Monitoring system voice howling suppression method

Abstract

The embodiment of the invention provides a voice howling suppression method for a monitoring system. The voice howling suppression method comprises the following steps: S1, processing a collected voice signal according to a preset rule to obtain a plurality of sub-bands; s2, respectively calculating frequency domain data of each sub-band to obtain frequency domain flatness and average energy ratio; s3, performing howling judgment according to a first preset threshold value and a second preset threshold value based on the frequency domain flatness and the average energy ratio of each sub-band; s4, when the squeal sub-band exists in the step S3, judging the total energy of the full frequency according to a third preset threshold value; and S5, when the full-frequency total energy is higher than a threshold value, performing suppression processing on the howling sub-band to obtain a sound signal Y (t). According to the monitoring system voice howling suppression method provided by the invention, the judgment is performed by comparing the characteristic values of the voice signals, and the sub-band which is judged to be howling is suppressed, so that effective and high-quality howling suppression is realized.

Description

Technical Field

[0001] The present invention relates to the field of monitoring technology, and in particular to a method for suppressing voice howling in a monitoring system. Background Art

[0002] In the monitoring system, the speaker's voice is collected by the microphone in the monitoring room, the signal is processed by the audio processor, and then through the power amplifier, the speaker's voice is finally played out by the monitor's speakers. However, the sound played by the speakers will be transmitted back to the microphone through air vibrations and reflections from the walls of the sound amplification site. When the echo path gain of one or several frequency bands is greater than 1, howling will occur.

[0003] In the existing technical method, the distance between the device with microphone and the device with sound is often set far away.

[0004] However, in some application scenarios, the microphone and the speaker cannot be set at a very far distance due to objective reasons. In this case, the howling problem cannot be solved by setting the distance very far. Summary of the invention

[0005] In view of the above problems, embodiments of the present invention are proposed to provide a method for suppressing voice howling in a monitoring system that overcomes the above problems or at least partially solves the above problems.

[0006] In order to solve the above problem, an embodiment of the present invention discloses a method for suppressing voice howling in a monitoring system, comprising the following steps:

[0007] S1. Processing the collected speech signal according to preset rules to obtain several sub-bands;

[0008] S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio;

[0009] S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold;

[0010] S4, when there is a howling sub-band in S3, judging the total energy of the whole frequency according to the third preset threshold;

[0011] S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

[0012] Furthermore, before step S1, the method further includes the following steps:

[0013] S0. Collect the voice signal of the monitoring system.

[0014] Furthermore, the step S1 includes the following steps:

[0015] S11, performing short-time Fourier transform on the speech signal to obtain a spectrum;

[0016] S12. Divide the spectrum into several sub-bands.

[0017] Furthermore, the step S11 includes: performing a short-time Fourier transform on the speech signal by the following transformation formula to obtain a spectrum;

[0018] Transformation formula:

[0019]

[0020] Among them, ω(n) is the window function, which is used to intercept a signal of the window width of x(n) at time n.

[0021] Furthermore, the window function of the short-time Fourier transform is a Hamming window.

[0022] Furthermore, the step of suppressing in step S5 includes:

[0023] S51, testing the local monitoring system and the on-site environment to obtain the M frequency points with the greatest howling impact and the optimal energy attenuation value corresponding to each frequency point;

[0024] S52, performing discretized howling suppression processing on the howling sub-bands determined as the howling sub-bands in the M frequency points to obtain a processed far-end discrete digital signal Y(n);

[0025] S53, performing notch and conversion processing on the remote discrete digital signal Y(n) to obtain an analog electrical signal R(t);

[0026] S54, the analog electrical signal R(t) is amplified to obtain a sound signal Y(t).

[0027] Furthermore, the step S54 further includes the following steps:

[0028] S55, the echo signal E(t) generated by the sound signal Y(t) after being affected by the environment is mixed with the near-end voice signal and collected, and then the process jumps to step S51 to continue.

[0029] Furthermore, the howling judgment in step S3 is specifically the following steps:

[0030] S31, setting the threshold of frequency domain flatness to P, and setting the threshold of average energy ratio to N;

[0031] S32: If the frequency domain flatness of a certain sub-band is lower than P and the average energy ratio is higher than N, it can be determined as a howling sub-band.

[0032] Furthermore, before making a full-frequency total energy determination, the following steps are also included:

[0033] The total energy of the whole frequency spectrum is calculated by the following formula;

[0034]

[0035] Where: X(n) is the complex amplitude of the nth frequency component;

[0036] N is the total number of frequency components.

[0037] The present invention also constructs a monitoring system, the monitoring system is used to implement the method according to claims 1-8, the monitoring system comprises: a sound transceiver module, an audio processor, a camera and a power amplifier;

[0038] The sound transceiver module includes a microphone and a speaker;

[0039] The two sound transceiver modules are respectively arranged at the camera and at a preset position away from the camera, and are used to collect voice signals and send sound signals;

[0040] The audio processor is electrically connected to the microphone and is used to process the collected voice signal;

[0041] The power amplifier is connected to the audio processor to further process the voice signal;

[0042] The power amplifier is electrically connected to the sound box and is used to play the secondary processed voice signal.

[0043] The embodiments of the present invention include the following advantages:

[0044] Through the logical relationship proposed in the present invention, the howling sub-band is determined more accurately, and the potential howling sub-band is accurately determined, so as to suppress it. On the one hand, the amount of calculation for processing all sub-band data is reduced, and on the other hand, the effect of suppressing howling is greatly improved. The howling sub-band can be processed by various iterative methods, so as to output a better quality and more realistic speech signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flowchart of a method for suppressing voice howling in a monitoring system according to an embodiment of the present invention;

[0046] Figure 2 This is a diagram showing the causes of howling in the present invention;

[0047] Figure 3 The present invention is a structural block diagram of a monitoring system voice howling suppression method embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] Example 1

[0050] Reference Figure 1 and Figure 3 , shows a flowchart of an embodiment of a method for suppressing voice howling in a monitoring system of the present invention, which may specifically include the following steps:

[0051] S1. Processing the collected speech signal according to preset rules to obtain several sub-bands;

[0052] S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio;

[0053] S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold;

[0054] S4, when there is a howling subband in S3, judging the total energy of the whole frequency according to the third preset threshold;

[0055] S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

[0056] Before step S1, the method further includes the following steps: S0, collecting voice signals of the monitoring system. First, the voice signals of the monitoring system are collected, and the collected voice signals are processed through the above steps.

[0057] It should be noted that according to Figure 2 , which shows the reason why the howling of the present invention is generated. It is because the sound of the microphone and the sound transmitted to the speaker through the microphone are transmitted back to the microphone, and a feedback loop is formed. This is repeated. Once the gain in the feedback loop reaches or exceeds the critical value, the howling will be triggered within tens to hundreds of milliseconds, thereby generating howling.

[0058] The core of the present invention lies in the improvement of the method for finding the howling sub-band. First, the collected speech signal is processed according to preset rules to obtain several sub-bands, and then the frequency domain flatness and average energy ratio of each sub-band are calculated based on a formula, and then compared with the preset threshold, that is, a judgment is made. When a certain relationship is satisfied, it is determined that there is a howling sub-band, and then a full-frequency total energy judgment is made. When the full-frequency total energy is higher than the threshold, the howling sub-band is suppressed, and after suppression, the audio is emitted through the speaker, thereby completing the howling suppression.

[0059] Example 2

[0060] Reference Figure 1 and Figure 3 , shows a flowchart of an embodiment of a method for suppressing voice howling in a monitoring system of the present invention, which may specifically include the following steps:

[0061] S1. Processing the collected speech signal according to preset rules to obtain several sub-bands;

[0062] S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio;

[0063] S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold;

[0064] S4, when there is a howling subband in S3, judging the total energy of the whole frequency according to the third preset threshold;

[0065] S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

[0066] Before step S1, the method further includes the following steps: S0, collecting voice signals of the monitoring system. First, the voice signals of the monitoring system are collected, and the collected voice signals are processed through the above steps.

[0067] Furthermore, the step S1 includes the following steps:

[0068] S11, performing short-time Fourier transform on the speech signal to obtain a spectrum;

[0069] S12. Divide the spectrum into several sub-bands.

[0070] Furthermore, the step S11 includes: performing a short-time Fourier transform on the speech signal by the following transformation formula to obtain a spectrum;

[0071] Transformation formula:

[0072]

[0073] Among them, ω(n) is the window function, which is used to intercept a signal of the window width of x(n) at time n.

[0074] The collected speech signal is a time domain signal. The time domain signal is converted into a frequency domain signal through the short-time Fourier transform formula. The obtained frequency domain signal can be divided into several sub-bands according to the frequency.

[0075] As a specific implementation method, the collected speech signal R(n) is a time domain signal. First, it is framed, generally a few milliseconds to tens of milliseconds per frame, and the overlap rate is set, usually 25% or 50%, and a time window function ω(n) is selected. The time window function is generally a rectangular window and a Hamming window. The shape of the rectangular window changes sharply at the boundary, and the Gibbs effect will appear in the frequency domain. In order to reduce spectrum leakage, the Hamming window is preferably used as the window function of STFT. The signal to be transformed is newly windowed. Windowing is to multiply the window function by the signal to be transformed point by point. The following is a specific short-time Fourier transform formula:

[0076]

[0077] Among them, R(m,k) is the converted frequency domain signal, m is the discrete digital voice signal frame number, H represents the overlap rate, N = (frame length * sampling rate) / overlap rate, ω(n) represents the window function, the length of the window function must be equal to N, and k represents the kth frequency point.

[0078] The time domain signal is converted into a frequency domain signal through short-time Fourier transform. The frequency domain function obtained by this transform is called Fourier spectrum, which represents the size (or intensity) of each frequency domain component contained in the time domain waveform. The frequency domain signal is divided into various sub-bands according to the frequency in the figure. After calculating the frequency domain flatness and average energy ratio of each sub-band, the howling judgment is made according to the first preset threshold and the second preset threshold; if it is judged that there is a howling sub-band, the howling sub-band is suppressed to obtain a sound signal Y(t) without howling.

[0079] Example 3

[0080] Reference Figure 1 and Figure 3 , shows a flowchart of an embodiment of a method for suppressing voice howling in a monitoring system of the present invention, which may specifically include the following steps:

[0081] S1. Processing the collected speech signal according to preset rules to obtain several sub-bands;

[0082] S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio;

[0083] S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold;

[0084] S4, when there is a howling sub-band in S3, judging the total energy of the whole frequency according to the third preset threshold;

[0085] S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

[0086] Before step S1, the method further includes the following steps: S0, collecting voice signals of the monitoring system. First, the voice signals of the monitoring system are collected, and the collected voice signals are processed through the above steps.

[0087] Furthermore, the step of suppressing in step S5 includes:

[0088] S51, testing the local monitoring system and the on-site environment to obtain the M frequency points with the greatest howling impact and the optimal energy attenuation value corresponding to each frequency point;

[0089] Specifically, the above steps are completed by using a sine wave generator, and the sine wave signal frequency test range [f min ,f max ], where f min and f max The sine wave frequency is tested for the minimum and maximum values ​​respectively; then the sine wave generator emits the minimum frequency f in a time-sharing manner. min Increase to the maximum frequency f max The equal-amplitude sine wave signals emitted at different frequencies have equal amplitudes, so the initial signal energies are all equal; the sine wave signals are processed and finally propagated into the environment.

[0090] Then, the sinusoidal wave signal is collected by the microphone and processed by short-time Fourier transform to obtain the frequency domain signal S(m,k); wherein S(m,k) represents the frequency domain signal of the current mth frame signal at the kth frequency point, k=0, 1,..., N-1; N represents the total number of frequency points; thus, a total of N frequency domain signals S(m,k) are obtained.

[0091] Calculate the energy P(m,k) of each frequency domain signal S(m,k), sort the calculated energies P(m,k) of the N frequency domain signals S(m,k), select the M frequency domain signals S(m,k) with the highest energy values, where M is less than or equal to N, and the corresponding frequency points are recorded as the M frequency points with the greatest howling influence obtained by screening, and calculate the energy average value P0 of the other NM frequency domain signals, and subtract the energy of each frequency point screened out from the energy average value P0, and the difference is the optimal energy attenuation value of the frequency point.

[0092] Preferably, the speech signal is preliminarily screened in step S51 to obtain the frequency points most likely to generate howling, and then howling is suppressed on the screened frequency points. In this way, on the one hand, too many sub-bands do not need to be processed at the same time, and on the other hand, the howling sub-bands are further subdivided, making the howling processing more efficient and high-quality.

[0093] S52, performing discretized howling suppression processing on the howling sub-bands determined as the howling sub-bands in the M frequency points to obtain a processed far-end discrete digital signal Y(n);

[0094] Specifically, not all signals are processed. Only the sound signals determined as howling sub-bands among the M frequency points with the greatest howling impact will be subjected to howling suppression processing. The far-end discrete digital signal Y(n) is obtained through Fourier transform, which greatly reduces the difficulty of data processing and the amount of data.

[0095] S53, performing notch and conversion processing on the remote discrete digital signal Y(n) to obtain an analog electrical signal R(t);

[0096] Specifically, the remote discrete digital signal Y(n) is transmitted to the notch filter for notch processing. The notch filter obtains the M frequency points that have the greatest howling effect and the optimal energy attenuation value corresponding to each frequency point. Then, the corresponding M frequency points are read from the remote discrete digital signal Y(n), and the energy of each read frequency point is attenuated. The attenuation value is the corresponding optimal energy attenuation value, thereby obtaining the signal after notch processing. The signal after notch processing is then converted by the audio processing module and output as an analog electrical signal R(t).

[0097] S54, the analog electrical signal R(t) is amplified to obtain a sound signal Y(t).

[0098] Specifically, the processed signal is processed twice to obtain an analog electrical signal R(t), which is further divided into two branches. The signal of one branch is processed by various data to obtain the reference channel signal for the mixed signal collected at the next moment to perform howling suppression processing, and the signal of the other branch passes through the power amplifier and is played through the speaker to produce the sound signal Y(t). At this time, Y(t) is the signal after howling processing.

[0099] It should be added that the time from the sound being collected by the microphone to the sound being emitted by the speaker is one cycle. The time from the sound being collected by the microphone is recorded as t0. The time from the sound being processed by the audio processor to the sound being emitted by the speaker is t1. The sound emitted by the speaker is then transmitted to the microphone through the environment and is collected by the microphone together with the sound emitted by the sound source at this time. This time is recorded as t0. ’, t1-t0 is a sound processing cycle, and this goes back and forth and affects each other. The next moment mentioned above often means there is a previous moment. For detailed explanation, taking this cycle as an example, a certain moment is recorded as t0. The sound collected by the microphone at this time is the sound emitted by the speaker at the previous moment, which is finally collected by the microphone after being affected by the environment, and the sound emitted by the sound source at this time. Note that this sound source refers to the sound source other than the speaker. The one earlier than t0 is the previous moment, and the one later than t0 is the next moment. Specifically, if a sound processing cycle is 0.5s, the timing starts when the stopwatch shows 60s. At 60s, the microphone collects sound, and after 0.5s, the speaker emits sound. If the sound is transmitted through the air to the microphone position, it also takes 0.2s. When it is 60.7s, this is the next moment, and 59.3s is the previous moment.

[0100] Furthermore, the step S54 further includes the following steps:

[0101] S55, the echo signal E(t) generated by the sound signal Y(t) after being affected by the environment is mixed with the near-end voice signal and collected, and then the process jumps to step S51 to continue.

[0102] The above steps are continuously cycled, and the sound signal generated at the previous moment and released by the speaker is collected by the microphone and the voice signal emitted by the sound source other than the speaker at the next step, and then the howling judgment is performed, which is continuously cycled and repeated to achieve voice howling suppression of the monitoring system.

[0103] Example 4

[0104] Reference Figure 1 and Figure 3 , shows a flowchart of an embodiment of a method for suppressing voice howling in a monitoring system of the present invention, which may specifically include the following steps:

[0105] S1. Processing the collected speech signal according to preset rules to obtain several sub-bands;

[0106] S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio;

[0107] S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold;

[0108] S4, when there is a howling sub-band in S3, judging the total energy of the whole frequency according to the third preset threshold;

[0109] S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

[0110] Before step S1, the method further includes the following steps: S0, collecting voice signals of the monitoring system. First, the voice signals of the monitoring system are collected, and the collected voice signals are processed through the above steps.

[0111] Furthermore, the howling judgment in step S3 is specifically the following steps:

[0112] S31, setting the threshold of frequency domain flatness to P, and setting the threshold of average energy ratio to N;

[0113] S32: If the frequency domain flatness of a certain sub-band is lower than P and the average energy ratio is higher than N, it can be determined as a howling sub-band.

[0114] It should be noted that the threshold of frequency domain flatness can be first recorded in the absence of howling, and then the system gain is gradually increased until howling occurs, and the frequency domain flatness at this time is recorded. The threshold should be set between the two to ensure that there is enough distinction between the state with howling and the state without howling; a large amount of audio data without howling can also be analyzed, the distribution of its frequency domain flatness can be calculated, and then a statistically reliable threshold can be selected, such as using the mean value minus a certain multiple of the standard deviation. The initial threshold can be set more conservatively to avoid misjudgment, and then adjusted according to actual performance.

[0115] The threshold of the average energy ratio can be set based on the historical data of the average energy and energy ratio of the audio signal over a period of time. For example, if it is observed that the energy ratio rarely exceeds 1.5 under normal circumstances, the threshold can be set above 1.5. The threshold should have a certain safety margin to prevent misjudgment due to instantaneous signal fluctuations.

[0116] Furthermore, the frequency domain flatness calculation formula is:

[0117]

[0118] Among them, σ E is the standard deviation of all subband energies, μ E is the average value of all sub-band energies, and if the frequency domain flatness is below a certain threshold, this indicates that the spectrum may have abnormal energy spikes, which may be a sign of howling.

[0119] Furthermore, the average energy ratio calculation formula is:

[0120]

[0121] Among them, E c is the energy of the current frame, E a It is the average value of frame energy in the past period of time.

[0122] Furthermore, before making a full-frequency total energy determination, the following steps are also included:

[0123] The total energy of the whole frequency spectrum is calculated by the following formula;

[0124]

[0125] Where: X(n) is the complex amplitude of the nth frequency component;

[0126] N is the total number of frequency components.

[0127] It should be noted that the total energy of the entire frequency is calculated by the above formula, and the existence of the howling sub-band is judged. If the total energy of the entire frequency of the howling sub-band calculated is greater than the threshold, the suppression process is performed.

[0128] Example 5

[0129] Reference Figure 1 and Figure 3 , shows a flowchart of an embodiment of a method for suppressing voice howling in a monitoring system of the present invention, which may specifically include the following steps:

[0130] S1. Processing the collected speech signal according to preset rules to obtain several sub-bands;

[0131] S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio;

[0132] S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold;

[0133] S4, when there is a howling sub-band in S3, judging the total energy of the whole frequency according to the third preset threshold;

[0134] S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

[0135] Before step S1, the method further includes the following steps: S0, collecting voice signals of the monitoring system. First, the voice signals of the monitoring system are collected, and the collected voice signals are processed through the above steps.

[0136] Furthermore, it also includes a monitoring system, which is used to implement the method described in claims 1-8, and the monitoring system includes: a sound transceiver module, an audio processor, a camera and a power amplifier; the sound transceiver module includes a microphone and a speaker; the two sound transceiver modules are respectively arranged on the camera and at a preset position away from the camera, for collecting voice signals and emitting sound signals; the audio processor is electrically connected to the microphone for processing the collected voice signals; the power amplifier is connected to the audio processor for further secondary processing of the voice signal; the power amplifier is electrically connected to the speaker for playing the secondary processed voice signal.

[0137] It should be noted that the monitoring system in actual use often includes a monitoring room and a camera. The user observes in the monitoring room, and the camera is set in various outdoor scenes. Through the sound transceiver module, which is respectively set in the camera and the monitoring room, two-way communication can be achieved. Of course, there is at least a microphone set in the monitoring room and a speaker set in the camera, so as to achieve one-way voice input. When the microphone and the speaker are set close to each other, it is necessary to strictly prevent the generation of howling. However, through the method of claims 1-8, the generation of howling can be effectively suppressed.

[0138] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0139] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0140] Those skilled in the art will appreciate that the embodiments of the present invention may be provided as methods, devices or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects.

[0141] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0142] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0143] The above is a detailed introduction to a method for suppressing voice howling in a monitoring system provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technicians in this field, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for suppressing voice howling in a monitoring system, characterized in that: The following steps are involved: S1. Processing the collected speech signal according to preset rules to obtain several sub-bands; S2. Calculate the frequency domain data of each sub-band respectively to obtain the frequency domain flatness and average energy ratio; S3, based on the frequency domain flatness and average energy ratio of each sub-band, making a howling decision according to a first preset threshold and a second preset threshold; S4, when there is a howling sub-band in S3, judging the total energy of the whole frequency according to the third preset threshold; S5. When the total energy of the entire frequency range is higher than the threshold, the howling subband is suppressed to obtain a sound signal Y(t).

2. The method for suppressing voice howling in a monitoring system according to claim 1, characterized in that: Before step S1, the following steps are also included: S0. Collect the voice signal of the monitoring system.

3. The method for suppressing voice howling in a monitoring system according to claim 1, characterized in that: The step S1 comprises the following steps: S11, performing short-time Fourier transform on the speech signal to obtain a spectrum; S12. Divide the spectrum into several sub-bands.

4. The method for suppressing voice howling in a monitoring system according to claim 3, characterized in that: The step S11 comprises: performing short-time Fourier transform on the speech signal by the following transformation formula to obtain a spectrum; Transformation formula: Among them, ω(n) is the window function, which is used to intercept a signal of the window width of x(n) at time n.

5. The method for suppressing voice howling in a monitoring system according to claim 4, characterized in that: The window function of the short-time Fourier transform is a Hamming window.

6. The method for suppressing voice howling in a monitoring system according to claim 1, characterized in that: The step of suppressing in step S5 includes: S51, testing the local monitoring system and the on-site environment to obtain the M frequency points with the greatest howling impact and the optimal energy attenuation value corresponding to each frequency point; S52, performing discretized howling suppression processing on the howling sub-bands determined as the howling sub-bands in the M frequency points to obtain a processed far-end discrete digital signal Y(n); S53, performing notch processing on the remote discrete digital signal Y(n) to obtain an analog electrical signal R(t); S54, the analog electrical signal R(t) is amplified to obtain a sound signal Y(t).

7. The method for suppressing voice howling in a monitoring system according to claim 6, characterized in that: The step S54 further includes the following steps: S55, the echo signal E(t) generated by the sound signal Y(t) after being affected by the environment is mixed with the near-end voice signal and collected, and then the process jumps to step S51 to continue.

8. The method for suppressing voice howling in a monitoring system according to claim 1, characterized in that: The howling judgment in step S3 is specifically the following steps: S31, setting the threshold of frequency domain flatness to P, and setting the threshold of average energy ratio to N; S32: If the frequency domain flatness of a certain sub-band is lower than P and the average energy ratio is higher than N, it can be determined as a howling sub-band.

9. The method for suppressing voice howling in a monitoring system according to claim 1, characterized in that: Before making a full-frequency total energy judgment, the following steps are also included: The total energy of the entire frequency spectrum is calculated using the following formula: Where: X(n) is the complex amplitude of the nth frequency component; N is the total number of frequency components.

10. A monitoring system, characterized in that: The monitoring system is used to implement the method described in claims 1-8, and the monitoring system includes: a sound transceiver module, an audio processor, a camera and a power amplifier; The sound transceiver module includes a microphone and a speaker; The two sound transceiver modules are respectively arranged at the camera and at a preset position away from the camera, and are used to collect voice signals and send sound signals; The audio processor is electrically connected to the microphone and is used to process the collected voice signal; The power amplifier is connected to the audio processor to further process the voice signal; The power amplifier is electrically connected to the sound box and is used to play the secondary processed voice signal.

Images