Interphone noise suppression method and noise suppression device

Abstract

The invention provides an interphone noise suppression method and a noise suppression device, which are used for judging whether a sound signal is noise or normal voice by utilizing a decibel value of an actually acquired sound signal and / or a ratio of a voice signal to a total sound signal, and switching on or switching off an information channel between an interphone and an earphone according to a judgment result. Moreover, the decibel threshold value of the normal voice signal is dynamically updated according to the environmental data, so that the accuracy of noise discrimination and the noise suppression effect are improved. The noise suppression device designed by the invention is externally hung between the interphone and the earphone, when the noise suppression device is used for suppressing noise, the interior of the original interphone does not need to be disassembled or maintained, and the noise suppression device is particularly suitable for maintenance of old equipment interphones.

Description

Technical Field

[0001] This invention belongs to the field of telephone noise suppression technology, specifically relating to a telephone noise suppression method and a noise suppression device. Background Technology

[0002] Among various equipment, the intercom is a crucial voice communication terminal, and its communication quality directly affects the transmission of instructions and the execution of tasks by operators. However, in complex noisy environments (such as engine noise, mechanical vibration, electromagnetic interference, etc.), intercoms are often severely affected by noise interference, leading to decreased communication quality or a poor auditory experience for the operator, and even affecting mission safety. Currently, common noise suppression technologies mainly include:

[0003] 1. Active noise cancellation technology: It generates sound waves that are opposite to ambient noise to cancel it out. It is suitable for low-frequency steady-state noise (such as engine noise), but has a weaker ability to handle high-frequency or sudden noise.

[0004] 2. Passive noise reduction technology: It uses physical sound insulation materials to block the transmission of noise, but its effect on suppressing low-frequency noise is limited. In addition, due to the increase in the size of the device, it may affect the comfort of wearing the handset and the original structure of the handset.

[0005] In addition, most existing noise suppression methods often require complex hardware design or algorithm support, and noise reduction design is usually carried out in the early stages of product design. However, for old equipment that has been in service for a long time (such as aircraft, tanks, etc.), the communication quality of its communicator may deteriorate due to component aging, early design defects, or changes in the noise environment. However, traditional noise reduction methods are difficult to apply for the following reasons: (1) Hardware limitations: It is impossible to replace or significantly modify the original communicator structure. (2) Cost and feasibility: Complex noise reduction schemes require customized development and have high maintenance costs. (3) Real-time requirements: Low latency processing is required in combat and operational environments, which existing algorithms may not be able to meet. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing telephone noise suppression technologies, such as complex hardware design and unsuitability for noise reduction and maintenance of old equipment, and to provide a telephone noise suppression method and noise suppression device.

[0007] To achieve the above objectives, the technical solution provided by this invention is:

[0008] A method for suppressing noise in a communication device includes the following steps:

[0009] Step 1: Collect the audio signal output from the communicator and obtain the current decibel value of the audio signal;

[0010] Step 2: Determine whether the current decibel value of the sound signal is within the preset normal speech signal decibel threshold;

[0011] If the current sound signal decibel value is within the normal speech signal decibel threshold, the current sound signal is determined to be normal speech and the information channel between the communicator and the headset is opened; otherwise, the current sound signal is determined to be noise and the information channel between the communicator and the headset is closed.

[0012] The decibel threshold for normal speech signals is dynamically updated based on environmental data.

[0013] Furthermore, another noise suppression scheme is provided, which, after determining that the decibel value of the current sound signal is within the normal speech signal decibel threshold, also includes the following process:

[0014] The ratio of speech signal to total sound signal in the current sound signal is analyzed and obtained, and it is determined whether the ratio reaches a preset speech signal ratio threshold.

[0015] If this condition is met, the current audio signal is determined to be normal speech, and the information channel between the communicator and the headset is maintained.

[0016] If the signal is not reached, the current audio signal is determined to be noise and the information channel between the communicator and the headset is disconnected.

[0017] Furthermore, another noise suppression scheme is provided, which, in addition to steps 1 and 2, includes the following process:

[0018] Analyze and obtain the ratio of speech signal to total sound signal in the current sound signal;

[0019] The formula converts the current sound signal's decibel value into a percentage of its decibel value:

[0020]

[0021] In the formula, a represents the percentage of the decibel value; x represents the decibel value of the current sound signal; μ represents the median value of the decibel threshold of the normal speech signal. The decibel threshold of the normal speech signal is defined as [μ1,μ2], then μ=(μ1+μ2) / 2;

[0022] The weighted fusion value, calculated using the formula, is the percentage of the stated decibel value and the ratio of the speech signal to the total sound signal.

[0023] result=ρ1a+ρ2b

[0024] In the formula, result represents the weighted fusion value, a represents the percentage of the decibel value, b represents the ratio of the speech signal to the total sound signal; ρ1 is the weight of the percentage of the decibel value; ρ2 is the weight of the ratio of the speech signal to the total sound signal, and both are not 0.

[0025] The calculated weighted fusion value is compared with a preset weighted fusion result threshold. If the weighted fusion value is greater than the weighted fusion result threshold, the current sound signal is determined to be normal speech, and the information channel between the communicator and the headset is opened; otherwise, the current sound signal is determined to be noise, and the information channel between the communicator and the headset is cut off.

[0026] Among them, ρ1, ρ2, and the weighted fusion result threshold were obtained through multiple adjustments.

[0027] Furthermore, the process also includes the following: dynamically adjusting the decibel threshold of the normal speech signal by monitoring changes in environmental noise in real time through a decibel threshold learning module.

[0028] Furthermore, the process of dynamically adjusting the decibel threshold of the normal speech signal through the decibel threshold learning module is as follows:

[0029] Data acquisition: Simultaneously acquire environmental sound signals and environmental factor data under different environmental conditions, and record the decibel threshold of the normal speech signal when it is actually judged as normal speech or noise;

[0030] Feature extraction: The environmental factor data is processed to extract feature data including background noise level and environmental noise level.

[0031] Machine learning model training: Construct a machine learning model, using the aforementioned feature data as input data and the corresponding normal speech signal decibel threshold as output value, and train the constructed machine learning model; the machine learning model employs linear regression, decision tree, or support vector machine algorithms;

[0032] Threshold prediction: Obtain current environmental factor data, input it into a trained machine learning model, and predict the decibel threshold of the normal speech signal corresponding to the current environment.

[0033] A telephone noise suppression device is also provided, wherein the telephone noise suppression device is disposed between the telephone and the earpiece;

[0034] The communicator noise suppression device includes a sound decibel collector, an embedded controller, and a switch module;

[0035] The sound decibel collector is used to collect the decibel value of the current sound signal;

[0036] The embedded controller is configured with a normal voice signal decibel threshold, which is used to determine whether the current sound signal decibel value is within the preset normal voice signal decibel threshold, and output control commands according to the determination result.

[0037] The input terminal of the switch module is connected to the communicator, and the output terminal is connected to the earpiece; the control terminal of the switch module is connected to the embedded controller, and is used to switch the signal channel between the communicator and the earpiece on and off according to the control command output by the embedded controller.

[0038] Furthermore, it also includes a voice activity detection module and a decibel threshold learning module;

[0039] The voice activity detection module is used to analyze the ratio of voice signal to total voice signal in the current sound signal;

[0040] The decibel threshold learning module is used to dynamically update the decibel threshold of normal speech signals based on current environmental factors.

[0041] Furthermore, the embedded controller also includes a decibel conversion unit and a weighted fusion unit.

[0042] The decibel conversion unit is used to convert the decibel value of the current sound signal into a percentage of the decibel value;

[0043] The weighted fusion unit is used to calculate the weighted fusion value of the percentage of the decibel value and the ratio of the speech signal to the total sound signal, and to compare the weighted fusion value with a preset weighted fusion result threshold, and to determine whether the current sound signal belongs to normal speech or noise based on the comparison result.

[0044] Furthermore, when the communicator noise suppression device is applied to voice communication in the equipment cockpit, the normal voice signal decibel threshold is 40-60 decibels.

[0045] Furthermore, a computer-readable storage medium stores a computer program that, when executed by a processor, implements the aforementioned telephone noise suppression method.

[0046] The advantages of this invention are:

[0047] 1. The noise suppression method provided by this invention can achieve efficient adaptive suppression of intercom noise, and its specific advantages are as follows:

[0048] (1) In the method of this invention, noise and normal speech can be quickly identified by comparing the decibel value of the collected sound signal with the decibel threshold of the normal speech signal. Control commands are then output based on the identification results to connect or disconnect the information channel between the communicator and the headset. Based on this, this application proposes another solution: using speech activity detection and analysis technology to further obtain the ratio of the current sound signal to the total sound signal. Whether the ratio reaches a preset speech signal ratio threshold determines whether the current sound is normal speech or noise, improving the accuracy of speech signal identification and avoiding misjudging normal speech as noise. Simultaneously, by utilizing the decibel difference of the sound signal, low-decibel sounds are identified as noise and suppressed, while high-decibel sounds are retained as normal content, thus achieving a simple and efficient noise reduction effect. Combining speech activity detection and decibel difference, efficient noise identification and suppression are achieved.

[0049] (2) The method of the present invention incorporates a dynamic adjustment mechanism for the decibel threshold of the normal speech signal, which can automatically optimize the decibel threshold of the normal speech signal according to environmental changes, further improving the noise reduction effect and maintaining high-efficiency noise reduction performance in both quiet and noisy environments.

[0050] 2. The noise suppression device for the intercom provided by the present invention is externally mounted between the intercom and the headset, enabling external maintenance of the intercom without disassembling or maintaining the original intercom. It is particularly suitable for the maintenance of old equipment (such as old models of fighter jets, tanks, etc.).

[0051] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0052] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0053] Figure 1 This is a flowchart of the communication device noise suppression method provided in the embodiments of the present invention;

[0054] Figure 2 This is a block diagram of the communication device noise suppression device provided in the embodiments of the present invention. Detailed Implementation

[0055] The embodiments of the present invention are described in detail below. These embodiments are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0056] Taking the cockpit of a certain type of equipment as an example, sound signals below 20 decibels are considered quiet, sound signals between 20 and 40 decibels and sound signals between 40 and 60 decibels are considered normal speech signals.

[0057] Example 1

[0058] Example 1 provides a method for suppressing noise in a communication device, comprising the following steps:

[0059] Step 1: Collect the audio signal output from the communicator and obtain the current decibel value of the audio signal;

[0060] Step 2: Determine whether the current decibel value of the sound signal is within the preset normal speech signal decibel threshold;

[0061] If the current audio signal decibel value is within the normal speech signal decibel threshold, the current audio signal is determined to be normal speech, and the information channel between the communicator and the headset is opened; otherwise, the current audio signal is determined to be noise, and the information channel between the communicator and the headset is closed. The normal speech signal decibel threshold is dynamically updated based on environmental factors.

[0062] Example 2

[0063] Based on the method of Example 1, Example 2 further includes a step of analyzing the proportion of speech signal to total sound signal in the current sound signal. (Refer to...) Figure 1 The specific steps for the inhibition method are as follows:

[0064] Step 1: Collect the audio signal output by the communicator, obtain the current decibel value of the audio signal, and analyze the ratio of the speech signal to the total audio signal in the current audio signal;

[0065] Step 2: Determine whether the current decibel value of the sound signal is within the preset normal speech signal decibel threshold;

[0066] If the current audio signal decibel value is within the normal audio signal decibel threshold, proceed to step 3; otherwise, determine that the current audio signal is noise and disconnect the information channel between the communicator and the headset.

[0067] Step 3: Set a preset voice signal percentage threshold. Determine whether the ratio of voice signal to total voice signal in the current audio signal reaches the preset voice signal percentage threshold. If the ratio reaches the threshold, it is determined to be normal voice, and the information channel between the handset and the headset is maintained. Otherwise, it is determined to be noise, and the information channel between the handset and the headset is disconnected. In this embodiment, the voice signal percentage threshold is set to 60%.

[0068] Example 3

[0069] This embodiment provides another method for noise suppression based on the decibel value of the sound signal and the ratio of the speech signal to the total sound signal, including the following steps:

[0070] Step 1: Collect the audio signal output by the communicator, obtain the current decibel value of the audio signal, and analyze the ratio of the speech signal to the total audio signal in the current audio signal;

[0071] Step 2: Convert the current sound signal decibel value obtained in Step 1 into a percentage of the decibel value, and calculate the weighted fusion value of the percentage of the decibel value and the ratio of the speech signal to the total sound signal;

[0072] The formula for converting decibel values ​​to percentages is:

[0073]

[0074] In the formula, a represents the percentage of the decibel value; x represents the decibel value of the current sound signal; μ represents the median value of the decibel threshold of the normal speech signal. The decibel threshold of the normal speech signal is defined as [μ1, μ2], then μ = (μ1 + μ2) / 2.

[0075] The weighted fusion formula is:

[0076] result=ρ1a+ρ2b

[0077] In the formula, result represents the weighted fusion value, a represents the percentage of the decibel value, b represents the ratio of the speech signal to the total sound signal; ρ1 is the weight of the percentage of the decibel value; ρ2 is the weight of the ratio of the speech signal to the total sound signal, and all of them are not 0.

[0078] Step 3: Preset the weighted fusion result threshold. Compare the calculated weighted fusion value with the weighted fusion result threshold. If the weighted fusion value is greater than the weighted fusion result threshold, the current sound signal is determined to be normal speech, and the information channel between the communicator and the headset is opened; otherwise, the current sound signal is determined to be noise, and the information channel between the communicator and the headset is cut off.

[0079] Among them, ρ1, ρ2 and the weighted fusion result threshold are adjusted according to the actual sound judgment effect. Multiple adjustment tests are required to determine the optimal weight value and weighted fusion result threshold.

[0080] To achieve noise suppression in the handset, this invention provides a handset noise suppression device, externally attached between the handset and the headset, with one end connected to the output of the handset and the other end connected to the input of the headset. (See reference...) Figure 2The communicator noise suppression device includes a sound decibel sensor, an embedded controller, and a switch module. The sound decibel sensor collects the decibel value of the current sound signal output by the communicator. The embedded controller is configured with a normal speech signal decibel threshold. It receives the decibel value of the current sound signal output by the sound decibel sensor, determines whether the current sound signal decibel value is within the preset normal speech signal decibel threshold, and outputs a control command to the switch control module based on the determination result. In this embodiment, the normal speech signal decibel threshold is 40-60 decibels. The input terminal of the switch module is connected to the communicator, and the output terminal is connected to the headset. The control terminal of the switch module is connected to the embedded controller and is used to open and close the signal channel between the communicator and the headset according to the control command output by the embedded controller: if the embedded controller determines that the current sound signal is normal speech, it outputs a conduction control command to control the switch control module to open the signal channel between the communicator and the headset; if it determines that the sound signal is noise, it outputs a disconnection control command to control the switch module to disconnect the signal channel between the communicator and the headset, thereby achieving noise suppression.

[0081] Specifically, it also includes a decibel threshold learning module, which dynamically and adaptively adjusts the decibel threshold of normal speech signals configured in the embedded controller based on environmental data and machine learning algorithms. The specific adjustment steps include:

[0082] Data acquisition: Simultaneously acquire environmental sound signals and environmental factor data under different environmental conditions, and record the decibel threshold of the normal speech signal when it is actually judged as normal speech or noise.

[0083] Feature extraction: The collected environmental factor data is processed, including quantifying the background noise level and classifying the environmental noise level into different levels. Feature data including background noise level and environmental noise level are extracted and used as input data for the decibel threshold learning module.

[0084] Machine learning model training: A machine learning model is built using extracted feature data as input and the corresponding decibel threshold of a normal speech signal as output. The model is then trained. Learning algorithms for the machine learning model can include linear regression, decision trees, or support vector machines.

[0085] Threshold prediction: Obtain current environmental factor data, input it into a trained machine learning model, and the model predicts and outputs the decibel threshold of the normal speech signal corresponding to the current environment.

[0086] In a further design, the noise suppression device for the communicator of the present invention also includes a voice activity detection module, which is used to perform voice activity detection and analysis on the sound signal to obtain the ratio of the voice signal to the total sound signal.

[0087] In a further design, the embedded controller in the communication device of the present invention also includes a decibel conversion unit and a weighted fusion unit. The decibel conversion unit is used to convert the decibel value of the current sound signal into a percentage of the decibel value according to the following formula:

[0088]

[0089] Where: a represents the percentage of the decibel value; x represents the decibel value of the current sound signal; μ represents the median value of the decibel threshold of the normal speech signal. The decibel threshold of the normal speech signal is defined as [μ1, μ2], then μ = (μ1 + μ2) / 2.

[0090] The weighted fusion unit is used to calculate a weighted fusion value based on the percentage of the decibel value and the ratio of the speech signal to the total sound signal, according to the following formula:

[0091] result=ρ1a+ρ2b

[0092] Where: result represents the weighted fusion value, a represents the percentage of the decibel value, b represents the ratio of the speech signal to the total sound signal; ρ1 is the weight of the percentage of the decibel value; ρ2 is the weight of the ratio of the speech signal to the total sound signal. ρ1 and ρ2 are determined by adjusting based on the results of multiple actual sound signal judgments, and neither is 0. If any weight is zero, this formula is not applicable for judgment.

[0093] The present invention provides a method and device for suppressing noise in a handset. By analyzing the decibel value of the sound signal and / or the ratio of the voice signal to the total sound signal, the method determines whether the output sound signal of the handset is normal speech or noise. Based on this determination, it establishes or disconnects the information channel between the handset and the headset. Furthermore, the decibel threshold for normal speech signals is dynamically updated according to environmental factors, improving the accuracy of noise detection and the effectiveness of noise suppression. The noise suppression device is externally mounted, enabling external repair of the handset without requiring disassembly or maintenance of the original equipment. This makes it suitable for repairing handsets in older equipment.

[0094] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the scope of the technology disclosed in the present invention, and such modifications or substitutions should all be covered within the scope of protection of the present invention.

Claims

1. A method for suppressing noise in a communication device, characterized in that, Includes the following steps: Step 1: Collect the audio signal output from the communicator and obtain the current decibel value of the audio signal; Step 2: Determine whether the current decibel value of the sound signal is within the preset normal speech signal decibel threshold; If the current sound signal decibel value is within the normal speech signal decibel threshold, the current sound signal is determined to be normal speech and the information channel between the communicator and the headset is opened; otherwise, the current sound signal is determined to be noise and the information channel between the communicator and the headset is closed. The decibel threshold for normal speech signals is dynamically updated based on environmental data.

2. The communication device noise suppression method according to claim 1, characterized in that, After determining that the current sound signal's decibel value is within the normal speech signal decibel threshold, the following process is also included: The ratio of speech signal to total sound signal in the current sound signal is analyzed and obtained, and it is determined whether the ratio reaches a preset speech signal ratio threshold. If this condition is met, the current audio signal is determined to be normal speech, and the information channel between the communicator and the headset is maintained. If the signal is not reached, the current audio signal is determined to be noise and the information channel between the communicator and the headset is disconnected.

3. The communication device noise suppression method according to claim 1, characterized in that, It also includes the following processes: Analyze and obtain the ratio of speech signal to total sound signal in the current sound signal; The formula converts the current sound signal's decibel value into a percentage of its decibel value: In the formula, a represents the percentage of the decibel value; x represents the decibel value of the current sound signal; μ represents the median value of the decibel threshold of the normal speech signal. The decibel threshold of the normal speech signal is defined as [μ1,μ2], then μ=(μ1+μ2) / 2; The weighted fusion value, calculated using the formula, is the percentage of the stated decibel value and the ratio of the speech signal to the total sound signal. result=ρ1a+ρ2b In the formula, result represents the weighted fusion value, a represents the percentage of the decibel value, b represents the ratio of the speech signal to the total sound signal, and ρ1 is the weight of the percentage of the decibel value. ρ2 is the weight value of the ratio of speech signal to total sound signal, and none of them are 0; The calculated weighted fusion value is compared with a preset weighted fusion result threshold. If the weighted fusion value is greater than the weighted fusion result threshold, the current sound signal is determined to be normal speech, and the information channel between the communicator and the headset is opened; otherwise, the current sound signal is determined to be noise, and the information channel between the communicator and the headset is cut off. Among them, ρ1, ρ2, and the weighted fusion result threshold were obtained through multiple adjustments.

4. The communication device noise suppression method according to claim 1, characterized in that, It also includes the following process: monitoring changes in environmental noise in real time through a decibel threshold learning module and dynamically adjusting the decibel threshold of the normal speech signal.

5. The communication device noise suppression method according to claim 4, characterized in that, The process of dynamically adjusting the decibel threshold of the normal speech signal through the decibel threshold learning module is as follows: Data acquisition: Simultaneously acquire environmental sound signals and environmental factor data under different environmental conditions, and record the decibel threshold of the normal speech signal when it is actually judged as normal speech or noise; Feature extraction: The environmental factor data is processed to extract feature data including background noise level and environmental noise level. Machine learning model training: Construct a machine learning model, using the aforementioned feature data as input data and the corresponding normal speech signal decibel threshold as output value, and train the constructed machine learning model; the machine learning model employs linear regression, decision tree, or support vector machine algorithms; Threshold prediction: Obtain current environmental factor data, input it into a trained machine learning model, and predict the decibel threshold of the normal speech signal corresponding to the current environment.

6. A noise suppression device for a telephone, characterized in that, The noise suppression device for the communicator is disposed between the communicator and the earpiece; The communicator noise suppression device includes a sound decibel collector, an embedded controller, and a switch module; The sound decibel collector is used to collect the decibel value of the current sound signal; The embedded controller is configured with a normal voice signal decibel threshold, which is used to determine whether the current sound signal decibel value is within the preset normal voice signal decibel threshold, and output control commands according to the determination result. The input terminal of the switch module is connected to the communicator, and the output terminal is connected to the earpiece; the control terminal of the switch module is connected to the embedded controller, and is used to switch the signal channel between the communicator and the earpiece on and off according to the control command output by the embedded controller.

7. The intercom noise suppression device according to claim 6, characterized in that, It also includes a voice activity detection module and a decibel threshold learning module; The voice activity detection module is used to analyze the ratio of voice signal to total voice signal in the current sound signal; The decibel threshold learning module is used to dynamically update the decibel threshold of normal speech signals based on current environmental factors.

8. The communication device noise suppression apparatus according to claim 7, characterized in that: The embedded controller also includes a decibel conversion unit and a weighted fusion unit. The decibel conversion unit is used to convert the decibel value of the current sound signal into a percentage of the decibel value; The weighted fusion unit is used to calculate the weighted fusion value of the percentage of the decibel value and the ratio of the speech signal to the total sound signal, and to compare the weighted fusion value with a preset weighted fusion result threshold, and to determine whether the current sound signal belongs to normal speech or noise based on the comparison result.

9. The intercom noise suppression device according to any one of claims 6-8, characterized in that, When the communication device is used for voice communication in the equipment cockpit, the normal voice signal decibel threshold is 40-60 decibels.

10. A computer-readable storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the telephone noise suppression method according to any one of claims 1-5.

Images