Broadcast high-frequency noise detection method

By performing time-domain framing and feature extraction on broadcast audio data, the channel correlation is calculated, and the problems of low manual detection efficiency and insufficient accuracy in the prior art are solved, efficient broadcast high-frequency noise detection is achieved, saving labor costs and Improve detection accuracy.

CN119068915APending Publication Date: 2024-12-03NANTONG KERUI ENVIRONMENTAL TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411044255.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The existing broadcast audio quality evaluation noise detection methods rely on manual sampling, which are inefficient and inaccurate.

Method used

A broadcast high-frequency noise detection method is adopted. By performing time-domain frame-division and feature extraction of audio data, the frame feature vectors and correlations of each channel are calculated, whether it is high-frequency audio, and an audio detection report is generated to reduce manual inspection work. , improve detection efficiency.

Benefits of technology

It realizes comprehensive detection of broadcast high-frequency noise, saves labor costs, improves efficiency, avoids detection abnormalities, and ensures the accuracy of results.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a broadcast high-frequency noise detection method, which can comprehensively detect broadcast high-frequency audio noise, greatly reduce manual inspection work, save labor cost and improve efficiency. Obtaining an audio anomaly detection result in each channel device by calculating the correlation of the audio data collected by each channel; and detection abnormity caused by the same problem of all channels and various noises is avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of audio detection, and in particular relates to a method for detecting broadcast high-frequency noise. Background Art

[0002] Broadcasting refers to a news dissemination tool that transmits sound through radio waves or wires. It was born in the 1920s. Among them, the program transmitted by radio waves is called wireless broadcasting, and the program transmitted by wires is called cable broadcasting. The advantages of broadcasting are wide coverage, rapid dissemination, diverse functions, and strong appeal; the disadvantages are that it is fleeting, sequential listening, inability to select, and difficulty in listening if the language is not understood. Now, broadcasting has achieved the strategic goal of mainly using FM and medium wave and supplemented by short wave. This is a major change in the history of international broadcasting. Before broadcasting, the broadcast audio needs to be evaluated for audio quality and noise detection, but the existing detection method is generally through manual sampling detection, which is not only inefficient but also cannot guarantee the accuracy of the detection results. Summary of the invention

[0003] Purpose of the invention: In order to solve the deficiencies of the prior art, the present invention provides a method for detecting broadcast high-frequency noise.

[0004] Technical solution: A method for detecting broadcast high-frequency noise, comprising the following steps:

[0005] Step 1: first, the collected audio data is used to form an audio group to be tested, and then high-frequency audio data is collected, and the audio data in the audio group to be tested and the high-frequency audio data are subjected to time domain framing and feature extraction processing to obtain multiple frame feature vectors of each channel to be tested and multiple frame feature vectors of the reference channel;

[0006] Step 2: Calculate the correlation and make frame correlation decisions using the frame feature vectors of each channel to obtain multiple decision results for each channel; Calculate the correlation and make frame correlation decisions using the frame feature vectors of the reference channel and the frame feature vectors of the channel to be detected to obtain multiple decision results for the target channel;

[0007] Step 3: According to the multiple judgment results of each channel, determine whether the audio data of each channel is high-frequency audio, and obtain the final high-frequency noise detection result;

[0008] Step 4: Generate an audio detection report based on the internal high-frequency noise detection results of each channel to be detected.

[0009] As an optimization: in the step 1, when the audio data in the audio group to be tested and the audio data of the reference channel are subjected to time domain framing and feature extraction processing, a differential sequence of the audio signal to be detected is obtained through multiple high-frequency audio frames, and the differential sequence is used to indicate the degree of difference between the multiple audio frames; an autocorrelation operation is performed on the differential sequence, and multiple correlation values ​​are obtained from the autocorrelation operation results; and the multiple correlation values ​​are combined into a correlation value sequence.

[0010] As an optimization: the correlation value of the audio to be detected includes: dividing the correlation value sequence into multiple sequence vectors according to a preset vector dimension, obtaining the correlation coefficient of two adjacent sequence vectors in the multiple sequence vectors, and determining the average value of the correlation coefficients of two adjacent sequence vectors in the multiple sequence vectors as the average correlation coefficient of the correlation value sequence.

[0011] As an optimization: in the step 2, the frame feature vectors of every two channels to be detected are used to calculate the frame correlation coefficients of every two channels to be detected. The frame correlation coefficients of every two channels to be detected constitute a frame correlation coefficient matrix to be detected. It is determined whether the frame correlation coefficients in the frame correlation coefficient matrix to be detected are greater than a first preset threshold value. If so, 1 is used as the judgment result of the channel corresponding to the frame correlation coefficient; if not, 0 is used as the judgment result of the channel corresponding to the frame correlation coefficient.

[0012] As an optimization: in the step three, the sum of the judgment results of each channel of the device to be detected and other channels is counted to determine whether the sum of the judgment results of each channel of the device to be detected and other channels is greater than a preset value. If so, 1 is used as the detection result of the audio of the channel to be detected being judged as high-frequency audio. If not, 0 is used as the detection result of the audio of the channel to be detected being judged as not being a high-frequency audio.

[0013] Beneficial effects: The present invention can perform comprehensive detection of high-frequency audio noise in broadcasts, which can greatly reduce manual inspection work, save labor costs and improve efficiency; obtain audio anomaly detection results in each channel device by calculating the correlation of audio data collected by each channel; avoid the same problem in all channels and detection anomalies caused by various noises. DETAILED DESCRIPTION

[0014] The technical solutions in the embodiments of the present invention will be described clearly and completely below so that those skilled in the art can better understand the advantages and features of the present invention, thereby making a clearer definition of the protection scope of the present invention. The embodiments described in the present invention are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention.

[0015] Example

[0016] A method for detecting broadcast high-frequency noise comprises the following steps:

[0017] Step 1: first, the collected audio data is used to form an audio group to be tested, and then high-frequency audio data is collected, and the audio data in the audio group to be tested and the high-frequency audio data are subjected to time domain framing and feature extraction processing to obtain multiple frame feature vectors of each channel to be tested and multiple frame feature vectors of the reference channel;

[0018] In the step 1, when the audio data in the audio group to be tested and the audio data of the reference channel are subjected to time domain framing and feature extraction processing, a differential sequence of the audio signal to be detected is obtained through multiple high-frequency audio frames, and the differential sequence is used to indicate the degree of difference between the multiple audio frames; an autocorrelation operation is performed on the differential sequence, and multiple correlation values ​​are obtained from the autocorrelation operation results; and the multiple correlation values ​​are combined into a correlation value sequence.

[0019] The correlation value of the audio to be detected includes: dividing the correlation value sequence into multiple sequence vectors according to a preset vector dimension, obtaining the correlation coefficient of two adjacent sequence vectors in the multiple sequence vectors, and determining the average value of the correlation coefficients of two adjacent sequence vectors in the multiple sequence vectors as the average correlation coefficient of the correlation value sequence.

[0020] Step 2: Calculate the correlation and make frame correlation decisions using the frame feature vectors of each channel to obtain multiple decision results for each channel; Calculate the correlation and make frame correlation decisions using the frame feature vectors of the reference channel and the frame feature vectors of the channel to be detected to obtain multiple decision results for the target channel;

[0021] In the step 2, the frame feature vectors of every two channels to be detected are used to calculate the frame correlation coefficients of every two channels to be detected. The frame correlation coefficients of every two channels to be detected constitute a frame correlation coefficient matrix to be detected. It is determined whether the frame correlation coefficient in the frame correlation coefficient matrix to be detected is greater than a first preset threshold value. If so, 1 is used as the judgment result of the channel corresponding to the frame correlation coefficient; if not, 0 is used as the judgment result of the channel corresponding to the frame correlation coefficient.

[0022] Step 3: According to the multiple judgment results of each channel, determine whether the audio data of each channel is high-frequency audio, and obtain the final high-frequency noise detection result;

[0023] In the step three, the sum of the judgment results of each channel of the device to be detected and other channels is counted to determine whether the sum of the judgment results of each channel of the device to be detected and other channels is greater than a preset value. If so, 1 is used as the detection result that the audio of the channel to be detected is judged as a high-frequency audio. If not, 0 is used as the detection result that the audio of the channel to be detected is judged as not a high-frequency audio.

[0024] Step 4: Generate an audio detection report based on the internal high-frequency noise detection results of each channel to be detected.

[0025] The present invention can perform comprehensive detection of high-frequency audio noise of broadcasting, which can greatly reduce manual inspection work, save labor costs and improve efficiency; obtain audio anomaly detection results in each channel device by calculating the correlation of audio data collected by each channel; avoid the same problem in all channels and detection anomalies caused by various noises.

Claims

1. A method for detecting broadcast high-frequency noise, characterized in that: The steps include: Step 1: first, the collected audio data is used to form an audio group to be tested, and then high-frequency audio data is collected, and the audio data in the audio group to be tested and the high-frequency audio data are subjected to time domain framing and feature extraction processing to obtain multiple frame feature vectors of each channel to be tested and multiple frame feature vectors of the reference channel; Step 2: Calculate the correlation and make frame correlation decisions using the frame feature vectors of each channel to obtain multiple decision results for each channel; Calculate the correlation and make frame correlation decisions using the frame feature vectors of the reference channel and the frame feature vectors of the channel to be detected to obtain multiple decision results for the target channel; Step 3: According to the multiple judgment results of each channel, determine whether the audio data of each channel is high-frequency audio, and obtain the final high-frequency noise detection result; Step 4: Generate an audio detection report based on the internal high-frequency noise detection results of each channel to be detected.

2. The broadcast high-frequency noise detection method according to claim 1, characterized in that: In the step 1, when the audio data in the audio group to be tested and the audio data of the reference channel are subjected to time domain framing and feature extraction processing, a differential sequence of the audio signal to be detected is obtained through multiple high-frequency audio frames, and the differential sequence is used to indicate the degree of difference between the multiple audio frames; an autocorrelation operation is performed on the differential sequence, and multiple correlation values ​​are obtained from the autocorrelation operation results; and the multiple correlation values ​​are combined into a correlation value sequence.

3. The broadcast high-frequency noise detection method according to claim 1, characterized in that: The correlation value of the audio to be detected includes: dividing the correlation value sequence into multiple sequence vectors according to a preset vector dimension, obtaining the correlation coefficient of two adjacent sequence vectors in the multiple sequence vectors, and determining the average value of the correlation coefficients of two adjacent sequence vectors in the multiple sequence vectors as the average correlation coefficient of the correlation value sequence.

4. The broadcast high-frequency noise detection method according to claim 1, characterized in that: In the step 2, the frame feature vectors of every two channels to be detected are used to calculate the frame correlation coefficients of every two channels to be detected. The frame correlation coefficients of every two channels to be detected constitute a frame correlation coefficient matrix to be detected. It is determined whether the frame correlation coefficient in the frame correlation coefficient matrix to be detected is greater than a first preset threshold value. If so, 1 is used as the judgment result of the channel corresponding to the frame correlation coefficient; if not, 0 is used as the judgment result of the channel corresponding to the frame correlation coefficient.

5. The broadcast high-frequency noise detection method according to claim 1, characterized in that: In the step three, the sum of the judgment results of each channel of the device to be detected and other channels is counted to determine whether the sum of the judgment results of each channel of the device to be detected and other channels is greater than a preset value. If so, 1 is used as the detection result that the audio of the channel to be detected is judged as a high-frequency audio. If not, 0 is used as the detection result that the audio of the channel to be detected is judged as not a high-frequency audio.