://doi.org/10.1016/j.apacoust.2018.06.014
Received 15 March 2018; Received in revised form 7 June 2018; Accepted 15 June 2018
⁎
Corresponding author at: State Key Lab of Precision Measuring Technology and Instruments, Tianjin University, 92 Weijin Road, Nankai District, Tianjin, China.
1
Jia-jia Jiang and Ling-ran Bu contributed equally to this work and should be considered co-first authors.
E-mail address: lingranbu@tju.edu.cn (L.-r. Bu).
Applied Acoustics 141 (2018) 26–34
0003-682X/ © 2018 Elsevier Ltd. All rights reserved.
T