CN-121983074-A - Data enhancement method, device, electronic equipment, storage medium and program product

CN121983074ACN 121983074 ACN121983074 ACN 121983074ACN-121983074-A

Abstract

The invention provides a data enhancement method, a device, electronic equipment, a storage medium and a program product, and relates to the technical field of data processing, wherein the method comprises the steps of obtaining initial data to be enhanced; the method comprises the steps of matching first environmental noise data corresponding to initial data from a plurality of environmental noise data, carrying out energy alignment on the first environmental noise data and the initial data, and carrying out data enhancement on the initial data after energy alignment based on the first environmental noise data after energy alignment. The invention can improve the recognition accuracy of the environmental noise.

Inventors

CHEN XIAOLIANG
YU XIN
CHANG LE
HUANG BINHE
JING TENG

Assignees

北京中科声智科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260310

Claims (11)

1. A method of data enhancement, comprising: acquiring initial data to be reinforced; Matching first environmental noise data corresponding to the initial data from a plurality of environmental noise data; energy aligning the first ambient noise data with the initial data; And carrying out data enhancement on the initial data after the energy alignment based on the first environmental noise data after the energy alignment.
2. The method of claim 1, wherein said matching the first ambient noise data corresponding to the initial data from a plurality of ambient noise data comprises: calculating the similarity between the initial data and each of the plurality of environmental noise data; Determining the first environmental noise data from the plurality of environmental noise data, wherein the similarity between the first environmental noise data and the initial data is larger than the similarity between other noise data and the initial data, and the other noise data is environmental noise data except the first environmental noise data in the plurality of environmental noise data.
3. The method of claim 1, wherein said matching the first ambient noise data corresponding to the initial data from a plurality of ambient noise data comprises: determining a first noise type of the initial data; acquiring a second noise type corresponding to each environmental noise data in the plurality of environmental noise data; the first ambient noise data is determined from the plurality of ambient noise data, a first noise type of the initial data matching a second noise type of the first ambient noise data.
4. A method according to any one of claims 1 to 3, wherein said data enhancing said initial data after energy alignment based on said first ambient noise data after energy alignment comprises: calculating a first power of the first environmental noise data after energy alignment, and calculating a second power of the initial data after energy alignment; Sampling the first environmental noise data after energy alignment based on a preset power range and the first power to obtain a first sampling result, and sampling the initial data after energy alignment based on the preset power range and the second power to obtain a second sampling result; Calculating a signal-to-noise ratio based on the first sampling result and the second sampling result; and carrying out data enhancement on the initial data after energy alignment based on the signal-to-noise ratio and the first environmental noise data after energy alignment.
5. The method of claim 4, wherein the data enhancing the initial data after energy alignment based on the signal-to-noise ratio and the first ambient noise data after energy alignment comprises: Calculating a weight coefficient corresponding to the first environmental noise data after energy alignment based on the signal-to-noise ratio; Adjusting the first environmental noise data after energy alignment based on the weight coefficient to obtain first intermediate environmental noise data, wherein a parameter value of the first intermediate environmental noise data at a first moment is a product of a parameter value of the first environmental noise data at the first moment and the weight coefficient; mixing the first intermediate environmental noise data with the initial data with the aligned energy to obtain the initial data with the enhanced data, wherein the parameter value of the initial data with the enhanced data at the first moment is the sum of the parameter value of the initial data at the first moment and the parameter value of the first intermediate environmental noise data.
6. The method of claim 5, wherein said mixing said first intermediate ambient noise data with said energy aligned initial data to obtain data enhanced initial data comprises: Preprocessing the first intermediate environmental noise data to obtain second intermediate environmental noise data, wherein the preprocessing comprises filling processing or cutting processing, and the second intermediate environmental noise data is matched with the time length of the initial data after energy alignment; and mixing the second intermediate environmental noise data with the initial data with the aligned energy to obtain the initial data with the enhanced data.
7. The method of any one of claims 1 to 3, wherein the initial data comprises snore data and the plurality of ambient noise data comprises at least one of air conditioning ambient noise data, street ambient noise data, and indoor mute ambient noise data.
8. A data enhancement device, comprising: the acquisition module is used for acquiring initial data to be reinforced; the matching module is used for matching the first environmental noise data corresponding to the initial data from the plurality of environmental noise data; A processing module for energy aligning the first ambient noise data with the initial data; and the enhancement module is used for carrying out data enhancement on the initial data after the energy alignment based on the first environmental noise data after the energy alignment.
9. An electronic device comprising a processor, a memory and a program stored on the memory and executable on the processor, the program when executed by the processor implementing the steps of the data enhancement method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of the data enhancement method according to any of claims 1 to 7.
11. A computer program product comprising computer instructions which, when executed by a processor, implement the steps of the data enhancement method according to any of claims 1 to 7.

Description

Data enhancement method, device, electronic equipment, storage medium and program product Technical Field The present invention relates to the field of data processing technologies, and in particular, to a data enhancement method, a data enhancement device, an electronic device, a storage medium, and a program product. Background Data enhancement of audio data is a key technique for identifying noise in audio data. In the prior art, data enhancement is typically performed by directly superimposing random white gaussian noise on the original audio data. However, the random Gaussian white noise cannot simulate real environmental noise, so that the model obtained through training of the audio data after data enhancement cannot accurately identify the environmental noise, and the identification accuracy is low. Therefore, the problem of low recognition accuracy of environmental noise exists in the prior art. Disclosure of Invention The embodiment of the invention provides a data enhancement method, a device, electronic equipment, a storage medium and a program product, which are used for solving the problem of low recognition accuracy of environmental noise in the prior art. To solve the above problems, the present invention is achieved as follows: In a first aspect, an embodiment of the present invention provides a data enhancement method, including: acquiring initial data to be reinforced; Matching first environmental noise data corresponding to the initial data from a plurality of environmental noise data; energy aligning the first ambient noise data with the initial data; And carrying out data enhancement on the initial data after the energy alignment based on the first environmental noise data after the energy alignment. In one embodiment, the matching the first environmental noise data corresponding to the initial data from the plurality of environmental noise data includes: calculating the similarity between the initial data and each of the plurality of environmental noise data; Determining the first environmental noise data from the plurality of environmental noise data, wherein the similarity between the first environmental noise data and the initial data is larger than the similarity between other noise data and the initial data, and the other noise data is environmental noise data except the first environmental noise data in the plurality of environmental noise data. In one embodiment, the matching the first environmental noise data corresponding to the initial data from the plurality of environmental noise data includes: determining a first noise type of the initial data; acquiring a second noise type corresponding to each environmental noise data in the plurality of environmental noise data; the first ambient noise data is determined from the plurality of ambient noise data, a first noise type of the initial data matching a second noise type of the first ambient noise data. In one embodiment, the data enhancing the initial data after energy alignment based on the first ambient noise data after energy alignment includes: calculating a first power of the first environmental noise data after energy alignment, and calculating a second power of the initial data after energy alignment; Sampling the first environmental noise data after energy alignment based on a preset power range and the first power to obtain a first sampling result, and sampling the initial data after energy alignment based on the preset power range and the second power to obtain a second sampling result; Calculating a signal-to-noise ratio based on the first sampling result and the second sampling result; and carrying out data enhancement on the initial data after energy alignment based on the signal-to-noise ratio and the first environmental noise data after energy alignment. In one embodiment, the data enhancing the initial data after energy alignment based on the signal-to-noise ratio and the first ambient noise data after energy alignment includes: Calculating a weight coefficient corresponding to the first environmental noise data after energy alignment based on the signal-to-noise ratio; Adjusting the first environmental noise data after energy alignment based on the weight coefficient to obtain first intermediate environmental noise data, wherein a parameter value of the first intermediate environmental noise data at a first moment is a product of a parameter value of the first environmental noise data at the first moment and the weight coefficient; mixing the first intermediate environmental noise data with the initial data with the aligned energy to obtain the initial data with the enhanced data, wherein the parameter value of the initial data with the enhanced data at the first moment is the sum of the parameter value of the initial data at the first moment and the parameter value of the first intermediate environmental noise data. In one embodiment, the mixing the first intermediate ambient noise data and the initial data a