Audio feature extraction method and device thereof, training method and electronic equipment

A technology of audio features and extraction methods, applied in speech analysis, instruments, etc., can solve the problems of complex detection environment and low detection accuracy, and achieve the effect of improving accuracy and good recognition and classification.

Inactive Publication Date: 2020-05-05
普联国际有限公司
View PDF8 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] However, the detection environment is often more complicated. There may be other sounds in the surrounding environment, and the audio may contain environmental noises that are not baby crying. When the energy characteristics of other sounds are similar to the energy characteristics of baby crying, if only through Audio energy detects baby crying, it is likely to detect environmental noise as baby crying, and the detection accuracy is low

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
  • Audio feature extraction method and device thereof, training method and electronic equipment
  • Audio feature extraction method and device thereof, training method and electronic equipment
  • Audio feature extraction method and device thereof, training method and electronic equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0065] The present invention provides an audio feature extraction method, please refer to figure 1 , figure 1 It is a schematic flowchart of a preferred embodiment of an audio feature extraction method provided by the present invention; specifically, the method includes:

[0066] S1. Acquire the audio to be extracted according to the preset window length, and divide the audio to be extracted into M audio frames according to the preset frame length, M> 1;

[0067] S2. Calculate the frequency spectrum corresponding to each audio frame; wherein the frequency spectrum includes N frequency domain points, N> 1;

[0068] S3. Obtain a fitting slope and a fitting intercept corresponding to each frequency spectrum based on a linear fitting algorithm based on the N frequency domain points of each frequency spectrum;

[0069] S4. Calculate and obtain the spectral flatness of each frequency spectrum according to the frequency spectrum and a preset calculation formula;

[0070] S5. Divide each of t...

Embodiment 2

[0114] The invention also provides an audio feature extraction device, please refer to figure 2 , figure 2 It is a schematic structural diagram of a preferred embodiment of an audio feature extraction device provided by the present invention; specifically, the device includes:

[0115] The audio frame obtaining module 11 is configured to obtain the audio to be extracted according to a preset window length, and divide the audio to be extracted into M audio frames according to the preset frame length, M> 1;

[0116] The first calculation module 12 is configured to calculate the frequency spectrum corresponding to each audio frame; wherein, the frequency spectrum includes N frequency domain points, N> 1;

[0117] The fitting module 13 is configured to obtain a fitting slope and a fitting intercept corresponding to each frequency spectrum based on a linear fitting algorithm based on the N frequency domain points of each frequency spectrum;

[0118] The spectrum flatness calculation modul...

Embodiment 3

[0145] The present invention also provides a training method of audio classification model, please refer to image 3 , image 3 It is a schematic flowchart of a preferred embodiment of an audio classification model training method provided by the present invention; specifically, the method includes:

[0146] S9. Construct an initial audio classification model; wherein, P types of classification results corresponding to the audio classification initial model;

[0147] S10. Acquire several training audios corresponding to each of the classification results; wherein each of the training audios is pre-allocated with a classification identifier that matches the corresponding classification result;

[0148] S11. Using the training audio as the audio to be extracted, and extracting an audio feature corresponding to each training audio according to any one of the audio feature extraction methods provided in the first embodiment;

[0149] S12: Perform standardized processing on the audio featur...

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 an audio feature extraction method and a device thereof, a training method and electronic equipment, and the method comprises the steps: obtaining a to-be-extracted audio according to a preset window length, and dividing the to-be-extracted audio into M audio frames according to a preset frame length; calculating a frequency spectrum corresponding to each audio frame; according to the N frequency domain points of each frequency spectrum, obtaining a fitting slope and a fitting intercept corresponding to each frequency spectrum based on a linear fitting algorithm; calculating the frequency spectrum flatness of each frequency spectrum according to the frequency spectrum and a preset calculation formula; dividing each frequency spectrum into m frequency spectrum bands,calculating to obtain a logarithm frequency spectrum corresponding to each frequency spectrum band, and further calculating to obtain a frequency spectrum contrast ratio of each frequency spectrum; and according to the fitting slope, the fitting intercept, the spectrum flatness and the spectrum contrast of the audio frame, obtaining the characteristic quantity of the audio frame, and further extracting the audio characteristics of the audio to-be-extracted. When the extracted audio features are used for infant cry and other detection scenes, the accuracy of audio detection can be improved.

Description

Technical field [0001] The present invention relates to the field of audio processing technology, in particular to an audio feature extraction method, device, training method and electronic equipment. Background technique [0002] With the development of society, problems such as high labor costs have gradually emerged. Nowadays, the consumption cost or time cost of baby care is getting higher and higher. Care products such as baby care devices that can recognize baby crying or household security products are becoming more and more popular. Favor of parents. When the nursing product or home security product detects the baby's crying, it will automatically issue a warning to the nursing staff or parents so as to take care of the baby in time. [0003] In the prior art, when a baby cry is detected, it is usually determined whether there is a baby cry based on the energy characteristics of the detected audio. When the energy characteristics of the detected audio match the energy char...

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): G10L25/03G10L25/18
CPCG10L25/03G10L25/18
Inventor 何维祯
Owner 普联国际有限公司
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