CN117727314B - Filtering enhancement method for ecological audio information - Google Patents
Filtering enhancement method for ecological audio information Download PDFInfo
- Publication number
- CN117727314B CN117727314B CN202410179556.5A CN202410179556A CN117727314B CN 117727314 B CN117727314 B CN 117727314B CN 202410179556 A CN202410179556 A CN 202410179556A CN 117727314 B CN117727314 B CN 117727314B
- Authority
- CN
- China
- Prior art keywords
- time sequence
- audio
- frequency domain
- sequence interval
- bird
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000001914 filtration Methods 0.000 title claims abstract description 54
- 230000005236 sound signal Effects 0.000 claims abstract description 92
- 238000009499 grossing Methods 0.000 claims abstract description 38
- 230000000694 effects Effects 0.000 claims abstract description 25
- 238000001228 spectrum Methods 0.000 claims abstract description 18
- 238000010586 diagram Methods 0.000 claims description 75
- 230000010355 oscillation Effects 0.000 claims description 36
- 230000002708 enhancing effect Effects 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 6
- 238000009825 accumulation Methods 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 2
- 241000271566 Aves Species 0.000 description 11
- 230000007613 environmental effect Effects 0.000 description 7
- 238000012544 monitoring process Methods 0.000 description 5
- 230000003542 behavioural effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Circuit For Audible Band Transducer (AREA)
Abstract
The application relates to the technical field of voice processing, and provides a filtering enhancement method for ecological audio information, which comprises the following steps: acquiring bird audio signal data; acquiring a frequency spectrum energy difference index according to bird audio signal data; acquiring an audio noise receiving factor according to the spectrum energy difference index; acquiring a noise disturbance index according to the audio noise factor; acquiring a wind sound interference degree index according to bird audio signal data; acquiring a wind sound interference adjustment coefficient according to the wind sound interference degree index; obtaining a wiener filter smooth adjustment coefficient according to the noise disturbance index and the wind noise disturbance adjustment coefficient; obtaining a wiener filter smoothing coefficient according to the wiener filter smoothing adjustment coefficient; and acquiring bird audio signal data after the filtering enhancement based on the wiener filtering smoothing coefficient by using a wiener filtering algorithm. The application improves the filtering enhancement effect on bird audio signal data through the self-adaptive wiener filtering smoothing coefficient.
Description
Technical Field
The application relates to the technical field of voice processing, in particular to a filtering enhancement method for ecological audio information.
Background
With the development of society, people are increasingly conscious of ecological environment protection, people are becoming aware that many wild birds and their habitat are not protected, and more species may be extinct, so that people continuously strengthen the monitoring protection of birds. At present, an ecological system AI voiceprint monitoring technology is mostly adopted, bird audio information is monitored through voiceprint sensing equipment arranged in a field environment, and monitoring protection of birds is achieved through the bird audio information.
Because the positions of the voiceprint sensing equipment are different, the moving range of birds is larger, and the bird audio information is weaker in the process of collecting the audio data, so that the bird audio information is difficult to distinguish, and the filter enhancement methods such as Kalman filtering, wiener filtering and the like are generally required to be adopted, so that the identification degree of bird singing in the audio is improved.
In addition, the audio data capable of representing birds actually has related audio information of environmental noise besides the bird song, and the related audio information of the environmental noise generally affects the recognition degree of the bird song in the audio, so that the audio data needs to be removed as noise in the process of denoising. However, the traditional filtering enhancement algorithm has poor filtering enhancement effect on bird audio information, and misjudgment is generated on bird song detection. For example, in the traditional wiener filtering algorithm, the influence of environmental noise on the audio is not analyzed, the accuracy of the smoothing coefficient is low, and the filtering enhancement effect of bird audio information is poor.
Disclosure of Invention
The application provides a filtering enhancement method for ecological audio information, which aims to solve the problem of poor filtering enhancement effect of bird audio information, and adopts the following technical scheme:
an embodiment of the present application provides a filtering enhancement method for eco-audio information, the method comprising the steps of:
acquiring bird audio signal data;
Acquiring a frequency domain waveform diagram of each time sequence interval according to bird audio signal data, and acquiring all frequency component peak points and oscillation trailing wave peak points in the frequency domain waveform diagram of each time sequence interval according to the frequency domain waveform diagram of each time sequence interval; acquiring an oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval according to the frequency component peak point and the oscillation tailing peak point; acquiring noise disturbance indexes of bird audio signal data according to the oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval;
Acquiring all bird song existence time sequence intervals and ecological environment audio time sequence intervals according to the bird audio signal data; acquiring a wind-sound interference adjustment coefficient of bird audio signal data according to a bird song existence time sequence interval and an ecological environment audio time sequence interval; obtaining a wiener filter smooth adjustment coefficient according to the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data;
And acquiring wiener filter smoothing coefficients according to the wiener filter smoothing adjustment coefficients, and acquiring bird audio signal data after filter enhancement based on the wiener filter smoothing coefficients by using a wiener filter algorithm.
Preferably, the method for acquiring the frequency domain waveform diagram of each time sequence interval according to the bird audio signal data and acquiring all the frequency component peak points and the oscillation trailing peak points in the frequency domain waveform diagram of each time sequence interval according to the frequency domain waveform diagram of each time sequence interval comprises the following steps:
Taking the interval of each first preset parameter in the acquisition time of the bird audio signal data as a time sequence interval;
taking bird audio signal data in each time sequence interval as input of Fourier transform, and obtaining frequency domain data of each time sequence interval by utilizing the Fourier transform; drawing the frequency domain data of each time sequence interval by utilizing MATLAB software to obtain a frequency domain waveform diagram of each time sequence interval, wherein the horizontal axis of the frequency domain waveform diagram is frequency, and the vertical axis of the frequency domain waveform diagram is frequency domain energy;
For the frequency domain waveform diagram of each time sequence interval, taking the frequency domain energy of all wave crest points in the frequency domain waveform diagram as input of an Ojin threshold method, and obtaining a division threshold value by using the Ojin threshold method; and taking each wave peak point with the frequency domain energy larger than or equal to the division threshold value in the frequency domain waveform as a frequency component wave peak point, and taking each wave peak point with the frequency domain energy smaller than the division threshold value in the frequency domain waveform as an oscillation trailing wave peak point.
Preferably, the method for obtaining the oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval according to the frequency component peak point and the oscillation tailing peak point comprises the following steps:
Taking each frequency component peak point in the frequency domain waveform diagram of each time sequence interval as a target frequency component peak point, and taking a sequence formed by all oscillation tailing peak points between the target frequency component peak point and the next frequency component peak point according to the ascending order of the frequencies as an oscillation tailing effect characteristic sequence of the target frequency component peak point;
And regarding the frequency domain waveform diagram of each time sequence interval, taking the frequency domain energy which is the frequency domain energy multiplied by the second preset parameter of the minimum frequency domain energy of all the frequency component peak points in the frequency domain waveform diagram as bandwidth construction energy, and taking the width of the waveform corresponding to each frequency component peak point, where the transverse straight line where the bandwidth construction energy is located, passing through each frequency component peak point as the bandwidth of each frequency component peak point in the frequency domain waveform diagram.
Preferably, the method for obtaining the noise disturbance index of the bird audio signal data according to the oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval comprises the following steps:
Taking the difference value of the frequency domain energy of each frequency component peak point in the frequency domain waveform diagram and the frequency domain energy average value of all oscillation trailing wave peak points in the oscillation trailing effect characteristic sequence of each frequency component peak point as a molecule for the frequency domain waveform diagram of each time sequence interval; calculating the absolute value of the difference between the frequency domain energy of each frequency component peak point and the frequency domain energy mean value of all frequency component peak points in the frequency domain waveform chart, and taking the sum of the absolute value and a third preset parameter as a denominator; taking the average value of the sum of the absolute values of the ratios of the numerator and the denominator on the frequency domain waveform diagram as the spectrum energy difference index of the time sequence interval;
Acquiring an audio noise receiving factor of each time sequence interval according to the frequency spectrum energy difference index of each time sequence interval;
And taking the average value of the sum of the spectrum energy difference index and the audio noise factor and the sum of the spectrum energy difference index and the audio noise factor over all time sequence intervals as the noise disturbance index of the bird audio signal data.
Preferably, the method for obtaining the audio noise factor of each time sequence interval according to the spectrum energy difference index of each time sequence interval comprises the following steps:
Taking a spectrum energy difference index of each time sequence interval as a molecule for the frequency domain waveform diagram of each time sequence interval;
Calculating the absolute value of the difference between the bandwidth of each frequency component peak point in the frequency domain waveform diagram and the bandwidth average value of all frequency component peak points, calculating the accumulation sum of the absolute value on the frequency domain waveform diagram, and taking the product of the accumulation sum and the frequency domain energy of the fundamental frequency in the frequency domain waveform diagram as a denominator;
the ratio of the numerator to the denominator is used as the audio noise factor of the time sequence interval.
Preferably, the method for acquiring all the bird song existence time sequence intervals and the ecological environment audio time sequence intervals according to the bird audio signal data comprises the following steps:
Taking the amplitude values of all audio signals in the bird audio signal data as the input of an Ojin threshold method, and obtaining a bird song audio dividing threshold value by using the Ojin threshold method;
Calculating the average value of the amplitude values of all audio signals in the bird audio signal data in each time sequence interval, taking each time sequence interval with the average value of the amplitude values being more than or equal to a bird song audio frequency dividing threshold value as a bird song existence time sequence interval, and taking each time sequence interval with the average value of the amplitude values being less than the bird song audio frequency dividing threshold value as an ecological environment audio frequency time sequence interval.
Preferably, the method for obtaining the wind noise interference adjustment coefficient of the bird audio signal data according to the bird song existence time sequence interval and the ecological environment audio time sequence interval comprises the following steps:
Taking each ecological environment audio time sequence interval as a target ecological environment audio time sequence interval, and taking the total number of frequency component peak points in a frequency domain waveform diagram of the target ecological environment audio time sequence interval as bird behavior audio weight factors of the target ecological environment audio time sequence interval when the audio noise receiving factor corresponding to the target ecological environment audio time sequence interval is smaller than or equal to the audio noise receiving factor of the last time sequence interval; when the audio noise factor corresponding to the target ecological environment audio time sequence interval is larger than the audio noise factor of the previous time sequence interval, calculating the difference value of the audio noise factor corresponding to the target ecological environment audio time sequence interval and the audio noise factor of the previous time sequence interval, and taking the sum of the total number of the frequency component peak points and the difference value as a wind-up audio weight factor of the target ecological environment audio time sequence interval;
Acquiring a wind sound interference degree index of the target ecological environment audio time sequence interval according to a wind sound weighting factor of the target ecological environment audio time sequence interval;
And taking the average value of the wind noise interference degree indexes corresponding to all the ecological environment audio time sequence intervals as a wind noise interference adjustment coefficient of the bird audio signal data.
Preferably, the method for obtaining the wind noise interference degree index of the target ecological environment audio time sequence interval according to the wind noise weight factor of the target ecological environment audio time sequence interval comprises the following steps:
Calculating the amplitude average value of all audio signals in the bird audio signal data in each ecological environment audio time sequence interval, calculating the absolute value of the difference between the amplitude average value corresponding to the target ecological environment audio time sequence interval and the amplitude average value corresponding to each ecological environment audio time sequence interval, and taking the natural constant as a base number and taking the mapping result of the sum of the absolute value on the total number of the ecological environment audio time sequence intervals as an index as a first summation factor;
Taking the air-blast audio weight factors of the target ecological environment audio time sequence interval as molecules; calculating a measurement distance between a maximum amplitude point of the target ecological environment audio time sequence interval and a maximum amplitude point of each bird song existence time sequence interval, and taking the sum of the measurement distances on the total number of the bird song existence time sequence intervals as a denominator;
and taking the sum of the ratio of the numerator and the denominator and the first summation factor as an index of the wind noise interference degree of the target ecological environment audio time sequence interval.
Preferably, the method for obtaining the wiener filter smooth adjustment coefficient according to the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data comprises the following steps:
And calculating the product of the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data, and taking the product as a wiener filtering smooth adjustment coefficient.
Preferably, the method for obtaining wiener filter smoothing coefficients according to wiener filter smoothing adjustment coefficients and obtaining bird audio signal data after filter enhancement based on the wiener filter smoothing coefficients by using a wiener filter algorithm comprises the following steps:
Calculating the sum of the wiener filter smoothing adjustment coefficient and a fourth preset parameter, and taking the product of the sum and the preset smoothing coefficient as the wiener filter smoothing coefficient;
And taking the bird audio signal data and the wiener filter smoothing coefficient as inputs of a wiener filter algorithm respectively, and obtaining the bird audio signal data after the filter enhancement by using the wiener filter algorithm.
The beneficial effects of the application are as follows: through the frequency domain characteristics of noise interference of field audio data and analysis of behavior activity sound of wind noise interference, noise interference indexes and wind noise interference adjustment coefficients of the audio data are constructed, further wiener filter smooth adjustment coefficients are constructed, and smooth coefficients in a wiener filter algorithm are improved, so that the wiener filter algorithm can adaptively adjust the smooth coefficients according to noise conditions of the field audio and the existence condition of the wind noise interference audio, more noise is removed on the basis of resident bird types of behavior audio, more accurate field bird audio is obtained, and more accurate data support is provided for follow-up field bird automatic monitoring. Therefore, the application improves the filtering enhancement effect on bird audio signal data through the self-adaptive wiener filtering smoothing coefficient.
Drawings
In order to more clearly illustrate the embodiments of the application or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the application, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a flowchart of a filtering enhancement method for ecological audio information according to an embodiment of the present application;
Fig. 2 is a flowchart of an implementation of a filtering enhancement method for eco-audio information according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Referring to fig. 1, a flowchart of a filtering enhancement method for ecological audio information according to an embodiment of the present application is shown, the method includes the following steps:
and S001, acquiring bird audio information data in a field ecological environment.
The method comprises the steps of collecting bird audio signal data by arranging voiceprint sensing equipment in a field ecological environment, firstly collecting audio signal data for 5min, determining the maximum amplitude value of the audio signal data through a time sequence formed by the amplitude values of the audio signal data according to the time ascending order, formally starting to collect the bird audio signal data, setting the sampling rate to be twice the maximum amplitude value, collecting the total duration to be 720min, and recording the data as bird audio signal data.
Thus, bird audio signal data in the field ecological environment are obtained.
Step S002, obtaining a frequency spectrum energy difference index according to the bird audio signal data, obtaining an audio noise receiving factor according to the frequency spectrum energy difference index, and obtaining a noise interference index according to the audio noise receiving factor.
Because the positions of the voiceprint sensing equipment are relatively fixed, the moving range, time and the like of field birds are greatly changed, and the sound intensity and characteristics of the birds are weak in audio data collected in many times, the birds are difficult to accurately monitor relative to the bird audio data in the audio data. Thus, filtering enhancement is typically performed on the collected bird audio signal data.
In addition, because more than one bird may exist in the monitored area, the audio data segment with higher aliasing effect may be the result of multiple birds ringing at the same time, so that the actual noise influence needs to be analyzed to determine the influence of the noise on the bird audio signal data, thereby improving the filtering effect and enhancing the bird voice data.
Specifically, each 1min interval in the acquisition time of the acquired bird audio signal data is divided into a time sequence interval, bird audio signal data in each time sequence interval is used as input of Fourier transform, output of the Fourier transform is used as frequency domain data of each time sequence interval, MATLAB software is utilized to draw a frequency domain waveform diagram of each time sequence interval based on the frequency domain data of each time sequence interval, and the horizontal axis of the frequency domain waveform diagram is frequency and the vertical axis is frequency domain energy. The fourier transform is a well-known technique, and the process of the present application is not described in detail.
Further, in the frequency domain waveform diagram, each frequency corresponds to one frequency domain energy, the greater the degree of noise interference of the audio signal, the more additional frequency components are, and the greater the bandwidth of the fundamental frequency, the greater the difference of the fundamental frequency energy in each time sequence interval.
Specifically, for the frequency domain waveform diagram of the x time sequence interval, all wave peak points are obtained, wherein the wave peak points comprise all frequency component wave peak points and oscillation trailing wave peak points generated by frequency leakage, the frequency domain energy of all wave peak points is used as input of an Ojin threshold method, the output of the Ojin threshold method is used as a dividing threshold of the wave peak points, each wave peak point with the frequency domain energy being greater than or equal to the dividing threshold is used as a frequency component wave peak point, and each wave peak point with the frequency domain energy being smaller than the dividing threshold is used as an oscillation trailing wave peak point. The method of threshold value of Ojin is a well-known technique, and the process is not repeated in the present application.
Further, since the oscillation tailing is usually followed by each frequency component, the sequence of all the oscillation tailing peaks from each frequency component peak to the next frequency component peak in ascending order of frequency is used as the oscillation tailing characteristic sequence of each frequency component peak, and the oscillation tailing characteristic sequence of the c-th frequency component peak is recorded as. At the same time, the minimum frequency domain energy of all frequency component peak pointsThe frequency domain energy is taken as bandwidth construction energy, a transverse straight line where the bandwidth construction energy is located in the frequency domain waveform diagram passes through waveform diagrams of all frequency component peak points, and the width of the waveform passing through each frequency component peak point is recorded as the bandwidth of the frequency component peak point.
Calculating a noise disturbance index of the bird audio signal data:
In the method, in the process of the invention, For the spectrum energy difference index of the xth time sequence interval, V is the total number of the wave peak points of the frequency component in the frequency domain waveform diagram of the xth time sequence interval,/>Frequency domain energy of peak point of c-th frequency component in frequency domain waveform diagram of x-th time sequence interval,/>Is the frequency domain energy average value of all oscillation trailing wave peak points in the oscillation trailing effect characteristic sequence of the c-th frequency component wave peak point in the frequency domain waveform diagram of the x-th time sequence interval,/>Is the frequency domain energy average value of all frequency component peak points in the frequency domain waveform diagram of the xth time sequence interval,/>The error parameter is avoided to be 0, and the experience value of the error parameter is 0.1;
audio noise factor for the xth time interval,/> Frequency domain energy of fundamental frequency in frequency domain waveform diagram of xth time sequence interval,/>Bandwidth of peak point of c-th frequency component in frequency domain waveform diagram of x-th time sequence interval,/>The average value of the bandwidths of all the frequency component peak points in the frequency domain waveform diagram of the xth time sequence interval;
N is the noise disturbance index of the bird audio signal data, and N is the total number of time sequence intervals in the bird audio signal data.
When the difference of the frequency domain energy corresponding to each frequency component peak point in the frequency domain waveform diagram of the xth time sequence interval is smaller, the difference of the frequency domain energy corresponding to each frequency component peak point and the frequency domain energy average value generated by the oscillation tailing effect corresponding to each frequency component peak point is larger, namelyThe larger the value of the frequency component difference in the spectrogram of the audio data in the time sequence interval is, the larger the difference between the frequency component difference and the energy generated by the oscillation tail is, the more likely the audio data is interfered by noise, and the larger the spectrum energy difference index is; when the frequency domain energy corresponding to the fundamental frequency in the frequency domain waveform diagram of the xth time sequence interval is smaller, and the bandwidth difference corresponding to each frequency component peak point is smaller, namely/>The larger the value of the frequency component is, the smaller the ratio of the energy of the fundamental frequency in the time sequence interval to the total energy in the audio data is, and the smaller the bandwidth difference of the frequency components is, the larger the corresponding audio data receives noise interference degree is, and the larger the audio noise receiving factor is. The greater the degree of audio noise in the xth time sequence interval, and the more complex the frequency domain energy variation, i.e./>The larger the value of (c) is, the more aliased noise in the audio and the greater the interference level, i.e. the greater the N, the greater the smoothing coefficient should be.
Thus, the noise disturbance index of the bird audio signal data is obtained.
Step S003, obtaining a wind noise interference adjustment coefficient according to bird audio signal data, obtaining a wiener filter smooth adjustment coefficient according to a noise disturbance index and the wind noise interference adjustment coefficient, and obtaining the wiener filter smooth coefficient according to the wiener filter smooth adjustment coefficient.
However, in the outdoor sound collection process, the wind sound in the environment is the wind sound with the greatest interference to the bird sound, and compared with other environmental noises, the wind sound has the characteristics of variable duration, irregular appearance time and the like, and when the wind sound appears in the environment and the wind sound is larger, the influence on the definition of the bird sound is larger, so that the recognition and monitoring of the birds are further influenced.
In order to ensure that the bird voice data can be finally and clearly identified from the field voice data, besides the adjustment of the filtering effect by judging the noise interference degree received by the audio data, the wind noise condition should be analyzed. Specifically, when bird song is found in the field audio data collected by the voiceprint sensing device, the time sequence interval corresponding to the bird song is possibly influenced by wind sound, and then the audio of the bird song is more unclear.
In the field audio data before bird song, when birds do not generate sound of behavioral activities, only environmental noise should exist in the audio data, which is usually relatively stable, and when wind starts to generate in the field environment, the original environmental noise of the field audio data generates a certain change relative to the previous one, and the aliasing condition of the specific audio is enhanced, the sound source is increased, and a new larger amplitude condition may be generated.
Specifically, since the amplitude of the audio generated by the bird song is much higher than the amplitude generated by the environmental noise under normal conditions, the amplitude of all the audio signals in the bird audio signal data is used as the input of the dyadic threshold method, the output of the dyadic threshold method is used as the bird song audio dividing threshold value of the bird audio signal data, the amplitude average value of all the audio signals in the bird audio signal data in each time sequence interval is calculated, each time sequence interval, of which the amplitude average value is greater than or equal to the bird song audio dividing threshold value, is used as one bird song existence time sequence interval, and each time sequence interval, of which the amplitude average value is smaller than the bird song audio dividing threshold value, is used as one ecological environment audio time sequence interval. The method of threshold value of Ojin is a well-known technique, and the process is not repeated in the present application.
Calculating the wind-sound interference adjustment coefficient of the bird audio signal data:
In the method, in the process of the invention, Wind-up audio weight factor of the t-th ecological environment audio time sequence interval,/>The total number of the wave peak points of the frequency component in the frequency domain waveform diagram of the t time sequence interval,/>The audio noise receiving factors of the t-th ecological environment audio time sequence interval and the time sequence intervals above the t-th ecological environment audio time sequence interval are respectively obtained;
is the index of the wind noise interference degree of the t time sequence interval,/> The average value of the amplitude of all audio signals in the bird audio signal data in the t-th and y-th ecological environment audio time sequence intervals is/are respectivelyIs the total number of the time sequence intervals of the ecological environment audio frequency,/>The total number of time sequence intervals exists for bird song,/>Maximum amplitude points of the t-th ecological environment audio time sequence interval and the z-th bird song existence time sequence interval are respectively,/>As Euclidean distance function,/>The Euclidean distance between the maximum amplitude point of the time sequence interval of the t ecological environment audio frequency and the time sequence interval of the z-th bird song;
And D is the wind noise interference adjustment coefficient of the bird audio signal data.
When the noise reception of the t-th eco-environment audio time sequence interval is enhanced to some extent compared with the noise reception of the previous time sequence interval,The audio in the time sequence interval of the ecological environment audio is possibly added with new components, the stronger the aliasing effect of the time sequence interval of the ecological environment audio is, and when the aliasing effect of the time sequence interval of the t-th ecological environment audio is enhanced, the closer the time sequence interval exists to the bird song is, the greater the frequency of the time sequence interval isThe larger the value of the (2), the larger the amplitude average value corresponding to the time sequence interval of the ecological environment audio is compared with the amplitude average value of the rest time sequence intervals of the ecological environment audio, namely the first summation factor/>The larger the value of (i.e./>)The larger the time sequence interval is, the greater the possibility of existence of wind sound frequency is, and the greater the interference influence degree on the bird song frequency is. Thus, the greater the likelihood that bird behavioral audio is present in the field audio dataThe greater the value of D, the greater the degree of wind noise interference with the bird song in the field audio data, the greater the filter smoothing coefficient should be increased for the audio to better enhance the portion of the bird song in the audio.
Further, according to the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data, a wiener filtering smooth adjustment coefficient is constructed:
In the method, in the process of the invention, Smooth adjustment of coefficients for wiener filtering,/>The noise disturbance index of the bird audio signal data is represented by D, the wind noise disturbance adjustment coefficient of the bird audio signal data is represented by e, and the natural constant is represented by e.
When the influence degree of the bird voice frequency signal data by noise is larger, and the interference degree of the bird song voice frequency data by wind sound in the voice frequency is more obvious,The larger the value of (i.e./>)The larger the smoothing coefficient of wiener filtering should be increased.
Further, the adjusted wiener filter smoothing coefficients are calculated:
In the method, in the process of the invention, For the adjusted wiener filter smoothing coefficients,/>For the initial wiener filter smoothing coefficient, the empirical value is 0.55,/>Smooth adjustment of coefficients for wiener filtering,/>To adjust the parameters, the empirical value was taken to be 0.5.
So far, the adjusted wiener filter smoothing coefficient is obtained.
And S004, acquiring bird audio signal data after the filtering enhancement based on the wiener filtering smoothing coefficient by using a wiener filtering algorithm.
The bird audio signal data is subjected to filter enhancement based on wiener filtering by using the adjusted wiener filtering smoothing coefficient, the bird audio signal data is used as input of a wiener filtering algorithm, the adjusted wiener filtering smoothing coefficient is used as a smoothing parameter of the wiener filtering algorithm, and output of the wiener filtering algorithm is used as the bird audio signal data subjected to filter enhancement, and the wiener filtering algorithm is a known technology, so that the process of the wiener filtering algorithm is not repeated. A flow chart of an implementation of the present application is shown in fig. 2.
Thus, the filtering enhancement method for the ecological audio information is completed.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. The above description is only of the preferred embodiments of the present application and is not intended to limit the application, but any modifications, equivalent substitutions, improvements, etc. within the principles of the present application should be included in the scope of the present application.
Claims (10)
1. A filter enhancement method for eco-audio information, characterized in that the method comprises the steps of:
acquiring bird audio signal data;
Acquiring a frequency domain waveform diagram of each time sequence interval according to bird audio signal data, and acquiring all frequency component peak points and oscillation trailing wave peak points in the frequency domain waveform diagram of each time sequence interval according to the frequency domain waveform diagram of each time sequence interval; acquiring an oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval according to the frequency component peak point and the oscillation tailing peak point; acquiring noise disturbance indexes of bird audio signal data according to the oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval;
Acquiring all bird song existence time sequence intervals and ecological environment audio time sequence intervals according to the bird audio signal data; acquiring a wind-sound interference adjustment coefficient of bird audio signal data according to a bird song existence time sequence interval and an ecological environment audio time sequence interval; obtaining a wiener filter smooth adjustment coefficient according to the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data;
And acquiring wiener filter smoothing coefficients according to the wiener filter smoothing adjustment coefficients, and acquiring bird audio signal data after filter enhancement based on the wiener filter smoothing coefficients by using a wiener filter algorithm.
2. The method for enhancing the filtering of the ecological audio information according to claim 1, wherein the method for obtaining the frequency domain waveform of each time sequence interval according to the bird audio signal data and obtaining all the frequency component peak points and the oscillation trailing wave peak points in the frequency domain waveform of each time sequence interval according to the frequency domain waveform of each time sequence interval is as follows:
Taking the interval of each first preset parameter in the acquisition time of the bird audio signal data as a time sequence interval;
taking bird audio signal data in each time sequence interval as input of Fourier transform, and obtaining frequency domain data of each time sequence interval by utilizing the Fourier transform; drawing the frequency domain data of each time sequence interval by utilizing MATLAB software to obtain a frequency domain waveform diagram of each time sequence interval, wherein the horizontal axis of the frequency domain waveform diagram is frequency, and the vertical axis of the frequency domain waveform diagram is frequency domain energy;
For the frequency domain waveform diagram of each time sequence interval, taking the frequency domain energy of all wave crest points in the frequency domain waveform diagram as input of an Ojin threshold method, and obtaining a division threshold value by using the Ojin threshold method; and taking each wave peak point with the frequency domain energy larger than or equal to the division threshold value in the frequency domain waveform as a frequency component wave peak point, and taking each wave peak point with the frequency domain energy smaller than the division threshold value in the frequency domain waveform as an oscillation trailing wave peak point.
3. The method for enhancing the filtering of the ecological audio information according to claim 1, wherein the method for obtaining the oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval according to the frequency component peak point and the oscillation tailing peak point is as follows:
Taking each frequency component peak point in the frequency domain waveform diagram of each time sequence interval as a target frequency component peak point, and taking a sequence formed by all oscillation tailing peak points between the target frequency component peak point and the next frequency component peak point according to the ascending order of the frequencies as an oscillation tailing effect characteristic sequence of the target frequency component peak point;
And regarding the frequency domain waveform diagram of each time sequence interval, taking the frequency domain energy which is the frequency domain energy multiplied by the second preset parameter of the minimum frequency domain energy of all the frequency component peak points in the frequency domain waveform diagram as bandwidth construction energy, and taking the width of the waveform corresponding to each frequency component peak point, where the transverse straight line where the bandwidth construction energy is located, passing through each frequency component peak point as the bandwidth of each frequency component peak point in the frequency domain waveform diagram.
4. The method for enhancing filtering of ecological audio information according to claim 1, wherein the method for obtaining noise disturbance index of bird audio signal data according to the oscillation tailing effect characteristic sequence of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval and the bandwidth of each frequency component peak point in the frequency domain waveform diagram of each time sequence interval comprises the following steps:
Taking the difference value of the frequency domain energy of each frequency component peak point in the frequency domain waveform diagram and the frequency domain energy average value of all oscillation trailing wave peak points in the oscillation trailing effect characteristic sequence of each frequency component peak point as a molecule for the frequency domain waveform diagram of each time sequence interval; calculating the absolute value of the difference between the frequency domain energy of each frequency component peak point and the frequency domain energy mean value of all frequency component peak points in the frequency domain waveform chart, and taking the sum of the absolute value and a third preset parameter as a denominator; taking the average value of the sum of the absolute values of the ratios of the numerator and the denominator on the frequency domain waveform diagram as the spectrum energy difference index of the time sequence interval;
Acquiring an audio noise receiving factor of each time sequence interval according to the frequency spectrum energy difference index of each time sequence interval;
And taking the average value of the sum of the spectrum energy difference index and the audio noise factor and the sum of the spectrum energy difference index and the audio noise factor over all time sequence intervals as the noise disturbance index of the bird audio signal data.
5. The filtering enhancement method for ecological audio information as claimed in claim 4, wherein the method for obtaining the audio noise receiving factor of each time interval according to the spectrum energy difference index of each time interval is as follows:
Taking a spectrum energy difference index of each time sequence interval as a molecule for the frequency domain waveform diagram of each time sequence interval;
Calculating the absolute value of the difference between the bandwidth of each frequency component peak point in the frequency domain waveform diagram and the bandwidth average value of all frequency component peak points, calculating the accumulation sum of the absolute value on the frequency domain waveform diagram, and taking the product of the accumulation sum and the frequency domain energy of the fundamental frequency in the frequency domain waveform diagram as a denominator;
the ratio of the numerator to the denominator is used as the audio noise factor of the time sequence interval.
6. The method for enhancing filtering of ecological audio information according to claim 1, wherein the method for acquiring all of the bird song presence timing intervals and the ecological environment audio timing intervals from the bird audio signal data is as follows:
Taking the amplitude values of all audio signals in the bird audio signal data as the input of an Ojin threshold method, and obtaining a bird song audio dividing threshold value by using the Ojin threshold method;
Calculating the average value of the amplitude values of all audio signals in the bird audio signal data in each time sequence interval, taking each time sequence interval with the average value of the amplitude values being more than or equal to a bird song audio frequency dividing threshold value as a bird song existence time sequence interval, and taking each time sequence interval with the average value of the amplitude values being less than the bird song audio frequency dividing threshold value as an ecological environment audio frequency time sequence interval.
7. The filtering enhancement method for ecological audio information according to claim 1, wherein the method for acquiring the wind noise interference adjustment coefficient of the bird audio signal data according to the bird song presence time sequence interval and the ecological environment audio time sequence interval is as follows:
Taking each ecological environment audio time sequence interval as a target ecological environment audio time sequence interval, and taking the total number of frequency component peak points in a frequency domain waveform diagram of the target ecological environment audio time sequence interval as bird behavior audio weight factors of the target ecological environment audio time sequence interval when the audio noise receiving factor corresponding to the target ecological environment audio time sequence interval is smaller than or equal to the audio noise receiving factor of the last time sequence interval; when the audio noise factor corresponding to the target ecological environment audio time sequence interval is larger than the audio noise factor of the previous time sequence interval, calculating the difference value of the audio noise factor corresponding to the target ecological environment audio time sequence interval and the audio noise factor of the previous time sequence interval, and taking the sum of the total number of the frequency component peak points and the difference value as a wind-up audio weight factor of the target ecological environment audio time sequence interval;
Acquiring a wind sound interference degree index of the target ecological environment audio time sequence interval according to a wind sound weighting factor of the target ecological environment audio time sequence interval;
And taking the average value of the wind noise interference degree indexes corresponding to all the ecological environment audio time sequence intervals as a wind noise interference adjustment coefficient of the bird audio signal data.
8. The method for enhancing filtering of audio information according to claim 7, wherein the method for obtaining the wind noise interference level index of the target audio time sequence section according to the wind noise weight factor of the target audio time sequence section comprises:
Calculating the amplitude average value of all audio signals in the bird audio signal data in each ecological environment audio time sequence interval, calculating the absolute value of the difference between the amplitude average value corresponding to the target ecological environment audio time sequence interval and the amplitude average value corresponding to each ecological environment audio time sequence interval, and taking the natural constant as a base number and taking the mapping result of the sum of the absolute value on the total number of the ecological environment audio time sequence intervals as an index as a first summation factor;
Taking the air-blast audio weight factors of the target ecological environment audio time sequence interval as molecules; calculating a measurement distance between a maximum amplitude point of the target ecological environment audio time sequence interval and a maximum amplitude point of each bird song existence time sequence interval, and taking the sum of the measurement distances on the total number of the bird song existence time sequence intervals as a denominator;
and taking the sum of the ratio of the numerator and the denominator and the first summation factor as an index of the wind noise interference degree of the target ecological environment audio time sequence interval.
9. The method for enhancing the filtering of the ecological audio information according to claim 1, wherein the method for obtaining the wiener filter smoothing adjustment coefficient according to the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data is as follows:
And calculating the product of the noise disturbance index and the wind noise disturbance adjustment coefficient of the bird audio signal data, and taking the product as a wiener filtering smooth adjustment coefficient.
10. The method for enhancing the filtering of the ecological audio information according to claim 1, wherein the method for obtaining wiener filter smoothing coefficients according to wiener filter smoothing adjustment coefficients and obtaining the bird audio signal data after the filtering enhancement based on the wiener filter smoothing coefficients by using a wiener filter algorithm is as follows:
Calculating the sum of the wiener filter smoothing adjustment coefficient and a fourth preset parameter, and taking the product of the sum and the preset smoothing coefficient as the wiener filter smoothing coefficient;
And taking the bird audio signal data and the wiener filter smoothing coefficient as inputs of a wiener filter algorithm respectively, and obtaining the bird audio signal data after the filter enhancement by using the wiener filter algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410179556.5A CN117727314B (en) | 2024-02-18 | 2024-02-18 | Filtering enhancement method for ecological audio information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410179556.5A CN117727314B (en) | 2024-02-18 | 2024-02-18 | Filtering enhancement method for ecological audio information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117727314A CN117727314A (en) | 2024-03-19 |
| CN117727314B true CN117727314B (en) | 2024-04-26 |
Family
ID=90205709
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410179556.5A Active CN117727314B (en) | 2024-02-18 | 2024-02-18 | Filtering enhancement method for ecological audio information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117727314B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118675541B (en) * | 2024-08-21 | 2024-11-15 | 西安腾谦电子科技有限公司 | Audio data secure transmission method and system in complex environment |
| CN119296565B (en) * | 2024-12-10 | 2025-04-01 | 北京国旺盛源智能终端科技有限公司 | Detachable split operation device with audio data acquisition function |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103440871A (en) * | 2013-08-21 | 2013-12-11 | 大连理工大学 | Method for suppressing transient noise in voice |
| CN105513605A (en) * | 2015-12-01 | 2016-04-20 | 南京师范大学 | Voice enhancement system and method for cellphone microphone |
| CN108899042A (en) * | 2018-06-25 | 2018-11-27 | 天津科技大学 | A kind of voice de-noising method based on mobile platform |
| CN112750442A (en) * | 2020-12-25 | 2021-05-04 | 浙江弄潮儿智慧科技有限公司 | Nipponia nippon population ecosystem monitoring system with wavelet transformation and wavelet transformation method thereof |
| WO2021197566A1 (en) * | 2020-03-30 | 2021-10-07 | Harman Becker Automotive Systems Gmbh | Noise supression for speech enhancement |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11017798B2 (en) * | 2017-12-29 | 2021-05-25 | Harman Becker Automotive Systems Gmbh | Dynamic noise suppression and operations for noisy speech signals |
-
2024
- 2024-02-18 CN CN202410179556.5A patent/CN117727314B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103440871A (en) * | 2013-08-21 | 2013-12-11 | 大连理工大学 | Method for suppressing transient noise in voice |
| CN105513605A (en) * | 2015-12-01 | 2016-04-20 | 南京师范大学 | Voice enhancement system and method for cellphone microphone |
| CN108899042A (en) * | 2018-06-25 | 2018-11-27 | 天津科技大学 | A kind of voice de-noising method based on mobile platform |
| WO2021197566A1 (en) * | 2020-03-30 | 2021-10-07 | Harman Becker Automotive Systems Gmbh | Noise supression for speech enhancement |
| CN112750442A (en) * | 2020-12-25 | 2021-05-04 | 浙江弄潮儿智慧科技有限公司 | Nipponia nippon population ecosystem monitoring system with wavelet transformation and wavelet transformation method thereof |
Non-Patent Citations (1)
| Title |
|---|
| LCMV分频的改进维纳滤波后置波束形成算法;郭业才;陈小燕;王超;;电子测量与仪器学报;20171015(第10期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117727314A (en) | 2024-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117727314B (en) | Filtering enhancement method for ecological audio information | |
| KR101034831B1 (en) | Wind noise suppression system | |
| CN117033911B (en) | Step counting analysis method based on intelligent glasses data | |
| CN117727313B (en) | Intelligent noise reduction method for wild bird sound data | |
| CN113325277A (en) | Partial discharge processing method | |
| CN106504760B (en) | Wideband background noise and speech separation detection system and method | |
| CN106052852B (en) | A kind of detection method and device of pulse acoustical signal | |
| CN109932624B (en) | Cable partial discharge period narrow-band interference denoising method based on Gaussian scale space | |
| CN110974189A (en) | Method, device, equipment and system for detecting signal quality of pulse wave | |
| CN116626408A (en) | Power supply ripple noise detection method based on machine learning | |
| KR101649243B1 (en) | Method and apparatus for detecting correctness of pitch period | |
| CN103778921A (en) | Method for eliminating nonuniform noise in speech collected by radar | |
| CN105429719A (en) | Strong interference signal detection method based on power spectrum and multiple dimensioned wavelet transformation analysis | |
| CN112183407A (en) | Tunnel seismic wave data denoising method and system based on time-frequency domain spectral subtraction | |
| CN110287853B (en) | Transient signal denoising method based on wavelet decomposition | |
| CN117727332B (en) | Ecological population assessment method based on language spectrum feature analysis | |
| US10013992B2 (en) | Fast computation of excitation pattern, auditory pattern and loudness | |
| Qiao et al. | Spectral entropy based dolphin whistle detection algorithm and its possible application for biologically inspired communication | |
| CN118173107B (en) | Bird sound quality analysis method based on multi-mode depth feature level fusion | |
| CN116935880B (en) | Integrated machine man-machine interaction system and method based on artificial intelligence | |
| CN110542927B (en) | Variable window weighted seismic data spike noise suppression method | |
| CN116626760B (en) | Stratum discontinuity detection method and device based on self-adaptive high-order maximum entropy WVD | |
| CN113948088A (en) | Voice recognition method and device based on waveform simulation | |
| CN114114400A (en) | An Effective Signal Picking Method for Microseismic Events | |
| CN112327260A (en) | Method and device for suppressing pulsed interference signal in SAR echo data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |