Panda sound event detection method and system under mixed audio

A technology of event detection and audio mixing, which is applied in voice analysis, instruments, etc., can solve the problems of giant panda call types, large differences in acoustic characteristics, and different call durations, so as to achieve good detection results and improve The effect of recognition and high detection accuracy

Active Publication Date: 2021-05-14
SICHUAN UNIV +1
View PDF11 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Aiming at the defects in the prior art, the present invention provides a giant panda sound event detection method and system under mixed audio, which solves the problem that due to the changeable types of giant panda calls, the large difference in acoustic characteristics, and the different lengths of calls The problems caused, and improve the recognition and detection efficiency of giant panda calls

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Panda sound event detection method and system under mixed audio
  • Panda sound event detection method and system under mixed audio
  • Panda sound event detection method and system under mixed audio

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0032] like Figure 1-2 Shown:

[0033] 1. Extract the mel spectrum

[0034] The original audio signal needs to go through two steps to obtain the logarithmic Mel spectral features: short-time Fourier transform and Mel filtering.

[0035] 1. Short-time Fourier transform

[0036] The short-time Fourier transform can convert time-domain signals into frequency-domain signals. The steps of the short-time Fourier transform are framing, windowing, and Fourier transform. Framing allows a long time sequence signal to be divided into multiple short time unit signals. However, to ensure that the features are independent and related to each other, there must be overlap between frames, that is, frame shift. Each frame also needs to be windowed. The purpose of the windowing feature is to attenuate the side lobes, eliminate high-frequency interference and leakage energy, and be more periodic. The window function is:

[0037]

[0038] N is the number of frames, and each frame will b...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a panda sound event detection method and system under mixed audio. The panda sound event detection method comprises the following steps: acquiring audio data of a detected environment; extracting a logarithmic Mel spectrum from the audio data, and standardizing the logarithmic Mel spectrum; dividing the processed audio data into development set data and test set data; building a multi-scale attention convolutional recurrent neural network; training the multi-scale attention convolutional recurrent neural network through the development set data; and predicting the test set data through the trained multi-scale attention convolutional recurrent neural network, and generating a prediction result; According to the method, the multi-scale attention convolution module is utilized to extract the multi-scale feature information, and the recurrent neural network is utilized to fully utilize the context information, so that the method has relatively high detection precision for the giant panda sound, and has a very good detection effect for the giant panda sound with different cry types and cry durations.

Description

technical field [0001] The invention relates to the technical fields of audio detection technology, sound collection, recognition and detection, and in particular to a method and system for detecting giant panda sound events under mixed audio. Background technique [0002] Giant panda, English name Giant Panda, scientific name Ailuropoda melanoleuca, commonly known as "panda" or "cat bear", is a mammal belonging to the family Ursidae of the order Carnivora, with black and white body colors. Pandas are endemic to China, and their main existing habitats are the mountainous areas around the Sichuan Basin in central and western China and the Qinling Mountains in southern Shaanxi. There are about 2,060 wild giant pandas in the world (2016 data). Due to the low fertility rate, the giant panda is rated as an endangered species in the China Red Data Book of Endangered Animals and is a national treasure of China. The ancestors of giant pandas first appeared 2-3 million years ago an...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G10L17/26G10L17/18G10L25/18
CPCG10L17/18G10L17/26G10L25/18
Inventor 赵启军汤茂林陈鹏侯蓉闫蔚然郭龙银张艳秋刘鹏张珊
Owner SICHUAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products