A Cetacean Whistle Signal Detection Method Based on Mel Subband Spectral Entropy

Abstract

The invention provides a cetacean whistle signal detection method based on Mel subband spectrum entropy. From the perspective of spectral entropy, the present invention combines the Mel filter to realize the detection of the dolphin whistle signal by calculating the Mel sub-band spectral entropy, and further reduces the false alarm probability through the time threshold to improve the detection accuracy. The method of the invention is simple and easy, can realize the automatic detection and extraction of the whistle signal, and reduces the workload of manual extraction. The dolphin whistle signal is of great value for the research of bionic camouflage underwater acoustic communication, dolphin population classification and identification, etc. Therefore, the method of the present invention can help related researches such as bionic camouflage underwater acoustic communication and marine biology.

Description

technical field [0001] The invention relates to an underwater acoustic communication method, in particular to an animal whistle signal detection method. Background technique [0002] The dolphin whistle signal is the most important signal when dolphins communicate, and it is of great significance for passive acoustic detection of dolphins. However, in the actual collected sound signal, there may only be a few whistle signals, and the position of the whistle signal appears randomly, so how to detect whether there is a whistle signal in the sound signal and determine the position of the whistle signal, for the dolphin whistle Acoustic signal extraction and research are very important. [0003] The patent document whose publication number is CN104217722B discloses a method for extracting the time spectrum profile of a dolphin whistle signal. The method extracts the time spectrum profile curve of a dolphin whistle through short-time Fourier transform. However, this method is ba...

AI Technical Summary

Problems solved by technology

However, in the actual collected sound signal, there may only be a few whistle signals, and the position of the whistle signal appears randomly, so how to detect whether there is a whistle signal in the sound signal and determine the position of the whistle signal, for the dolphin whistle Acoustic signal extraction and research are very important

[0003] The patent document whose publication number is CN104217722B discloses a method for extracting the time spectrum profile of a dolphin whistle signal. The method extracts the time spectrum profile curve of a dolphin whistle through short-time Fourier transform. However, this method is based on the known dolphin whistle The acoustic signal segment is carried out, and how to detect and extract the whistle signal segment from the collected signal is not introduced; the patent document with the publication number CN107527625A discloses a dolphin whistle signal auditory feature based on the cochlear model in the bionic auditory system The extraction method, which is based on the Gammatone filter and the Meddis inner hair cell model, obtains the auditory features of the dolphin whistle signal, but this method is also based on known dolphin whistle signal fragments, and does not introduce the dolphin whistle signal detection and extraction Method; Publication No. discloses the communication device based on cetacean call pulse time length in the patent document of CN208257816U, and this device utilizes cetacean class call short-term energy spectrum to calculate cetacean class call pulse time length, but this device is mainly used for Communication, and the short-term energy spectrum is easily affected by noise to generate false alarms, but the device does not use the time threshold to further reduce the probability of false alarms

Images