CN117727330B - Biological diversity prediction method based on audio decomposition - Google Patents
Biological diversity prediction method based on audio decomposition Download PDFInfo
- Publication number
- CN117727330B CN117727330B CN202410179245.9A CN202410179245A CN117727330B CN 117727330 B CN117727330 B CN 117727330B CN 202410179245 A CN202410179245 A CN 202410179245A CN 117727330 B CN117727330 B CN 117727330B
- Authority
- CN
- China
- Prior art keywords
- audio
- ecological
- index
- interval
- bird song
- 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
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 68
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000005236 sound signal Effects 0.000 claims abstract description 107
- 230000004044 response Effects 0.000 claims abstract description 30
- 238000012545 processing Methods 0.000 claims abstract description 14
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 9
- 238000009432 framing Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 description 12
- 230000003044 adaptive effect Effects 0.000 description 5
- 230000007613 environmental effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 241000894007 species Species 0.000 description 4
- 238000001228 spectrum Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 210000000056 organ Anatomy 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 206010011469 Crying Diseases 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000037433 frameshift Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000009295 sperm incapacitation Effects 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Auxiliary Devices For Music (AREA)
Abstract
The invention relates to the technical field of audio signal processing, in particular to a biological diversity prediction method based on audio decomposition, which comprises the following steps: collecting each ecological audio signal at different moments in a day; extracting an ecological audio windowing signal; acquiring class width of each clustering class of the ecological audio Hanning window according to the frequency domain amplitude information of the ecological audio windowing signal and a clustering algorithm; constructing a time domain bird song index and a frequency domain bird song index of the ecological audio Hanning window; calculating a bird song complex index of the ecological audio Hanning window and a bird song response index of an audio interval based on the complex bird song complex index; and (3) obtaining the decomposition layer number of each audio interval when the VMD algorithm is decomposed, and constructing the biological abundance index by combining the Babbitt coefficient between modal components under each decomposition layer number after decomposition of different audio intervals and the acoustic diversity index of each audio interval. The invention can realize accurate assessment of biological diversity.
Description
Technical Field
The application relates to the technical field of audio signal processing, in particular to a biological diversity prediction method based on audio decomposition.
Background
Biodiversity refers to the abundance of life forms in a region, and is the sum of organisms and their ecological complexes that are regularly combined with the environment to form a stable ecological complex, and the various ecological processes associated therewith. Biodiversity is a necessary resource for maintaining the survival and development of human society, however, is influenced by human activities and industrialization, habitat environments of organisms are damaged, species and quantity of organisms are continuously disappearing, species extinction speed is increasing, and in order to effectively protect biodiversity, accurate grasp of the current situation of biodiversity is required, so rapid biodiversity monitoring, evaluation, protection, management, planning and the like are being carried out.
The bird diversity can well reflect the biological diversity due to wide distribution and sensitivity to the change of the ecological environment, and meanwhile, the bird diversity can well reflect the biological diversity, so that the biological diversity can be reflected by monitoring the bird biological diversity.
The traditional algorithm for decomposing the bird audio signal, such as VMD (Variational Mode Decomposition) algorithm, has better capability of adapting to the nonstationary signal, has certain advantages for decomposing the time-varying characteristics of the bird sound, and meanwhile, the VMD algorithm has a set of modal components as a decomposition result, each modal component corresponds to one frequency component of the original audio signal, so that the method has stronger interpretation, but the bird sound usually has complex frequency spectrum structure and transient sound, such as short-time chirp, ringing and the like, so that the problem that the number of decomposition layers is not easy to determine when the traditional VMD algorithm is used for decomposing the bird audio signal, and further, the modal components cannot be accurately extracted, and the biodiversity cannot be accurately estimated.
Disclosure of Invention
In order to solve the technical problems, the invention provides a biological diversity prediction method based on audio decomposition to solve the existing problems.
The biological diversity prediction method based on audio decomposition adopts the following technical scheme:
one embodiment of the present invention provides an audio decomposition-based biodiversity prediction method including the steps of:
collecting audio signals at different moments in a day, denoising, and recording the audio signals as ecological audio signals;
extracting each audio interval of each ecological audio signal; carrying out framing windowing treatment on the audio interval to obtain an ecological audio windowing signal; acquiring class width of each clustering class of the ecological audio Hanning window according to the frequency domain amplitude information of the ecological audio windowing signal and a clustering algorithm; obtaining a time domain bird song index of the ecological audio hanning window according to the peak value and the change of the audio signal in the ecological audio hanning window; obtaining a frequency domain bird song index of the ecological audio Hanning window according to class widths of all clustering classes in the ecological audio Hanning window and the amplitudes contained in all clustering classes; constructing a bird song complex index of the ecological audio Hanning window according to the time domain bird song index and the frequency domain bird song index; constructing a bird song response index of the audio interval according to the bird song complex index of each ecological audio Hanning window in the audio interval and the maximum value of the bird song complex index;
and according to the bird song response index of each audio interval, the maximum value of the bird song response index in the ecological audio signal of each audio interval and the preset VMD algorithm decomposition layer number maximum value, the decomposition layer number of each audio interval during VMD algorithm decomposition is obtained, and the biological abundance index is constructed by combining the Babbitt coefficient between modal components under each decomposition layer number after decomposition of different audio intervals and the acoustic diversity index of each audio interval, wherein the biological abundance index and the biological diversity form a positive correlation.
Further, each audio interval of the respective eco-audio signal includes:
for each eco-audio signal, the eco-audio signal is used as the input of the voice activation detection VAD algorithm, and the audio interval in which the voice activity signal exists in the eco-audio signal is output as each audio interval of the eco-audio signal.
Further, the step of performing frame windowing processing on the audio interval to obtain an ecological audio windowing signal includes:
and carrying out frame division processing on each audio interval, respectively applying a hanning window function to each frame to obtain an ecological audio hanning window, and forming an ecological audio windowing signal of the audio interval from audio signals of all frames of the ecological audio hanning windows of the audio interval.
Further, the obtaining the class width of each cluster class of the ecological audio hanning window according to the frequency domain amplitude information of the ecological audio windowing signal and the clustering algorithm comprises the following steps:
and taking the ecological audio windowing signal as the input of the fast Fourier transform to obtain an ecological audio windowing frequency domain signal, adopting an Ojin method to obtain a segmentation threshold value of the frequency domain signal amplitude, adopting a clustering algorithm to cluster the amplitude of the frequency domain signal which is greater than or equal to the segmentation threshold value to obtain each clustering class, calculating the absolute value of the difference value of any two frequencies of the frequency domain signal in the clustering class, and taking the maximum value of the absolute value of the difference value of all any two frequencies in the clustering class as the class width of the clustering class.
Further, the time domain bird song index of the ecological audio hanning window is obtained according to the peak value and the change of the audio signal in the ecological audio hanning window, and the method comprises the following steps:
calculating the average value of the left slope and the right slope of each peak value of the audio signal in the ecological audio hanning window, obtaining the product of the peak value and the average value, and taking the average value of the product of all the peak values in the ecological audio hanning window as the time domain bird song index of the ecological audio hanning window.
Further, the frequency domain bird song index of the ecological audio hanning window comprises:
calculating the ratio of the average value of the amplitude values of each clustering class to the class width in the ecological audio Hanning window, taking the sum value of the ratio of 1 as the true number of the logarithmic function taking the natural constant as the base number, and taking the sum value of the calculation results of the logarithmic function of all the clustering classes as the frequency domain bird-ringing index of the ecological audio Hanning window.
Further, the bird song complex index of the ecological audio hanning window is the sum of the time domain bird song index and the frequency domain bird song index.
Further, the construction of the bird song response index of the audio interval includes:
for each audio interval, counting the maximum value of the bird song complex indexes of all the ecological audio hanning windows in the audio interval, calculating the difference value of the maximum value and the bird song complex indexes of all the ecological audio hanning windows in the audio interval, obtaining the ratio of the bird song complex indexes of all the ecological audio hanning windows in the audio interval to the difference value, and taking the sum of the ratio of all the ecological audio hanning windows in the audio interval as the bird song response index of the audio interval.
Further, the obtaining the number of decomposition layers of each audio interval during the decomposition of the VMD algorithm includes:
counting the maximum value of the bird song response index in the ecological audio signal where the audio interval is located, obtaining the ratio of the bird song response index to the maximum value of each audio interval, calculating the difference value between the maximum value of the preset decomposition layer number and the minimum value of the preset decomposition layer number, obtaining the rounded value of the product of the difference value and the ratio, and taking the sum of the rounded value and the minimum value as the decomposition layer number of the audio interval when the VMD algorithm is decomposed.
Further, the constructing the biological abundance index includes:
calculating the Pasteur coefficients between each modal component of each audio interval of the ecological audio signal after the decomposition of the VMD algorithm and all modal components of other audio intervals, obtaining the maximum value of the Pasteur coefficient of each modal component of each audio interval, and calculating the sum of the reciprocal values of the maximum values of all modal components of each audio interval; calculating the product of the acoustic diversity index of each audio interval and the sum value;
and recording the sum value of the products of all the audio intervals of the ecological audio signals as ecological audio sum values, and summing the acquired ecological audio sum values of all the ecological audio signals to obtain the biological abundance index.
The invention has at least the following beneficial effects:
the invention constructs a bird song complex index by carrying out framing and windowing treatment on the ecological audio signal and analyzing the characteristics of bird sounding organs, parts and the like, and reflects the number characteristics of birds in a Hanning window; further, by analyzing the difference between the sounding characteristics of birds and part of environmental noise, a bird song response index is constructed based on the bird song complex index, the characteristics of the number of birds in an audio interval are reflected, and misjudgment of bird song and environmental noise is avoided; and then, the number of decomposition layers of the VMD algorithm is determined based on the bird song response index and the acoustic diversity index in a self-adaptive manner, so that the problem of inaccurate modal component extraction caused by over-decomposition or under-decomposition is avoided, different bird song can be represented by the decomposed modal components, the biological abundance index is further constructed, and the biological diversity is estimated more accurately.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of a method for audio decomposition-based biodiversity prediction according to the present invention;
fig. 2 is a schematic diagram of the VMD algorithm decomposition layer number adaptive determination.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description refers to the specific implementation, structure, characteristics and effects of the audio decomposition-based biological diversity prediction method according to the invention with reference to the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the audio decomposition-based biodiversity prediction method provided by the invention with reference to the accompanying drawings.
The method for predicting biodiversity based on audio decomposition according to an embodiment of the present invention specifically provides a method for predicting biodiversity based on audio decomposition, please refer to fig. 1, which includes the following steps:
and S001, collecting bird audio signals and preprocessing.
The embodiment aims at analyzing the audio information in the ecological natural protection area of the region so as to accurately evaluate the biodiversity in the region. For convenience of description, this embodiment will be described taking the area a as an example. In the ecological natural protection area of the A land, the microphone array formed by four paths of sound pick-up devices can effectively collect sound signals in a 200m multiplied by 200m circular area, the sampling frequency is recorded as f, the empirical value is taken as 44.1kHz in the embodiment, the duration of collecting audio signals once is recorded as T, the empirical value is taken as 1h in the embodiment, and because the activity time of different organisms can be different, the audio signals can be collected once respectively in 0:00, 8:00 and 16:00, and a set operator at the collection time can select the collected audio signals by himself, and the collected audio signals are respectively recorded as first, second and third original ecological audio signals.
Because the environment in the ecological natural protection area is complex, the collected original ecological audio signal may have more noise such as wind noise and rain noise, so as to avoid the influence of noise on the subsequent audio processing step, the obtained ecological audio signal needs to be subjected to denoising processing, and the common audio denoising technology has wavelet transform denoising, spectrum reduction algorithm, adaptive filtering denoising and the like, so that the noise in different environments is adapted, a better denoising effect is achieved, and the obtained ecological audio signal is subjected to denoising processing by using the adaptive filtering algorithm. The adaptive filtering algorithm is a known technology, and the specific process is not described in detail in this embodiment.
Thus, the preprocessed ecological audio signals are respectively recorded as first, second and third ecological audio signals.
Step S002, determining the decomposition layer number of the VMD algorithm based on the waveform characteristics of the ecological audio signal after framing and windowing, further decomposing the audio interval of the ecological audio signal, and realizing the evaluation of the biodiversity based on the Babbitt coefficient of the decomposed modal component combined with the acoustic diversity index.
First, the embodiment will identify the audio interval in each ecological audio signal, and the specific extraction process is as follows:
in natural environment, animals are safe to protect themselves and are prevented from being discovered by natural enemies, and the sounds are not continuously emitted, so that the first, second and third ecological audio signals obtained in the steps have mute parts, and in order to facilitate analysis of the non-mute parts in subsequent steps, the first, second and third ecological audio signals are used as input of a voice activation detection VAD (Voice Activity Detection) algorithm respectively, and an audio interval with voice activity signals in the first, second and third ecological audio signals is obtained. The VAD algorithm is a known technology, and the specific process of this embodiment is not described in detail.
In order to avoid the problem that partial modal energy is dominant when the subsequent VMD algorithm is used to decompose the audio signals of each audio interval in the first, second and third ecological audio signals, the peak normalization method is used to normalize each audio interval in the first, second and third ecological audio signals respectively, so that the audio intervals have a standard amplitude range. The peak normalization method is a known technique, and the specific process is not described in detail in this embodiment.
Then, in this embodiment, each ecological audio signal is subjected to frame windowing, and a bird song complex index is constructed, and the specific processes of windowing and constructing the bird song complex index are as follows:
because the size of the bird sound cavity is usually smaller, the sound emitted by the birds is usually sharper, the high-frequency part of the obtained audio signal is more remarkable, the time for the birds to emit the sound is usually shorter, the high-instantaneous energy of the audio signal is caused by the sharp bird sound in a short time, and sudden vibration or wave peaks appear in the audio signal; when different birds sound at the same time, the sounds of the different birds are overlapped together because the sounds of the different birds generally have different frequency ranges and frequency characteristics, and meanwhile, the sounds of the different birds have short duration, so that rapid waveform changes occur in the obtained audio signals. According to this embodiment, each ecological audio signal is subjected to frame windowing processing, and a bird song complex index is constructed, and for convenience of description, the first ecological audio signal is taken as an example for description, and the construction process of the bird song complex index is as follows:
and carrying out frame division processing on each audio interval of the first ecological audio signal, wherein the length of the audio signal of each frame is 20ms, the frame shift is 10ms, and each frame of the audio signal is respectively applied with a hanning window function to obtain an ecological audio hanning window which is used for reducing the amplitudes of two ends of the audio signal of each frame and avoiding spectrum distortion caused by spectrum leakage, thereby reducing the accuracy of the subsequent analysis result.
The audio signals of all frames of ecological audio Hanning windows after framing and windowing are recorded as ecological audio windowing signals, the ecological audio windowing signals are used as the input of FFT (fast Fourier transform) to obtain corresponding ecological audio windowing frequency domain signals, the amplitude of the ecological audio windowing frequency domain signals is used as the input of an Ojin threshold segmentation method, segmentation thresholds are output, the amplitude which is greater than or equal to the segmentation thresholds in the ecological audio windowing frequency domain signals is used as the input of a Canopy clustering algorithm, the sizes of a strong threshold and a weak threshold in the Canopy clustering algorithm are respectively obtained by taking the empirical values 1 and 3, the output of the Canopy clustering algorithm as each clustering category, then the absolute value of the difference value of any two frequencies of the ecological audio windowing frequency domain signals in each clustering category is calculated, and the maximum value of the absolute value of the difference value of any two frequencies in the clustering categories is used as the class width of the clustering category, wherein the fast Fourier transform, the Ojin method and the Canopy clustering algorithm are known techniques, and the specific processes are not repeated. Then a bird song complex index can be constructed based on this, with the following calculation formula:
wherein the method comprises the steps ofTime domain bird song index, +_f, representing the b-th ecological audio hanning window in the a-th audio interval in the first ecological audio signal>Representing the number of peaks in the hanning window of the b-th ecological audio in the a-th audio interval in the first ecological audio signal,/v>、/>Respectively representing the left slope and the right slope of the peak value of the c-th audio signal in the b-th ecological audio Hanning window in the a-th audio interval in the first ecological audio signal, wherein the calculation method of the left slope and the right slope is the slope of energy between the peak value of the c-th audio signal and one of the left and right nearest audio signal sampling points, and the calculation method comprises the steps of>Representing the size of the peak of the c-th audio signal in the b-th ecological audio hanning window in the a-th audio interval in the first ecological audio signal;
frequency domain bird song index, +_f, representing the b-th ecological audio hanning window in the a-th audio interval in the first ecological audio signal>Representing the clustering number of the clustering frequency domain signals of the ecological audio windowing frequency domain of the b-th ecological audio Hanning window in the a-th audio interval in the first ecological audio signal after the clustering in the steps, wherein ln () represents a logarithmic function based on a natural constant,、/>the method comprises the steps that the average value of the amplitude of the g category of the ecological audio windowing frequency domain signal of the b-th ecological audio Hanning window in the a-th audio interval in the first ecological audio signal is clustered;
a bird song complexity index representing a b-th eco-audio hanning window in an a-th audio interval in the first eco-audio signal.
In the first ecological audio signal, the larger the peak value in the ecological audio windowing signal is, namelyThe larger the slope around the peak in the eco-audio windowed signal is, the larger the slope is, i.e. +.>、/>The larger the peak value is, the more sharp the sound in the hanning window is, the shorter the sound duration is, and the more accords with the characteristics of bird crying, meanwhile, the larger the peak value is, the more the birds possibly sound at the same time, so that the corresponding time domain bird song index is larger; in the first ecological audio signal, the larger the average value of elements in clusters in the ecological audio windowing frequency domain signal is, namely +.>The larger the cluster, the smaller the class width within the cluster, i.e. +.>The smaller the size, the clearer the sound in the Hanning window, the less noise is present, and the more clusters are clustered, namely +.>The larger the bird species, the more birds that sound within the hanning window, so the calculated frequencyThe greater the domain bird song index; the greater the time domain and frequency domain bird song indexes, i.e. +.>、/>The larger the bird song, the more bird sounds that are present in the hanning window, so the greater the calculated bird song complexity index.
And carrying out the same processing on the ecological audio windowing signals of each audio interval in the second and third ecological audio signals according to the steps, so as to obtain the bird song complex index of the hanning window of each audio interval in the first, second and third ecological audio signals.
Further, in this embodiment, the bird song response index is constructed based on the bird song complex index, and the specific construction process is as follows:
the bird song complex index of each hanning window in each audio interval obtained through the steps reflects the characteristic that the bird song number is contained in the extremely short time period corresponding to each hanning window, in order to avoid that part of environmental noise such as stone collision is similar to the bird song characteristic, the bird song complex index calculated in the extremely short time corresponding to one hanning window is larger, misjudgment is caused, in nature, when birds call for puppets, announce the territory, send out vigilance and the like, interaction behaviors with other birds exist, namely, when one bird sends out a song with a certain purpose, other birds send out responses, so that bird song response indexes are not generally constructed only in the extremely short time corresponding to one hanning window, and the characteristic that the bird song number is reflected in the audio interval is based on the bird song complex index.
Wherein the method comprises the steps ofA bird song response index indicating the a-th audio interval in the first physiological audio signal, +.>Representing the number of hanning windows, +.>A bird song complexity index representing a b-th ecological audio hanning window in an a-th audio interval in the first ecological audio signal, +.>Representing the maximum value of the bird song complex index of all the ecological audio hanning windows in the a-th audio interval in the first ecological audio signal, +.>And represents a coordination coefficient, which is used for avoiding incapacitation caused by 0 denominator, and takes an empirical value of 1 in the embodiment.
The greater the bird song complexity index of the b-th ecological audio Hanning window in the a-th audio interval in the first ecological audio signal, namelyThe larger the Hanning window, the more likely the birds will be in the very short time, and the smaller the difference between the bird song complex index and the maximum value of the bird song complex index, i.e. the more than one ecological voice frequency Hanning windowThe smaller the bird song index is, the more birds are likely to be in the very short time corresponding to the hanning windows, namely the more the mutual response phenomenon among birds is likely to be caused, so the calculated bird song response index is larger.
And carrying out the same processing on each audio interval in the second and third ecological audio signals according to the steps, so as to obtain the bird song response index of each audio interval in the first, second and third ecological audio signals.
Finally, the embodiment determines the number of decomposition layers based on the bird song response index, and the specific determination process is as follows:
the bird song response index of each audio interval obtained through the steps reflects the number characteristics of bird song in the audio interval, when the number of bird song in the audio interval is larger, the number of decomposition layers when the VMD algorithm is used for decomposing the audio signal is larger, so that the song of different birds is decomposed, and accordingly, the number of decomposition layers when the VMD algorithm is used for decomposing the signal of each audio interval in the ecological audio signal can be determined, and the first ecological audio signal is taken as an example, and the calculation formula of the number of decomposition layers is as follows:
wherein the method comprises the steps ofRepresents the number of decomposition layers of the a-th audio interval of the first physiological audio signal, int () represents a rounding function,>、/>respectively representing the maximum value and the minimum value of the decomposition layer number, and in the embodiment, respectively taking the empirical values of 8 and 2 #, respectively>A bird song response index indicating the a-th audio interval in the first physiological audio signal, +.>The maximum value of the bird song response index of each audio interval in the first ecological audio signal is represented. The schematic diagram of the adaptive determination of the number of decomposition layers of the VMD algorithm when decomposing each audio interval is shown in FIG. 2.
And carrying out the same processing on the second and third ecological audio signals according to the steps, so as to obtain the decomposition layer number of each audio interval in the second and third ecological audio signals, respectively putting the decomposition layer numbers of each audio interval in the first, second and third ecological audio signals into three sequences according to the time sequence, and respectively marking the obtained sequences as the first, second and third ecological audio signal decomposition layer number sequences.
And step S003, decomposing each ecological audio signal by using a VMD variation modal decomposition algorithm based on the obtained decomposition layer number, and constructing a biological abundance index to evaluate biological diversity.
By the decomposition layer number sequence of each ecological audio signal obtained by the steps, each audio interval in each ecological audio signal can be used as the input of the VMD algorithm, the decomposition layer number corresponding to each audio interval in each ecological audio signal is used as the decomposition layer number of the VMD algorithm, and penalty factors in the VMD algorithmSize-checked value 2000, convergence tolerance +.>Size takes the experience valueOutputting the modal components of each audio interval in the physiological audio signal, then the barycenter coefficient between each modal component can be calculated, so as to construct a biological abundance index and evaluate the biological diversity, wherein the calculation of the barycenter coefficient is a known technology, and the detailed process is not repeated in this embodiment. The calculation formula of the biological abundance index is as follows:
wherein the method comprises the steps ofRepresenting the biological abundance index,/->Representing the number of collected eco-audio signals, the empirical value of 3,/in this embodiment is taken>Represents the ithThe number of audio intervals of the eco-audio signal, +.>Representing the acoustic diversity index calculated by the jth audio interval in the ith eco-audio signal,/->Represents the decomposition layer number, [ solution ] of the jth audio interval in the ith ecological audio signal when the jth audio interval is decomposed by the VMD algorithm>Representing the maximum value of the Babbitt coefficient between the kth modal component of the jth audio interval in the ith ecological audio signal after the jth audio interval is decomposed by the VMD algorithm and all modal components of all audio intervals of the ith ecological audio signal, and>the coordination coefficient is expressed to avoid that the denominator is 0, and the empirical value is 1 in this embodiment. The calculation of the acoustic diversity index is a well-known technique, and the detailed process is not repeated in this embodiment. For convenience of description, the following will be madeAnd recording the ecological audio sum value, and summing the ecological sum values of all the acquired ecological audio signals to obtain the final biological abundance index.
The greater the acoustic diversity index within an audio interval, i.eThe larger the number of bird species or individuals is, the more the corresponding bird diversity is, the more the corresponding biodiversity is, and the smaller the inter-modal component Babbitt coefficient after VMD algorithm decomposition is, namely +.>The smaller the difference between the decomposed modal components is, the more likely the decomposed modal components belong to different organisms, the more the corresponding biodiversity is, and the estimation isThe greater the calculated biological abundance index.
In summary, the embodiment of the invention constructs the bird song complex index by carrying out framing windowing treatment on the ecological audio signal and analyzing the characteristics of bird sounding organs, parts and the like, and reflects the number characteristics of birds in a hanning window; further, by analyzing the difference between the sounding characteristics of birds and part of environmental noise, a bird song response index is constructed based on the bird song complex index, the characteristics of the number of birds in an audio interval are reflected, and misjudgment of bird song and environmental noise is avoided;
further, the number of decomposition layers of the VMD algorithm is determined based on the bird song response index and the acoustic diversity index in a self-adaptive manner, so that the problem of inaccurate modal component extraction caused by over-decomposition or under-decomposition is avoided, different bird song can be represented by the decomposed modal components, the biological abundance index is further constructed, and the biological diversity is estimated more accurately.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. And the foregoing description has been directed to specific embodiments of this specification. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and the same or similar parts of each embodiment are referred to each other, and each embodiment mainly describes differences from other embodiments.
The above embodiments are only for illustrating the technical solution of the present application, and are not limiting; the technical solutions described in the foregoing embodiments are modified or some of the technical features are replaced equivalently, so that the essence of the corresponding technical solutions does not deviate from the scope of the technical solutions of the embodiments of the present application, and all the technical solutions are included in the protection scope of the present application.
Claims (5)
1. A method for biodiversity prediction based on audio decomposition, the method comprising the steps of:
collecting audio signals at different moments in a day, denoising, and recording the audio signals as ecological audio signals;
extracting each audio interval of each ecological audio signal; carrying out framing windowing treatment on the audio interval to obtain an ecological audio windowing signal; acquiring class width of each clustering class of the ecological audio Hanning window according to the frequency domain amplitude information of the ecological audio windowing signal and a clustering algorithm; obtaining a time domain bird song index of the ecological audio hanning window according to the peak value and the change of the audio signal in the ecological audio hanning window; obtaining a frequency domain bird song index of the ecological audio Hanning window according to class widths of all clustering classes in the ecological audio Hanning window and the amplitudes contained in all clustering classes; constructing a bird song complex index of the ecological audio Hanning window according to the time domain bird song index and the frequency domain bird song index; constructing a bird song response index of the audio interval according to the bird song complex index of each ecological audio Hanning window in the audio interval and the maximum value of the bird song complex index;
acquiring the decomposition layer number of each audio interval when the VMD algorithm is decomposed according to the bird song response index of each audio interval, the maximum value of the bird song response index in the ecological audio signal of each audio interval and the preset VMD algorithm decomposition layer number maximum value, and constructing a biological abundance index by combining the Babbitt coefficient between modal components under each decomposition layer number after decomposition of different audio intervals and the acoustic diversity index of each audio interval, wherein the biological abundance index and the biological diversity form a positive correlation;
the obtaining the class width of each clustering class of the ecological audio hanning window according to the frequency domain amplitude information of the ecological audio windowing signal and the clustering algorithm comprises the following steps:
using the ecological audio windowing signal as the input of the fast Fourier transform to obtain an ecological audio windowing frequency domain signal, adopting an Ojin method to obtain a segmentation threshold value of the frequency domain signal amplitude, adopting a clustering algorithm to cluster the amplitude of the frequency domain signal which is greater than or equal to the segmentation threshold value to obtain each clustering class, calculating the absolute value of the difference value of any two frequencies of the frequency domain signal in the clustering class, and taking the maximum value of the absolute value of the difference value of all any two frequencies in the clustering class as the class width of the clustering class;
the time domain bird song index of the ecological audio hanning window is obtained according to the peak value and the change of the audio signal in the ecological audio hanning window, and the method comprises the following steps:
calculating the average value of the left slope and the right slope of each peak value of an audio signal in the ecological audio hanning window, obtaining the product of the peak value and the average value, and taking the average value of the product of all the peak values in the ecological audio hanning window as the time domain bird song index of the ecological audio hanning window;
the frequency domain bird song index of the ecological audio hanning window comprises:
calculating the ratio of the average value of the amplitude values of each clustering class to the class width in the ecological audio Hanning window, taking the sum of the ratio of 1 as the true number of the logarithmic function taking the natural constant as the base, and taking the sum of the calculation results of the logarithmic functions of all the clustering classes as the frequency domain bird-ringing index of the ecological audio Hanning window;
the bird song complex index of the ecological audio Hanning window is the sum value of the time domain bird song index and the frequency domain bird song index;
the construction of the bird song response index of the audio interval comprises the following steps:
for each audio interval, counting the maximum value of the bird song complex indexes of all the ecological audio hanning windows in the audio interval, calculating the difference value of the maximum value and the bird song complex indexes of all the ecological audio hanning windows in the audio interval, obtaining the ratio of the bird song complex indexes of all the ecological audio hanning windows in the audio interval to the difference value, and taking the sum of the ratio of all the ecological audio hanning windows in the audio interval as the bird song response index of the audio interval.
2. The audio decomposition-based bio-diversity prediction method of claim 1, wherein each audio interval of the respective eco-audio signal comprises:
for each eco-audio signal, the eco-audio signal is used as the input of the voice activation detection VAD algorithm, and the audio interval in which the voice activity signal exists in the eco-audio signal is output as each audio interval of the eco-audio signal.
3. The method for predicting biodiversity based on audio decomposition of claim 2, wherein the framing and windowing the audio interval to obtain the eco-audio windowed signal comprises:
and carrying out frame division processing on each audio interval, respectively applying a hanning window function to each frame to obtain an ecological audio hanning window, and forming an ecological audio windowing signal of the audio interval from audio signals of all frames of the ecological audio hanning windows of the audio interval.
4. The method for predicting biodiversity based on audio decomposition of claim 1, wherein the obtaining the number of decomposition layers for each audio interval when the VMD algorithm decomposes comprises:
counting the maximum value of the bird song response index in the ecological audio signal where the audio interval is located, obtaining the ratio of the bird song response index to the maximum value of each audio interval, calculating the difference value between the maximum value of the preset decomposition layer number and the minimum value of the preset decomposition layer number, obtaining the rounded value of the product of the difference value and the ratio, and taking the sum of the rounded value and the minimum value as the decomposition layer number of the audio interval when the VMD algorithm is decomposed.
5. The audio decomposition-based biodiversity prediction method of claim 1, wherein the constructing a biological abundance index includes:
calculating the Pasteur coefficients between each modal component of each audio interval of the ecological audio signal after the decomposition of the VMD algorithm and all modal components of other audio intervals, obtaining the maximum value of the Pasteur coefficient of each modal component of each audio interval, and calculating the sum of the reciprocal values of the maximum values of all modal components of each audio interval; calculating the product of the acoustic diversity index of each audio interval and the sum value;
and recording the sum value of the products of all the audio intervals of the ecological audio signals as ecological audio sum values, and summing the acquired ecological audio sum values of all the ecological audio signals to obtain the biological abundance index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410179245.9A CN117727330B (en) | 2024-02-18 | 2024-02-18 | Biological diversity prediction method based on audio decomposition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410179245.9A CN117727330B (en) | 2024-02-18 | 2024-02-18 | Biological diversity prediction method based on audio decomposition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117727330A CN117727330A (en) | 2024-03-19 |
| CN117727330B true CN117727330B (en) | 2024-04-16 |
Family
ID=90203853
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410179245.9A Active CN117727330B (en) | 2024-02-18 | 2024-02-18 | Biological diversity prediction method based on audio decomposition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117727330B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR19990070595A (en) * | 1998-02-23 | 1999-09-15 | 이봉훈 | How to classify voice-voice segments in flattened spectra |
| EP2211335A1 (en) * | 2009-01-21 | 2010-07-28 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Apparatus, method and computer program for obtaining a parameter describing a variation of a signal characteristic of a signal |
| WO2015162645A1 (en) * | 2014-04-25 | 2015-10-29 | パナソニックＩｐマネジメント株式会社 | Audio processing apparatus, audio processing system, and audio processing method |
| CN109409308A (en) * | 2018-11-05 | 2019-03-01 | 中国科学院声学研究所 | A method of the birds species identification based on birdvocalization |
| CN116434774A (en) * | 2023-02-28 | 2023-07-14 | 招联消费金融有限公司 | Speech recognition method and related device |
| CN117037840A (en) * | 2023-08-09 | 2023-11-10 | 东风汽车集团股份有限公司 | Abnormal sound source identification method, device, equipment and readable storage medium |
-
2024
- 2024-02-18 CN CN202410179245.9A patent/CN117727330B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR19990070595A (en) * | 1998-02-23 | 1999-09-15 | 이봉훈 | How to classify voice-voice segments in flattened spectra |
| EP2211335A1 (en) * | 2009-01-21 | 2010-07-28 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Apparatus, method and computer program for obtaining a parameter describing a variation of a signal characteristic of a signal |
| WO2015162645A1 (en) * | 2014-04-25 | 2015-10-29 | パナソニックＩｐマネジメント株式会社 | Audio processing apparatus, audio processing system, and audio processing method |
| CN109409308A (en) * | 2018-11-05 | 2019-03-01 | 中国科学院声学研究所 | A method of the birds species identification based on birdvocalization |
| CN116434774A (en) * | 2023-02-28 | 2023-07-14 | 招联消费金融有限公司 | Speech recognition method and related device |
| CN117037840A (en) * | 2023-08-09 | 2023-11-10 | 东风汽车集团股份有限公司 | Abnormal sound source identification method, device, equipment and readable storage medium |
Non-Patent Citations (4)
| Title |
|---|
| PSO-DVMD-WT的变形信号去噪方法研究;陈竹安;熊鑫;李亦佳;;测绘科学;20200616(第08期);全文 * |
| 基于深度学习的城市森林公园鸟声识别方法研究;李清荣;中国优秀硕士学位论文全文数据库 基础科学辑;20240215(第02期);全文 * |
| 多人对话场景下的说话人分割聚类研究;朱唯鑫;中国优秀硕士学位论文全文数据库信息科技辑;20171115(第11期);全文 * |
| 洞庭湖鸟类资源分布及其栖息地质量评估;关蕾等;北京林业大学学报;20160731;第38卷(第07期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117727330A (en) | 2024-03-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jiang et al. | Whistle detection and classification for whales based on convolutional neural networks | |
| Priyadarshani et al. | Birdsong denoising using wavelets | |
| Wang et al. | Environmental sound classification with parallel temporal-spectral attention | |
| González-Hernández et al. | Marine mammal sound classification based on a parallel recognition model and octave analysis | |
| CN111261189B (en) | Vehicle sound signal feature extraction method | |
| CN101976564A (en) | Method for identifying insect voice | |
| Venter et al. | Automatic detection of African elephant (Loxodonta africana) infrasonic vocalisations from recordings | |
| CN109741759B (en) | Acoustic automatic detection method for specific bird species | |
| Liu et al. | EMG burst presence probability: a joint time–frequency representation of muscle activity and its application to onset detection | |
| Lostanlen et al. | Long-distance detection of bioacoustic events with per-channel energy normalization | |
| Hidayat et al. | A Modified MFCC for Improved Wavelet-Based Denoising on Robust Speech Recognition. | |
| CN114129163B (en) | Emotion analysis method and system for multi-view deep learning based on electroencephalogram signals | |
| CN114403897A (en) | Human body fatigue detection method and system based on electroencephalogram signals | |
| CN117727330B (en) | Biological diversity prediction method based on audio decomposition | |
| Jaafar et al. | MFCC based frog identification system in noisy environment | |
| Reddy et al. | Categorization of environmental sounds | |
| Zhang et al. | Robust acoustic event recognition using AVMD-PWVD time-frequency image | |
| CN111862991A (en) | Method and system for identifying baby crying | |
| CN115481689A (en) | System and method for simultaneously recognizing user gesture and identity based on ultrasonic waves | |
| CN115376540A (en) | Biological radar voice enhancement method and system based on variational modal decomposition | |
| Cruz et al. | An incremental algorithm based on multichannel non-negative matrix partial co-factorization for ambient denoising in auscultation | |
| Zhang et al. | Automatic bioacoustics noise reduction method based on a deep feature loss network | |
| Adavanne et al. | Convolutional recurrent neural networks for bird audio detection | |
| Barker et al. | Ultrasound-coupled semi-supervised nonnegative matrix factorisation for speech enhancement | |
| Zia et al. | Noise detection and elimination for improved acoustic detection of coronary artery disease |
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 |