Unlock instant, AI-driven research and patent intelligence for your innovation.

Sound event classification method and system based on dynamic selection of time-frequency matrix

A classification method and time-frequency technology, applied in speech analysis, computer parts, instruments, etc., can solve problems such as large amount of calculation, inability to select time-frequency matrix for sound signals, complexity, etc., to increase high-frequency resolution, improve Feature extraction algorithm, simple and easy to implement effect

Active Publication Date: 2022-03-01
SHANDONG UNIV
View PDF14 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The inventors found that in the existing image signal feature extraction method, the pure sound and the noise-containing sound adopt different feature extraction processes, which is relatively complicated; and all the sound signals are uniformly adjusted into a fixed-size time-frequency matrix, which cannot be targeted Selecting a suitable time-frequency matrix for sound signals of different types and lengths requires a large amount of calculation

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
  • Sound event classification method and system based on dynamic selection of time-frequency matrix
  • Sound event classification method and system based on dynamic selection of time-frequency matrix
  • Sound event classification method and system based on dynamic selection of time-frequency matrix

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] Sound scene recognition and classification has various applications, including smart environment scene recognition, sound-based monitoring systems, home automation, machine hearing, etc. This sound event classification method can be applied to environmental scene recognition applications, various sound detection systems, home automation and other applications, and realizes corresponding operations in combination with specific application scenarios. Taking sound monitoring through sound event classification as an example, the monitoring system recognizes and classifies all kinds of sounds collected in real time to judge dangerous events, and then can take countermeasures, such as gunshots and explosions, which will trigger the alarm system; For example, in the auditorium, the color and type of lights are automatically changed according to the recognized applause sound; and in home automation, each smart home automatically performs preset operations according to the recogn...

Embodiment 2

[0073] In one or more embodiments, a sound event classification system based on dynamic selection of a time-frequency matrix is ​​disclosed, comprising:

[0074] A device for collecting sound signal data in a set area environment and performing preprocessing of sound signal data;

[0075] A device for generating a spectrogram for the preprocessed sound signal;

[0076] A device for gradually reducing the original spectrogram to generate multiple time-frequency matrices of different sizes;

[0077] A device for calculating the similarity between each time-frequency matrix and the original spectrogram, finding the optimal time-frequency matrix, and converting the image into a graph signal;

[0078] A device for extracting feature events from image signals; sending the extracted feature events to a classifier to obtain a sound event classification result.

Embodiment 3

[0080] In one or more embodiments, a terminal device is disclosed, including a server, the server includes a memory, a processor, and a computer program stored on the memory and operable on the processor, and the processor executes the The program realizes the sound event classification method based on the dynamic selection of the time-frequency matrix in the first embodiment. For the sake of brevity, details are not repeated here.

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 sound event classification method and system based on time-frequency matrix dynamic selection, comprising: collecting sound signal data in a set area environment, and performing preprocessing of the sound signal data; for the preprocessed sound signal, generating The spectrogram; the original spectrogram is gradually reduced to generate multiple time-frequency matrices of different sizes; the similarity between each time-frequency matrix and the original spectrogram is found, and the optimal time-frequency matrix is ​​found. The frequency matrix is ​​converted into a graph signal; the feature event is extracted from the graph signal; the extracted feature event is sent to the classifier to obtain the classification result of the sound event. The beneficial effect of the present invention is that the feature extraction process is simplified, and at the same time, an appropriate dynamic threshold is set to ensure the integrity of the extracted features. The method based on the graph signal is used to calculate the similarity between two images, and the time-frequency matrix is ​​dynamically selected for each sound signal, and the high-energy spectrum information of the sound signal is preserved as much as possible while reducing the amount of calculation.

Description

technical field [0001] The invention belongs to the technical field of sound recognition and classification, in particular to a sound event classification method and system based on dynamic selection of a time-frequency matrix. Background technique [0002] There are various sound scenes around us in daily life, including the sound of objects colliding with each other, the sound of gas jets, the sound of falling particles, the sound of objects rubbing, various bells, horns, electronic musical instruments Sound, etc., the recognition and classification of non-linguistic sound events in real environments has attracted more and more attention. Sound scene recognition and classification has various applications, including smart environment scene recognition, sound-based monitoring systems, home automation, machine hearing, etc. [0003] Nowadays, speech recognition technology is very common, but there are relatively few technologies for recognizing and classifying sound scenes,...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G10L17/26G10L25/18G06K9/00
CPCG10L17/26G10L25/18G06F2218/08G06F2218/12
Inventor 魏莹刘迎港
Owner SHANDONG UNIV