Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

GMM-UBM-based impact sound model establishment method and system, and impact sound detection method and system

A GMM-UBM and model building technology, applied in the field of audio signal acquisition and processing, can solve problems such as difficult detection of impact sound

Pending Publication Date: 2021-05-18
西安合谱声学科技有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The object of the present invention is to provide a GMM-UBM-based impact sound model establishment, impact sound detection method and system, to solve the problem in the prior art that the impact sound is difficult to detect in a noisy environment

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
  • GMM-UBM-based impact sound model establishment method and system, and impact sound detection method and system
  • GMM-UBM-based impact sound model establishment method and system, and impact sound detection method and system
  • GMM-UBM-based impact sound model establishment method and system, and impact sound detection method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0050]In this embodiment, a method of establishing an impact acoustic model based on GMM-UBM includes the following steps:

[0051]Step 1: Get the non-impact sound signal sample dataset, calculate the MFCC parameter of each non-impact sound signal to obtain an MFCC feature vector of the non-impact sound signal;

[0052]Step 2: Establish a UBM model according to the MFCC feature vector of the non-rinc-resistant sound signal, use the maximum desired algorithm to train UBM models, obtain the training of UBM models and training good model parameters, will train the UBM model as a background sound model;

[0053]Step 3: Get the impact sound signal sample data set, calculate the MFCC parameter of each impact sound signal to obtain the MFCC feature vector of the impact sound signal;

[0054]Step 4: Establish a GMM model based on the MFCC feature vector of the impact sound signal, and the model parameters of step 2 training are passed to the GMM model through the maximum sub-test estimation method, obt...

Embodiment 2

[0094]In this embodiment, an impact acoustic model established system based on GMM-UBM, including sample acquisition modules, model establishment modules, training modules, and parameter transfer modules;

[0095]The sample acquisition module is used to obtain non-impact sound signal sample data sets and impact sound signal sample data sets;

[0096]The model establishment module is used to calculate the MFCC parameter of each non-impact sound signal, obtain the MFCC feature vector of the non-impact sound signal, and is also used to calculate the MFCC parameter of each impact sound signal to obtain an MFCC feature vector of the impact sound signal. The UBM model is established according to the MFCC feature vector of the non-rincphone signal, and the GMM model is established according to the MFCC feature vector of the impact sound signal;

[0097]The training module is used to train UBM models using the maximum desired algorithm to obtain a well-trained UBM model and a well-trained model para...

Embodiment 3

[0108]Such asfigure 1 As shown, this embodiment provides an impact sound model establishment method, including:

[0109]Step 110: Get the audio signal sample picked up by the microphone; divide the sample into two categories, a type of audio sample signal of the non-shock sound, a type of audio sample signal of the impact;

[0110]Step 120: UBM sample training, the non-impact sound audio sample signal as a sample of UBM training, such asimage 3 Indicated;

[0111]Feature extraction of all audio samples of UBM training, such asfigure 2 As shown, the feature extraction includes: pre-weight, divided fraction window, Fourier transform to obtain a spectrogram, then perform MEL filter to make the spectrum map more compact, finally perform a fallback analysis (tubular number and discrete cosine transformation) and differentials (Provide a dynamic feature) Get 42-dimensional MFCC feature vectors;

[0112]Calculate the parameter set of Gaussian mixed model {λ1, ..., λ1, ..., λM}, Λi= (Ωiμi, Σi), i ∈ [1,...

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 belongs to the field of acquisition and processing of audio signals, and discloses a GMM-UBM-based impact sound model establishment method and system and an impact sound detection method and system. The model establishment method comprises the following steps: acquiring an audio signal picked up by a microphone; performing MFCC calculation on the audio signal; and carrying out GMM-UBM data training on the sample data, wherein the impact sound detection method comprises impact sound judgment of the test audio. The method has a good detection effect on impact sound. The method has the advantages that the detection result is robust to environmental noise, the signal-to-noise ratio is robust, and the method is low in operation complexity, easy to implement and the like.

Description

Technical field[0001]The present invention belongs to the field of acquisition and processing of the audio signal, and specifically, it is specifically related to the establishment, impact acoustic detection method and system based on the impact acoustic model of GMM-UBM.Background technique[0002]With the development of information and network technology, safety monitoring has become more and more prominent in national defense and social security, and the sound signal is everywhere in daily life. The amount of information contains is large, and the voice of public occasion can Effective characterization of the safety of the scene. Impact sound referred to some of the surprising sounds in the environment, such as gunshots, rushing brakes, explosions, screams, etc. Since the sound is all-to-direction, it is not affected by the light, so the impact sound is theoretically. But there is a wide variety of impact sounds, and there is a scream, and the gunshots are different from voices. Ev...

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 Applications(China)
IPC IPC(8): G10L25/51G10L25/03G10L25/21G10L25/27
CPCG10L25/51G10L25/03G10L25/27G10L25/21
Inventor 刘芳向阳黄绍锋王向辉
Owner 西安合谱声学科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com