A method and system for secure audio data transmission in complex environments
Through spectral graph analysis and Mel filter adjustment, the problem of inaccurate voice recognition in audio data transmission in complex environments is solved, which improves transmission efficiency and accuracy, and reduces the impact of interference.
Patent Information
- Application Number
- CN202411148550.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-21
AI Technical Summary
In complex environments such as rescue or field activities, audio data transmission is affected by other sound effects interference, resulting in inaccurate identification of the voice area and affecting the transmission effect.
By obtaining the audio data of voice calls, converting it into a spectral chart, analyzing high-energy areas, calculating the energy distribution difference coefficient and frequency information perception, adjusting the Mel filter to filter and compress audio data, and reducing interference effects.
It improves the accuracy and transmission efficiency of audio analysis, reduces the impact of complex environmental interference, and retains the vocal area information.
Smart Images

Figure CN118675541B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of speech analysis, and in particular to a method and system for securely transmitting audio data in a complex environment. Background Art
[0002] Complex environments are usually rescue environments or outdoor activity environments. During rescue activities, audio data can effectively and conveniently transmit information, making it easier for commanders in the rear to command and dispatch rescuers on the front line, so that rescue activities can be carried out faster and better. Effective analysis of audio data in complex environments is an important operation to ensure stable communication during rescue.
[0003] In the related technology, speech analysis is achieved by directly intercepting the human voice frequency and compressing it. In this way, due to the interference of other sound effects in the complex environment, such as thunder, wind and other complex environmental sounds, the frequency is reused and the overall complex environment has a greater impact, resulting in inaccurate recognition of the human voice area, which in turn makes the audio analysis effect poor, affecting the subsequent transmission process. Summary of the invention
[0004] In order to solve the technical problem in the related art that due to the interference of other sound effects in complex environments, such as thunder, wind and other complex environments, frequency reuse is caused, the overall complex environment has a greater impact, resulting in inaccurate recognition of human voice areas, and thus making the audio transmission effect poor, affecting the subsequent transmission process, the present invention provides a method and system for secure transmission of audio data in a complex environment, and the technical solutions adopted are as follows:
[0005] The present invention proposes a method for securely transmitting audio data in a complex environment, the method comprising:
[0006] Acquire audio data of different segments of voice calls using an intercom device, and convert each segment of the audio data into a spectrogram; wherein the audio data includes field audio data of field personnel and central audio data of the command center;
[0007] Determine the high energy area according to the color characteristics of the image in each of the spectrograms; select a target frequency, and determine the distribution index of the high energy area under the target frequency according to the position distribution of the high energy area in each of the spectrograms and the high energy characteristics of the target frequency;
[0008] Determine the energy distribution difference coefficient between the external audio data and the central audio data according to the difference in distribution indexes of the high energy area of the spectrograms from different sources at the target frequency; determine the adjustment parameters of the frequency information perception corresponding to the target frequency according to the distribution index and energy distribution difference coefficient of the two other frequencies closest to the target frequency;
[0009] The frequency compression index of the target frequency is determined according to the adjustment parameters and the network bandwidth, the Mel filter is adjusted according to the frequency compression index of all frequencies to obtain the target filter, the spectrogram is filtered according to the target filter to obtain the Mel spectrogram, the filtered audio corresponding to the Mel spectrogram is determined, and the filtered audio is compressed and transmitted.
[0010] Furthermore, the determining of the high energy area according to the color feature of the image in each of the spectrograms includes:
[0011] Determine a grayscale image of the spectrogram, perform maximum between-class variance analysis on the grayscale values of all pixels in the grayscale image, and determine an energy classification threshold;
[0012] Pixels in the grayscale image whose grayscale values are greater than the energy classification threshold are regarded as high-energy pixels, and the area formed by the high-energy pixels is regarded as a high-energy area.
[0013] Furthermore, the method for acquiring the distribution index of the high energy area at the target frequency includes:
[0014] The morphological center point of each high-energy region is used as the position point of the corresponding high-energy region, and any two position points are connected to obtain the regional distance. The negative number of the variance of all the regional distances is normalized as the high-energy distribution concentration index;
[0015] Normalizing the mean grayscale values of the pixels corresponding to the target frequency at different times to obtain a high energy feature index;
[0016] The product of the high energy distribution concentration index and the high energy characteristic index is calculated as the distribution index of the high energy area at the target frequency.
[0017] Furthermore, the method for obtaining the energy distribution difference coefficient of the outfield audio data and the center audio data includes:
[0018] Determine that the variance of the distribution index corresponding to all the outfield audio data is the outfield distribution variance, and the variance of the distribution index corresponding to all the center audio data is the center distribution variance;
[0019] The absolute value of the difference between the external field distribution variance and the central distribution variance is calculated, and the absolute value of the difference is normalized as the energy distribution difference coefficient between the external field audio data and the central audio data.
[0020] Furthermore, the method for obtaining the adjustment parameters of the frequency information perception situation corresponding to the target frequency includes:
[0021] The two frequencies closest to the target frequency are taken as adjacent frequencies, and the inverse of the energy distribution difference coefficient between the target frequency and the adjacent frequencies is determined as the distribution influencing factor of the corresponding frequency;
[0022] Determine the product of the distribution impact factor and the distribution index at the same frequency as the initial adjustment index at the corresponding frequency;
[0023] The sum of the target frequency and the initial adjustment indexes of two adjacent frequencies is normalized to obtain an adjustment parameter of the frequency information perception situation corresponding to the target frequency.
[0024] Furthermore, the method for obtaining the frequency compression index of the target frequency includes:
[0025] Normalize the maximum and minimum values of the network bandwidth to obtain the bandwidth impact coefficient;
[0026] The frequency compression index is determined according to the bandwidth influence coefficient and the adjustment parameter, wherein both the bandwidth influence coefficient and the adjustment parameter are positively correlated with the frequency compression index.
[0027] Further, determining the frequency compression index according to the bandwidth influence coefficient and the adjustment parameter includes:
[0028] A normalized value of the product of the bandwidth influence coefficient and the adjustment parameter is calculated as a frequency compression index of the target frequency.
[0029] Furthermore, the Mel filter is adjusted according to the frequency compression index of all frequencies to obtain a target filter, including:
[0030] Determine the product of the frequency compression index of each frequency and the Mel scale at the corresponding frequency in the Mel curve as the adjustment scale;
[0031] Construct the target filter based on all adjustment scales.
[0032] Further, determining the filtered audio corresponding to the Mel-level spectrogram includes:
[0033] The Mel-level spectrogram is subjected to inverse short-time Fourier transform processing to be converted into filtered audio.
[0034] On the other hand, the present invention also provides a system for securely transmitting audio data in a complex environment, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, it implements the steps of a method for securely transmitting audio data in a complex environment as described in any one of the preceding items.
[0035] The present invention has the following beneficial effects:
[0036] The present invention analyzes the energy distribution of the audio segment through a spectrogram, and then determines the degree of influence of the complex environment and the energy distribution of different frequencies in each segment, and analyzes the distribution index of the high energy area at different frequencies, which can characterize the importance of the corresponding frequency in the time period corresponding to the spectrogram; analyzes the entire voice call in combination with different spectrograms, analyzes the difference in the distribution index of the high energy area, determines the energy distribution difference coefficient of the outfield audio data and the center audio data, and then determines the adjustment parameter in combination with the energy distribution difference coefficient and the distribution index. The adjustment parameter not only characterizes the importance of each frequency in the voice call process, but also characterizes the situation affected by the complex environment interference. Therefore, the Mel filter is adjusted in combination with the network bandwidth to obtain the target filter, and then the filtered audio corresponding to the Mel spectrogram is obtained by filtering, and the filtered audio is compressed and transmitted. In the embodiment of the present invention, an adaptive adjustment parameter is determined for each frequency, and then an adaptive Mel filter analysis is performed on different frequencies. Compared with directly performing overall audio filtering compression, the present application can retain a more complete and effective human voice area, reduce the influence of complex environmental interference, improve the audio analysis effect, and enhance subsequent transmission efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0038] Figure 1 A flow chart of a method for securely transmitting audio data in a complex environment provided by an embodiment of the present invention;
[0039] Figure 2 A schematic diagram of a Mel curve provided by an embodiment of the present invention;
[0040] Figure 3 A structural diagram of a system for securely transmitting audio data in a complex environment provided by one embodiment of the present invention. DETAILED DESCRIPTION
[0041] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a method and system for secure transmission of audio data in a complex environment proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0042] 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.
[0043] The following is a detailed description of a method for securely transmitting audio data in a complex environment provided by the present invention in conjunction with the accompanying drawings.
[0044] See also Figure 1 , which shows a flow chart of a method for securely transmitting audio data in a complex environment provided by an embodiment of the present invention, the method comprising:
[0045] S101: Acquire audio data of different segments of a voice call using an intercom device, and convert each segment of the audio data into a spectrogram; wherein the audio data includes field audio data of field personnel and central audio data of a command center.
[0046] One implementation scenario of the present invention is an intercom scenario, such as a complex intercom scenario such as a front-line rescue scenario or an outdoor adventure. In this scenario, the two parties intercoming are usually field personnel and a corresponding command center. It should be noted that since field personnel are in a complex environment and need to move, their corresponding signal quality is subject to interference from a variety of high-frequency noises, which will affect the judgment of the human voice. Therefore, certain adjustments are needed to enhance the frequency that the human ear can perceive and reduce the interference frequency.
[0047] It should be noted that audio data is generally obtained through human ears during information exchange. Humans' perception of frequency is not linear. The human ear has different perception capabilities for sounds of different frequencies, and is more sensitive to low-frequency signals than high-frequency signals. For example, people can easily find the difference between 500Hz and 1000Hz, but it is difficult to find the difference between 7500Hz and 8000Hz.
[0048] Therefore, in an embodiment of the present invention, when transmitting audio data, the human ear's perception of the acquired information of different frequencies and the interference to the acquired data in each frequency range can be taken into consideration, and the acquired data can be adjusted to weaken the frequencies that are not strongly perceived by the human ear, thereby retaining the information content to a greater extent and reducing the amount of audio data.
[0049] The audio data is the sound frequency data obtained by the intercom device, which can be divided into different segments in a fixed order, such as 2 seconds per segment, for the convenience of analysis. It should be noted that, since it is an intercom scene, the audio data includes the field audio data of the field personnel and the central audio data of the command center. In the embodiment of the present invention, the field audio data and the central audio data belong to audio data from different sources.
[0050] Among them, each audio data segment is converted into a spectrogram. The specific method of converting the audio segment into a spectrogram is a method well known to technical personnel in this field, that is, the acquired audio data is converted into a spectrogram by short-time Fourier transform.
[0051] S102: Determine the high energy area according to the color characteristics of the image in each spectrogram; select a target frequency, and determine the distribution index of the high energy area under the target frequency according to the position distribution of the high energy area in each spectrogram and the high energy characteristics of the target frequency.
[0052] It can be understood that in the spectrogram, there are features of high energy and low energy, that is, the more the color in the spectrogram tends to be blue-green, the low energy area, and the yellow-red is the high energy area. Therefore, analysis can be carried out according to the color features of the image in the spectrogram to determine the high energy area. It can be understood that grayscale can effectively characterize color features. In order to facilitate specific analysis, the spectrogram can be grayscaled to determine the grayscale image of the spectrogram, and the maximum between-class variance analysis is performed on the grayscale values of all pixels in the grayscale image to determine the energy classification threshold; the pixels in the grayscale image whose grayscale values are greater than the energy classification threshold are regarded as high-energy pixels, and the area composed of high-energy pixels is regarded as a high-energy area.
[0053] Among them, the maximum inter-class variance method is an algorithm well known to technical personnel in this field. The corresponding energy classification threshold is determined by the maximum inter-class variance. The energy classification threshold is the corresponding grayscale value threshold. The pixels in the grayscale image whose grayscale values are greater than the energy classification threshold are regarded as high-energy pixels, and the high-energy area is the area composed of high-energy pixels.
[0054] It should be noted that, for the convenience of analysis, the frequencies in the embodiments of the present invention are distinguished by positive integer frequencies, such as 200 Hz, 201 Hz, etc. Therefore, any target frequency, that is, any positive integer frequency is selected as the target frequency, and the corresponding adjacent frequencies in the subsequent description are the frequencies corresponding to the adjacent numbers. Taking 200 Hz as a specific example, its adjacent frequencies are 199 Hz and 201 Hz.
[0055] It should be noted that, since the propagation of sound can easily show a variety of continuous and similar regional features on the spectrogram, the high-energy area is usually a regularly arranged distribution of similar areas in a spectrogram containing voice information. The embodiment of the present invention specifically analyzes the distribution characteristics of the high-energy area.
[0056] Furthermore, in some embodiments of the present invention, a method for obtaining a distribution index of a high-energy region at a target frequency includes: taking the morphological center point of each high-energy region as the position point of the corresponding high-energy region, connecting any two position points to obtain the region distance, and normalizing the negative of the variance of all region distances to obtain a high-energy distribution concentration index; normalizing the grayscale value mean of the pixel points corresponding to the target frequency at different times to obtain a high-energy feature index; and calculating the product of the high-energy distribution concentration index and the high-energy feature index as the distribution index of the high-energy region at the target frequency.
[0057] Among them, the morphological center can be used to characterize the regional position corresponding to the high-energy area. Of course, in other embodiments of the present invention, different methods such as the center of gravity or the center point can also be used. The reliability of using the morphological center for position analysis is stronger. In the embodiment of the present invention, the morphological center point can be used as the position point corresponding to the high-energy area, and then, all high-energy areas are traversed, and the Euclidean distance between the position points of any two high-energy areas is used as the regional distance, and the variance of all regional distances is used as the distribution discreteness of the high-energy area. That is, the larger the variance, the more discrete the distribution of the high-energy area. Therefore, the negative number of the variance is normalized as a high-energy distribution concentration index. That is, the larger the value of the high-energy distribution concentration index, the more concentrated the distribution of the high-energy area.
[0058] It should be noted that since human speech is usually in a relatively concentrated area, and the sudden appearance of noise will cause a difference effect due to its high frequency, the human speech can be specifically analyzed by calculating the degree of concentration.
[0059] It can be understood that a spectrogram is an image with time as the horizontal axis and frequency as the vertical axis, that is, the pixel points corresponding to the target frequency at different times are the pixel points on a straight line with the same vertical axis. In the embodiment of the present invention, the mean grayscale value of all pixel points on the straight line is calculated. The larger the value, the higher the energy of the corresponding target frequency, and a high-energy feature index is obtained.
[0060] Therefore, in an embodiment of the present invention, the product of the high-energy distribution concentration index and the high-energy characteristic index is calculated as the distribution index of the high-energy area at the target frequency. The larger the value of the distribution index is, the higher the energy corresponding to the target frequency is, so that the energy characteristics of each frequency can be analyzed, which is convenient for subsequent energy analysis.
[0061] S103: Determine the energy distribution difference coefficient between the outfield audio data and the center audio data based on the difference in distribution indicators of the high energy area of the spectrograms from different sources at the target frequency; determine the adjustment parameters of the frequency information perception corresponding to the target frequency based on the distribution indicators and energy distribution difference coefficients of the two other frequencies closest to the target frequency.
[0062] Among them, since each spectrogram has a corresponding distribution feature, and since each spectrogram is a spectrogram of a corresponding time period, its distribution index may be affected by instantaneous noise. Therefore, the embodiment of the present invention combines multiple spectrograms from different sources to achieve specific analysis.
[0063] In the embodiment of the present invention, the environment surrounding the audio data acquired by the command center is relatively stable so the interference to the acquired audio data is relatively small, while the front-line rescue personnel need to face various harsh environments and the interference to the audio data acquired may be relatively serious. When judging the frequency distribution of the acquired audio data affected by complex environmental interference, the difference in frequency energy distribution of the speech in the two cases can be analyzed to quantify the frequency distribution characteristics affected by complex environmental interference.
[0064] Therefore, by distinguishing the field audio data of field personnel and the central audio data of the command center, the impact of interference in complex environments can be accurately analyzed.
[0065] Furthermore, in some embodiments of the present invention, a method for obtaining the energy distribution difference coefficient of outside field audio data and center audio data includes: determining the variance of the distribution index corresponding to all outside field audio data as the outside field distribution variance, and the variance of the distribution index corresponding to all center audio data as the center distribution variance; calculating the absolute value of the difference between the outside field distribution variance and the center distribution variance, and normalizing the absolute value of the difference as the energy distribution difference coefficient of the outside field audio data and the center audio data.
[0066] The distribution characteristics are characterized by variance, and the absolute value of the difference is calculated for normalization to obtain the energy distribution difference coefficient. Of course, in other embodiments of the present invention, the analysis of the energy distribution difference coefficient can be realized in combination with other features, such as mean, median, etc., which is not limited to this.
[0067] The larger the energy distribution difference coefficient is, the greater the difference between the external audio data and the central audio data is. Since the central audio data can be determined as audio data in a stable environment and is less affected by complex environmental interference, it further indicates that the external audio data is more interfered.
[0068] Thus, the adjustment parameters can be determined based on the energy distribution difference coefficient. Further, in some embodiments of the present invention, the method for obtaining the adjustment parameters of the frequency information perception situation corresponding to the target frequency includes: taking the two frequencies closest to the target frequency as adjacent frequencies, determining the inverse of the energy distribution difference coefficient between the target frequency and the adjacent frequencies as the distribution influence factor of the corresponding frequency; determining the product of the distribution influence factor and the distribution index at the same frequency as the initial adjustment index at the corresponding frequency; normalizing the sum of the target frequency and the initial adjustment index of the two adjacent frequencies to obtain the adjustment parameters of the frequency information perception situation corresponding to the target frequency.
[0069] In combination with the adjacent frequencies mentioned above, which are the frequencies corresponding to the adjacent numbers, taking 200 Hz as a specific example, its adjacent frequencies are 199 Hz and 201 Hz. Thus, the data analysis within the corresponding frequency range is determined by the adjacent frequencies and the target frequency itself. The adjacent frequencies in other embodiments of the present invention can also represent other numbers of frequencies that are closest to each other, such as 5, 10, etc., which are not limited to this. The present invention uses 2 for specific analysis.
[0070] Among them, the inverse of the energy distribution difference coefficient between the target frequency and the adjacent frequency is determined as the distribution influence factor of the corresponding frequency, that is, the inverse of the energy distribution difference coefficient of the target frequency is used as the distribution influence factor of the target frequency, and the inverse of the energy distribution difference coefficient of any adjacent frequency is used as the distribution influence factor of the adjacent frequency. Thus, the distribution influence factor of each frequency is obtained. Since the larger the energy distribution difference coefficient is, the greater the possibility of being affected by complex environmental interference, the lower the importance of the frequency affected by the corresponding interference when making adjustments, and the more lossy compression processing can be performed, the larger the value of the distribution influence factor is, and the smaller the compression degree in the subsequent compression process is.
[0071] The larger the distribution index is, the higher the corresponding energy is, and the more important the information it contains is. Therefore, in an embodiment of the present invention, the product of the distribution influence factor and the distribution index at the same frequency is used as the initial adjustment index at the corresponding frequency, which represents the adjustment of the frequency. Since the frequency will cause mutual interference between neighbors, in an embodiment of the present invention, the sum of the target frequency and the initial adjustment indexes of the two adjacent frequencies is normalized to obtain the adjustment parameter of the frequency information perception corresponding to the target frequency, that is, the adjustment parameter taking into account the interference effect is represented by the sum value. The larger the value of the adjustment parameter is, the more important the target frequency is.
[0072] S104: Determine a frequency compression index of a target frequency according to adjustment parameters and network bandwidth, adjust the Mel filter according to the frequency compression index of all frequencies to obtain a target filter, filter the spectrogram according to the target filter to obtain a Mel spectrogram, determine the filtered audio corresponding to the Mel spectrogram, and compress and transmit the filtered audio.
[0073] It should be noted that when performing audio transmission, not only the distribution of the overall audio frequency should be considered, but also the network bandwidth needs to be analyzed, thereby determining the frequency compression index of the target frequency according to the adjustment parameter and the network bandwidth. Further, in some embodiments of the present invention, the method for obtaining the frequency compression index of the target frequency includes: normalizing the maximum and minimum values of the network bandwidth to obtain a bandwidth influence coefficient; determining the frequency compression index according to the bandwidth influence coefficient and the adjustment parameter, wherein the bandwidth influence coefficient and the adjustment parameter are both positively correlated with the frequency compression index.
[0074] Among them, the larger the bandwidth, the better the transmission effect during data transmission, and the more complete the audio data can be transmitted. The smaller the bandwidth, the more lossy compression of audio data is required for efficient transmission to adapt to the data transmission efficiency. Therefore, in the embodiment of the present invention, the maximum and minimum values of the network bandwidth are normalized to obtain the bandwidth influence coefficient, which characterizes the degree of influence of the bandwidth on the compression effect.
[0075] Then, in the embodiment of the present invention, the normalized value of the product of the bandwidth influence coefficient and the adjustment parameter can be calculated as the frequency compression index of the target frequency. The frequency compression index characterizes the compression degree when the data conversion compression is performed on the corresponding target frequency. The larger the value of the frequency compression index, the more important the corresponding frequency is, and the less interference from the complex environment. At this time, a smaller compression rate can be used for data compression to ensure the effective transmission of important information. Therefore, the Mel filter can be adjusted based on the frequency compression index.
[0076] Furthermore, in some embodiments of the present invention, the Mel filter is adjusted according to the frequency compression index of all frequencies to obtain a target filter, including: determining the product of the frequency compression index of each frequency and the Mel scale at the corresponding frequency in the Mel curve as an adjustment scale; and constructing the target filter according to all adjustment scales.
[0077] The Mel curve is a spectrum adjustment curve well known to those skilled in the art, such as Figure 2 As shown, Figure 2 A schematic diagram of a Mel curve provided by an embodiment of the present invention. Adaptive adjustment and compression of different frequencies are performed through frequency compression indicators, thereby being more suitable for complex intercom scenarios, thereby being able to better compress audio data and improve the subsequent transmission efficiency of audio data.
[0078] Furthermore, in some embodiments of the present invention, determining the filtered audio corresponding to the Mel-gram spectrogram includes: performing an inverse short-time Fourier transform process on the Mel-gram spectrogram to convert it into the filtered audio.
[0079] The inverse short-time Fourier transform of the Mel-level spectrogram is a technology well known to those skilled in the art. The corresponding spectrogram can be converted into audio data through the inverse short-time Fourier transform, thereby obtaining the filtered audio in the embodiment of the present invention.
[0080] It should be noted that the acquired audio data needs to be processed when the audio data is transmitted securely. In order to ensure the safe and effective transmission of the audio data, the audio data can be compressed. The acquired audio data is lossily compressed to reduce the amount of acquired data. Transmitting the compressed data can ensure stable transmission in complex environments. However, the above-mentioned lossy compression will cause information loss of the acquired audio data. When the degree of compression is too large, important information will be seriously lost.
[0081] Therefore, in the embodiment of the present invention, a specific analysis is made on the interference problem in a complex environment, so that the human voice frequency is retained as much as possible during the compression process, and other influencing frequencies in the complex environment are reduced, thereby reducing the amount of data while retaining the stability of voice transmission, and improving the subsequent compression effect. Therefore, in the embodiment of the present invention, the compression process of the filtered audio can be lossy compression, thereby improving the transmission efficiency while ensuring the stability of the audio data.
[0082] The present invention analyzes the energy distribution of the audio segment through a spectrogram, and then determines the degree of influence of the complex environment and the energy distribution of different frequencies in each segment, and analyzes the distribution index of the high energy area at different frequencies, which can characterize the importance of the corresponding frequency in the time period corresponding to the spectrogram; analyzes the entire voice call in combination with different spectrograms, analyzes the difference in the distribution index of the high energy area, determines the energy distribution difference coefficient of the outfield audio data and the center audio data, and then combines the energy distribution difference coefficient and the distribution index to determine the adjustment parameter, then the adjustment parameter not only characterizes the importance of each frequency in the voice call process, but also characterizes the situation affected by the complex environment interference, thus, the Mel filter is adjusted in combination with the network bandwidth to obtain the target filter, and then the filtered audio corresponding to the Mel spectrogram is obtained by filtering, and the filtered audio is compressed and transmitted. In the embodiment of the present invention, an adaptive adjustment parameter is determined for each frequency, and then an adaptive Mel filter analysis is performed on different frequencies. Compared with directly performing overall audio filtering compression, the present application can retain a more complete and effective human voice area, reduce the influence of complex environmental interference, and improve the effect of audio analysis and subsequent transmission.
[0083] This embodiment also provides a system for securely transmitting audio data in a complex environment. Figure 3 , Figure 3 A structural diagram of a system for securely transmitting audio data in a complex environment provided for an embodiment of the present invention, wherein the system 600 includes a memory 601, a processor 602, and a computer program 603 stored in the memory 601 and executable on the processor 602. When the processor 602 executes the computer program 603, the steps of a method for securely transmitting audio data in a complex environment as described above are implemented.
[0084] This embodiment also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is executed on a computer, the computer executes the above-mentioned related method steps to implement a method for securely transmitting audio data in a complex environment provided in the above embodiment.
[0085] This embodiment also provides a computer program product. When the computer program product runs on a computer, the computer executes the above-mentioned related steps to implement a method for securely transmitting audio data in a complex environment provided by the above embodiment.
[0086] Among them, the system, computer-readable storage medium or computer program product provided in this embodiment is used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here.
[0087] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0088] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
Claims
1. A method for secure transmission of audio data in complex environments. It is characterized in that The method comprises: Acquire audio data of different segments of voice calls using an intercom device, and convert each segment of the audio data into a spectrogram; wherein the audio data includes field audio data of field personnel and central audio data of the command center; Determine the high energy area according to the color characteristics of the image in each of the spectrograms; select a target frequency, and determine the distribution index of the high energy area under the target frequency according to the position distribution of the high energy area in each of the spectrograms and the high energy characteristics of the target frequency; Determine the energy distribution difference coefficient between the external audio data and the central audio data according to the difference in distribution indexes of the high energy area of the spectrograms from different sources at the target frequency; determine the adjustment parameters of the frequency information perception corresponding to the target frequency according to the distribution index and energy distribution difference coefficient of the two other frequencies closest to the target frequency; Determine a frequency compression index of a target frequency according to the adjustment parameter and the network bandwidth, adjust the Mel filter according to the frequency compression index of all frequencies to obtain a target filter, filter the spectrogram according to the target filter to obtain a Mel spectrogram, determine a filtered audio corresponding to the Mel spectrogram, and compress and transmit the filtered audio; The method for obtaining the energy distribution difference coefficient of the outfield audio data and the center audio data includes: Determine that the variance of the distribution index corresponding to all the outfield audio data is the outfield distribution variance, and the variance of the distribution index corresponding to all the center audio data is the center distribution variance; Calculating the absolute value of the difference between the external field distribution variance and the central distribution variance, and normalizing the absolute value of the difference as the energy distribution difference coefficient between the external field audio data and the central audio data; The method for obtaining adjustment parameters of the frequency information perception situation corresponding to the target frequency includes: The two frequencies closest to the target frequency are taken as adjacent frequencies, and the inverse of the energy distribution difference coefficient between the target frequency and the adjacent frequencies is determined as the distribution influencing factor of the corresponding frequency; Determine the product of the distribution impact factor and the distribution index at the same frequency as the initial adjustment index at the corresponding frequency; Normalizing the sum of the target frequency and the initial adjustment indexes of two adjacent frequencies to obtain an adjustment parameter of the frequency information perception situation corresponding to the target frequency; The method for obtaining the frequency compression index of the target frequency includes: Normalize the maximum and minimum values of the network bandwidth to obtain the bandwidth impact coefficient; Determining the frequency compression index according to the bandwidth influence coefficient and the adjustment parameter, wherein both the bandwidth influence coefficient and the adjustment parameter are positively correlated with the frequency compression index; Determining the frequency compression index according to the bandwidth influence coefficient and the adjustment parameter includes: A normalized value of the product of the bandwidth influence coefficient and the adjustment parameter is calculated as a frequency compression index of the target frequency.
2. A method for securely transmitting audio data in a complex environment as claimed in claim 1, It is characterized in that Determining the high energy area according to the color features of the image in each of the spectrograms comprises: Determine a grayscale image of the spectrogram, perform maximum inter-class variance analysis on the grayscale values of all pixels in the grayscale image, and determine an energy classification threshold; Pixels in the grayscale image whose grayscale values are greater than the energy classification threshold are regarded as high-energy pixels, and the area formed by the high-energy pixels is regarded as a high-energy area.
3. A method for securely transmitting audio data in a complex environment as claimed in claim 2, It is characterized in that The method for obtaining the distribution index of the high energy area at the target frequency includes: The morphological center point of each high-energy region is used as the position point of the corresponding high-energy region, and any two position points are connected to obtain the regional distance. The negative number of the variance of all the regional distances is normalized as the high-energy distribution concentration index; Normalizing the mean grayscale values of the pixels corresponding to the target frequency at different times to obtain a high energy feature index; The product of the high energy distribution concentration index and the high energy characteristic index is calculated as the distribution index of the high energy area at the target frequency.
4. A method for securely transmitting audio data in a complex environment as claimed in claim 1, It is characterized in that The Mel filter is adjusted according to the frequency compression index of all frequencies to obtain a target filter, including: Determine the product of the frequency compression index of each frequency and the Mel scale at the corresponding frequency in the Mel curve as the adjustment scale; Construct the target filter based on all adjustment scales.
5. A method for securely transmitting audio data in a complex environment as claimed in claim 1, It is characterized in that Determining the filtered audio corresponding to the Mel-language spectrogram includes: The Mel-level spectrogram is subjected to inverse short-time Fourier transform processing to be converted into filtered audio.
6. A system for securely transmitting audio data in a complex environment, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.
