An automatic threshold controlled mixing calculation method

Abstract

The application discloses a kind of automatic door limit control mixing calculation method, it is related to mixing control technical field, by the signal correlation between automatic calculation signal threshold and channel is controlled each channel gain to mix sound;The whole process includes: collecting the sound signal of multiple microphone channels, the RMS value is calculated by sound signal, and the difference of maximum and minimum of RMS value, signal-to-noise ratio and zero-crossing rate are jointly calculated to obtain speech energy value, so as to divide main microphone channel and vice microphone channel according to speech energy value, the final gain of vice microphone channel is determined by the cosine similarity of MFCC feature of main microphone channel and vice microphone channel, so as to suppress the same sound source of vice microphone channel, reduce local echo and loudspeaker echo.

Description

Technical Field

[0001] This application relates to the field of mixing control technology, specifically to a mixing calculation method with automatic threshold control. Background Technology

[0002] In scenarios such as audio conferencing, karaoke, live streaming, and film and television post-production, it is often necessary to mix multiple audio signals (such as signals from multiple microphones) into a single output. Traditional mixing methods typically use simple additive mixing, and traditional thresholds rely on on-site adjustments and presets, which cannot cope with changes in the environment. When one microphone is speaking, the sound will be picked up by neighboring microphones, and the sound from the loudspeaker will also be picked up by other microphones, leading to unclear speech and, in severe cases, feedback. Summary of the Invention

[0003] The purpose of this application is to provide an automatic threshold control mixing calculation method, which solves the problems of crosstalk and echo in multi-microphone conferences, unclear voice, and increased risk of feedback.

[0004] This application is achieved through the following technical solution:

[0005] A mixing calculation method with automatic threshold control includes:

[0006] Collect the audio signals corresponding to each microphone channel, calculate the RMS value corresponding to the audio signals corresponding to each microphone channel, and track the maximum and minimum values ​​of the RMS values.

[0007] Based on the maximum and minimum values ​​of the RMS value, the RMS extreme values ​​are obtained, and the noise is tracked according to the RMS extreme values;

[0008] The zero-crossing rate and signal-to-noise ratio are obtained based on the tracked noise, and the speech energy value is obtained based on the zero-crossing rate and signal-to-noise ratio.

[0009] The voice energy value is compared with a preset threshold to determine the operating status of each microphone channel; the operating status is either active or inactive.

[0010] Acquire all target microphone channels that are active and calculate the MFCC features of each target microphone channel. At the same time, determine the target microphone channel with the largest voice energy value as the main microphone channel and the rest as secondary microphone channels.

[0011] Obtain the MFCC feature similarity between the secondary microphone channel and the primary microphone channel, and obtain the similarity gain based on the MFCC feature similarity;

[0012] Based on the similarity gain, speech energy value, and RMS value corresponding to the secondary microphone channel, the final gain corresponding to each microphone channel is obtained, and gain control is performed on each microphone channel based on the final gain to realize mixing calculation based on automatic threshold control.

[0013] In one possible implementation, the audio signal corresponding to each microphone channel is acquired, the RMS value corresponding to the audio signal corresponding to each microphone channel is calculated, and the maximum and minimum values ​​of the RMS values ​​are tracked, including:

[0014] Collect the audio signals corresponding to each microphone channel. Where n is the microphone channel number and k is the time point index;

[0015] Calculate the RMS value of the audio signal corresponding to each microphone channel:

[0016] ;

[0017] in, The RMS value of the m-th frame. This refers to the number of data points, specifically 128 points per frame.

[0018] The maximum and minimum values ​​of the RMS value are tracked as follows:

[0019] ;

[0020] ;

[0021] in, This is the minimum value tracked in the m-th frame. This is the maximum value tracked in frame m. To find the minimum value, To find the maximum value, The RMS value of the (m-1)th frame. Let be the RMS value of the m-L+1th frame, where L is the number of windows tracking the RMS value.

[0022] In one possible implementation, the RMS extreme values ​​are obtained based on the maximum and minimum values ​​of the RMS values, including:

[0023] To obtain the maximum and minimum values ​​of the smoothed value:

[0024] ;

[0025] ;

[0026] in, This represents the smoothed maximum value corresponding to the m-th frame. This shows the smoothing minimum value corresponding to the m-th frame. This is the minimum value tracked in the m-th frame. This is the maximum value tracked in frame m. For smoothing coefficients, The maximum smoothed value corresponding to the (m-1)th frame. This is the smoothed minimum value corresponding to the (m-1)th frame;

[0027] Based on the smoothed maximum and smoothed minimum values, the RMS extreme values ​​are obtained as follows:

[0028] ;

[0029] in, is the RMS extreme value of the m-th frame.

[0030] In one possible implementation, tracking noise based on the RMS extrema includes:

[0031] ;

[0032] in, The noise tracked in the m-th frame, For the noise tracked in the (m-1)th frame, For tracking coefficients, This is the minimum value tracked in the m-th frame. This is the minimum value tracked in the (m-1)th frame. The minimum value tracked at frame m-L+1 is denoted by , where L is the number of windows tracking the RMS value.

[0033] In one possible implementation, the zero-crossing rate and signal-to-noise ratio are obtained based on the tracked noise, including:

[0034] The reference data obtained based on the tracked noise is as follows:

[0035] ;

[0036] in, This is the reference data for the m-th frame; This refers to the number of data points, specifically 128 points per frame. For symbolic functions, Let n be the signal of the nth microphone at time k. Let n be the signal of the nth microphone at time k-1. The noise tracked in the (m-1)th frame;

[0037] The zero-crossing rate is obtained based on the reference data as follows:

[0038] ;

[0039] in, Let m be the zero-crossing rate of the m-th frame. Calculate the coefficient for the zero-crossing rate. The zero-crossing rate of the (m-1)th frame;

[0040] The signal-to-noise ratio is obtained based on the tracked noise:

[0041] ;

[0042] in, Let be the signal-to-noise ratio of the m-th frame. The RMS value of the m-th frame. The noise tracked in the m-th frame.

[0043] In one possible implementation, obtaining the speech energy value based on the zero-crossing rate and the signal-to-noise ratio includes:

[0044] Based on the zero-crossing rate and signal-to-noise ratio, the characteristic value is obtained as follows:

[0045] ;

[0046] in, Let m be the feature value of the m-th frame. As the first weight value, This is the second weight value. As the third weight value, The RMS extreme value of the m-th frame. Let m be the zero-crossing rate of the m-th frame. Let be the signal-to-noise ratio of the m-th frame. The range threshold, For zero-crossing threshold, This is the signal-to-noise ratio threshold.

[0047] The speech energy value is obtained by using feature values:

[0048] ;

[0049] in, Let be the speech energy value of the m-th frame, 'a' be the first decision parameter, and 'b' be the second decision parameter.

[0050] In one possible implementation, the voice energy value is compared with a preset threshold to determine the operating status of each microphone channel, including:

[0051] Determine whether the voice energy value corresponding to the microphone channel is greater than a preset threshold. If so, determine that the microphone channel is in an active state; otherwise, determine that the microphone channel is in an inactive state.

[0052] Iterate through each microphone channel to obtain the operating status of each microphone channel.

[0053] In one possible implementation, obtaining the MFCC feature similarity between the secondary microphone channel and the primary microphone channel includes:

[0054] ;

[0055] in, Here, 'ch' represents the MFCC feature similarity, and 'ch' represents the secondary microphone channel whose MFCC feature similarity is calculated with the primary microphone channel. MFCC characteristics of the main microphone channel, This is the MFCC feature for the secondary microphone channel.

[0056] In one possible implementation, obtaining a similarity gain based on the MFCC feature similarity includes:

[0057] ;

[0058] in, For similarity gain, The RMS value of the main microphone channel in the m-th frame. This represents the RMS value of the secondary microphone channel in the m-th frame.

[0059] In one possible implementation, the final gain for each microphone channel is obtained based on the similarity gain, speech energy value, and RMS value corresponding to the secondary microphone channel, including:

[0060] ;

[0061] in, This represents the final gain of the secondary microphone channel in frame m. This represents the speech energy value of the secondary microphone channel in the m-th frame. This represents the RMS value of the i-th microphone channel in the m-th frame. Let represent the speech energy value of the i-th microphone channel in the m-th frame, and NN represent the total number of active microphone channels.

[0062] Compared with the prior art, this application has the following advantages and beneficial effects:

[0063] This application provides an automatic threshold control mixing calculation method, which controls the gain of each channel for mixing by automatically calculating the signal correlation between the signal threshold and the channel. The whole process includes: acquiring sound signals from multiple microphone channels, calculating the RMS value of the sound signals, and calculating the speech energy value by jointly calculating the difference between the maximum and minimum RMS values, the signal-to-noise ratio, and the zero-crossing rate. Based on the speech energy value, the main microphone channel and the secondary microphone channel are divided. The final gain of the secondary microphone channel is determined by the cosine similarity of the MFCC features of the main microphone channel and the secondary microphone channel, thereby suppressing the same sound source in the secondary microphone channel and reducing local crosstalk and speaker echo. Attached Figure Description

[0064] To more clearly illustrate the technical solutions of the exemplary embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0065] Figure 1 A flowchart illustrating an automatic threshold control mixing calculation method provided in this application embodiment.

[0066] Figure 2 This is a schematic diagram of mixing control provided in an embodiment of this application. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this application are only for explaining this application and are not intended to limit this application.

[0068] like Figure 1 As shown in the figure, this application provides a mixing calculation method for automatic threshold control, including:

[0069] S101. Collect the audio signal corresponding to each microphone channel, calculate the RMS value corresponding to the audio signal corresponding to each microphone channel, and track the maximum and minimum values ​​of the RMS value.

[0070] S102. Based on the maximum and minimum values ​​of the RMS value, obtain the RMS extreme values, and track the noise according to the RMS extreme values;

[0071] S103. Obtain the zero-crossing rate and signal-to-noise ratio based on the tracked noise, and obtain the speech energy value based on the zero-crossing rate and signal-to-noise ratio.

[0072] S104. Compare the voice energy value with a preset threshold to determine the operating state of each microphone channel; the operating state is either active or inactive.

[0073] S105. Obtain all target microphone channels that are active and calculate the MFCC characteristics of each target microphone channel. At the same time, determine the target microphone channel with the largest voice energy value as the main microphone channel and the rest as secondary microphone channels.

[0074] S106. Obtain the MFCC feature similarity between the secondary microphone channel and the primary microphone channel, and obtain the similarity gain based on the MFCC feature similarity.

[0075] S107. Based on the similarity gain, speech energy value and RMS value corresponding to the secondary microphone channel, obtain the final gain corresponding to each microphone channel, and perform gain control on each microphone channel based on the final gain to realize mixing calculation based on automatic threshold control.

[0076] This application provides an automatic threshold control mixing calculation method, which controls the gain of each channel for mixing by automatically calculating the signal correlation between the signal threshold and the channel. The whole process includes: acquiring sound signals from multiple microphone channels, calculating the RMS value of the sound signals, and calculating the speech energy value by jointly calculating the difference between the maximum and minimum RMS values, the signal-to-noise ratio, and the zero-crossing rate. Based on the speech energy value, the main microphone channel and the secondary microphone channel are divided. The final gain of the secondary microphone channel is determined by the cosine similarity of the MFCC features of the main microphone channel and the secondary microphone channel, thereby suppressing the same sound source in the secondary microphone channel and reducing local crosstalk and speaker echo.

[0077] To facilitate understanding of the technical solutions described in the embodiments of this application by those skilled in the art, the hardware system used in the embodiments of this application will be introduced first. The hardware system consists of a multi-channel ADC acquisition board and a DSP audio processing board. The system is connected to a computer via a serial port, and the real-time spectrum is monitored on the PC interface.

[0078] like Figure 2 As shown, by processing the audio signals of each microphone channel, calculating the ratio weight of each microphone throughput to the total active microphone energy, the final gain corresponding to each microphone channel is obtained. Finally, gain control of each microphone channel is performed based on the final gain, which can effectively suppress interference sources.

[0079] In one possible implementation, the audio signal corresponding to each microphone channel is acquired, the RMS value corresponding to the audio signal corresponding to each microphone channel is calculated, and the maximum and minimum values ​​of the RMS values ​​are tracked, including:

[0080] Collect the audio signals corresponding to each microphone channel. Where n is the microphone channel number and k is the time point index;

[0081] The RMS value (root mean square value) of the audio signal corresponding to each microphone channel is calculated as follows:

[0082] ;

[0083] in, The RMS value of the m-th frame. This refers to the number of data points, specifically 128 points per frame.

[0084] The maximum and minimum values ​​of the RMS value are tracked as follows:

[0085] ;

[0086] ;

[0087] in, This is the minimum value tracked in the m-th frame. This is the maximum value tracked in frame m. To find the minimum value, To find the maximum value, The RMS value of the (m-1)th frame. Let be the RMS value of the m-L+1th frame, where L is the number of windows tracking the RMS value.

[0088] For example, by inputting data into an ADC acquisition board, real-time audio data x(n) is acquired, buffered at 128 points, and sampled at a frequency of 48kHz. The room sound signal can be acquired via the ADC and passed through a low-pass filter with a frequency of 1000Hz to obtain the sound signal. Where n represents the nth microphone, and in this example, the number of microphones is 8, and k represents the value at the kth time point of the channel.

[0089] In one possible implementation, the RMS extreme values ​​are obtained based on the maximum and minimum values ​​of the RMS values, including:

[0090] To obtain the maximum and minimum values ​​of the smoothed value:

[0091] ;

[0092] ;

[0093] in, This represents the smoothed maximum value corresponding to the m-th frame. This shows the smoothing minimum value corresponding to the m-th frame. This is the minimum value tracked in the m-th frame. This is the maximum value tracked in frame m. For smoothing coefficients, The maximum smoothed value corresponding to the (m-1)th frame. This is the smoothed minimum value corresponding to the (m-1)th frame;

[0094] Based on the smoothed maximum and smoothed minimum values, the RMS extreme values ​​are obtained as follows:

[0095] ;

[0096] in, is the RMS extreme value of the m-th frame.

[0097] Based on sound signals The RMS value, maximum value, and minimum value are calculated sequentially. A sliding window can be used to calculate the maximum and minimum values ​​of the tracking RMS value. In this embodiment, one frame of data contains 128 points, the number of tracking RMS value windows L is set to 24, and the smoothing coefficient... It can be set to 0.98.

[0098] In one possible implementation, tracking noise based on the RMS extrema includes:

[0099] ;

[0100] in, The noise tracked in the m-th frame, For the noise tracked in the (m-1)th frame, For tracking coefficients, This is the minimum value tracked in the m-th frame. This is the minimum value tracked in the (m-1)th frame. The minimum value tracked at frame m-L+1 is denoted by , where L is the number of windows tracking the RMS value.

[0101] In one possible implementation, the zero-crossing rate and signal-to-noise ratio are obtained based on the tracked noise, including:

[0102] The reference data obtained based on the tracked noise is as follows:

[0103] ;

[0104] in, This is the reference data for the m-th frame; This refers to the number of data points, specifically 128 points per frame. For symbolic functions, Let n be the signal of the nth microphone at time k. Let n be the signal of the nth microphone at time k-1. The noise tracked in the (m-1)th frame;

[0105] The zero-crossing rate is obtained based on the reference data as follows:

[0106] ;

[0107] in, Let m be the zero-crossing rate of the m-th frame. Calculate the coefficient for the zero-crossing rate. The zero-crossing rate of the (m-1)th frame;

[0108] The signal-to-noise ratio is obtained based on the tracked noise:

[0109] ;

[0110] in, Let be the signal-to-noise ratio of the m-th frame. The RMS value of the m-th frame. The noise tracked in the m-th frame.

[0111] In the embodiments of this application, the tracking coefficient It can be set to 0.92, and the zero-crossing rate calculation coefficient can be set to 0.82.

[0112] In one possible implementation, obtaining the speech energy value based on the zero-crossing rate and the signal-to-noise ratio includes:

[0113] Based on the zero-crossing rate and signal-to-noise ratio, the characteristic value is obtained as follows:

[0114] ;

[0115] in, Let m be the feature value of the m-th frame. As the first weight value, This is the second weight value. As the third weight value, The RMS extreme value of the m-th frame. Let m be the zero-crossing rate of the m-th frame. Let be the signal-to-noise ratio of the m-th frame. The range threshold can be set to 8; The zero-crossing threshold can be set to 015; The signal-to-noise ratio threshold can be set to 18.

[0116] In this embodiment of the application, the first weight value It can be set to 0.7, the second weight value. It can be set to 0.5, the third weight value. It can be set to 0.7. It can be set to 8. It can be set to 0.15. It can be set to 18.

[0117] The speech energy value is obtained by using feature values:

[0118] ;

[0119] in, Let be the speech energy value of the m-th frame, 'a' be the first decision parameter, and 'b' be the second decision parameter.

[0120] In this embodiment of the application, the first decision parameter a can be set to 1.464237, and the second decision parameter b can be set to 1.459344.

[0121] In one possible implementation, the voice energy value is compared with a preset threshold to determine the operating status of each microphone channel, including:

[0122] Determine whether the voice energy value corresponding to the microphone channel is greater than a preset threshold. If so, determine that the microphone channel is in an active state; otherwise, determine that the microphone channel is in an inactive state.

[0123] Iterate through each microphone channel to obtain the operating status of each microphone channel.

[0124] For example, assuming the preset threshold is set to 0.3, then when the voice energy value corresponding to the microphone channel... If the value is greater than 0.3, the corresponding microphone channel can be considered to be in an active state; when the voice energy value corresponding to the microphone channel... If the value is less than or equal to 0.3, the corresponding microphone channel can be considered to be inactive.

[0125] according to Identify the active microphone channels, calculate the MFCC feature vector of the active microphone channels, find the microphone with the highest energy, and determine it as the primary microphone channel. The other active microphone channels are designated as secondary microphone channels.

[0126] In one possible implementation, obtaining the MFCC feature similarity between the secondary microphone channel and the primary microphone channel includes:

[0127] ;

[0128] in, Here, 'ch' represents the MFCC feature similarity, and 'ch' represents the secondary microphone channel whose MFCC feature similarity is calculated with the primary microphone channel. MFCC characteristics of the main microphone channel, This is the MFCC feature for the secondary microphone channel.

[0129] In one possible implementation, obtaining a similarity gain based on the MFCC feature similarity includes:

[0130] ;

[0131] in, For similarity gain, The RMS value of the main microphone channel in the m-th frame. This represents the RMS value of the secondary microphone channel in the m-th frame.

[0132] In one possible implementation, the final gain for each microphone channel is obtained based on the similarity gain, speech energy value, and RMS value corresponding to the secondary microphone channel, including:

[0133] ;

[0134] in, This represents the final gain of the secondary microphone channel in frame m. This represents the speech energy value of the secondary microphone channel in the m-th frame. This represents the RMS value of the i-th microphone channel in the m-th frame. This represents the speech energy value of the i-th microphone channel in the m-th frame, and NN represents the total number of active microphone channels. When no active microphone is involved, the microphone of the previous speaker will be forced to be on. If all microphones are not activated for a long time, the gain will be forced to 0 (if the duration of inactivity exceeds a preset time threshold, it is considered that all microphones have been inactive for a long time). For microphone channels that are inactive, the final gain of the corresponding microphone channel will be set to 0.

[0135] The final result is an 8-channel automatic mixer. The threshold feature value controls the number of mixing channels, and together with the similarity, it controls the gain of the microphone channels to achieve automatic microphone management.

[0136] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above description is only a specific embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A mixing calculation method with automatic threshold control, characterized in that, include: Collect the audio signals corresponding to each microphone channel, calculate the RMS value corresponding to the audio signals corresponding to each microphone channel, and track the maximum and minimum values ​​of the RMS values. Based on the maximum and minimum values ​​of the RMS value, the RMS extreme values ​​are obtained, and the noise is tracked according to the RMS extreme values; The zero-crossing rate and signal-to-noise ratio are obtained based on the tracked noise, and the speech energy value is obtained based on the zero-crossing rate and signal-to-noise ratio. The voice energy value is compared with a preset threshold to determine the operating status of each microphone channel; the operating status is either active or inactive. Acquire all target microphone channels that are active and calculate the MFCC features of each target microphone channel. At the same time, determine the target microphone channel with the largest voice energy value as the main microphone channel and the rest as secondary microphone channels. Obtain the MFCC feature similarity between the secondary microphone channel and the primary microphone channel, and obtain the similarity gain based on the MFCC feature similarity; Based on the similarity gain, speech energy value, and RMS value corresponding to the secondary microphone channel, the final gain corresponding to each microphone channel is obtained, and gain control is performed on each microphone channel based on the final gain to realize mixing calculation based on automatic threshold control.

2. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, Acquire the audio signals corresponding to each microphone channel, calculate the RMS value corresponding to the audio signal of each microphone channel, and track the maximum and minimum values ​​of the RMS values, including: Collect the audio signals corresponding to each microphone channel. Where n is the microphone channel number and k is the time point index; Calculate the RMS value of the audio signal corresponding to each microphone channel: ； in, The RMS value of the m-th frame. This refers to the number of data points, specifically 128 points per frame. The maximum and minimum values ​​of the RMS value are tracked as follows: ； ； in, This is the minimum value tracked in the m-th frame. This is the maximum value tracked in frame m. To find the minimum value, To find the maximum value, The RMS value of the (m-1)th frame. Let be the RMS value of the m-L+1th frame, where L is the number of windows tracking the RMS value.

3. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, Based on the maximum and minimum values ​​of the RMS values, the extreme values ​​of RMS are obtained, including: To obtain the maximum and minimum values ​​of the smoothed value: ； ； in, This represents the smoothed maximum value corresponding to the m-th frame. This shows the smoothing minimum value corresponding to the m-th frame. This is the minimum value tracked in the m-th frame. This is the maximum value tracked in frame m. For smoothing coefficients, The maximum smoothed value corresponding to the (m-1)th frame. This is the smoothed minimum value corresponding to the (m-1)th frame; Based on the smoothed maximum and smoothed minimum values, the RMS extreme values ​​are obtained as follows: ； in, is the RMS extreme value of the m-th frame.

4. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, Based on the RMS extreme value tracking noise, including: ； in, The noise tracked in the m-th frame, For the noise tracked in the (m-1)th frame, For tracking coefficients, This is the minimum value tracked in the m-th frame. This is the minimum value tracked in the (m-1)th frame. The minimum value tracked at frame m-L+1 is denoted by , where L is the number of windows tracking the RMS value.

5. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, The zero-crossing rate and signal-to-noise ratio are obtained based on the tracked noise, including: The reference data obtained based on the tracked noise is as follows: ； in, This is the reference data for the m-th frame; This refers to the number of data points, specifically 128 points per frame. For symbolic functions, Let n be the signal of the nth microphone at time k. Let n be the signal of the nth microphone at time k-1. The noise tracked in the (m-1)th frame; The zero-crossing rate is obtained based on the reference data as follows: ； in, Let m be the zero-crossing rate of the m-th frame. Calculate the coefficient for the zero-crossing rate. The zero-crossing rate of the (m-1)th frame; The signal-to-noise ratio is obtained based on the tracked noise: ； in, Let be the signal-to-noise ratio of the m-th frame. The RMS value of the m-th frame. The noise tracked in the m-th frame.

6. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, The speech energy value is obtained based on the zero-crossing rate and the signal-to-noise ratio, including: Based on the zero-crossing rate and signal-to-noise ratio, the characteristic value is obtained as follows: ； in, Let m be the feature value of the m-th frame. As the first weight value, This is the second weight value. As the third weight value, The RMS extreme value of the m-th frame. Let m be the zero-crossing rate of the m-th frame. Let be the signal-to-noise ratio of the m-th frame. The range threshold, For zero-crossing threshold, The signal-to-noise ratio threshold; The speech energy value is obtained by using feature values: ； in, Let be the speech energy value of the m-th frame, 'a' be the first decision parameter, and 'b' be the second decision parameter.

7. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, The voice energy value is compared with a preset threshold to determine the operating status of each microphone channel, including: Determine whether the voice energy value corresponding to the microphone channel is greater than a preset threshold. If so, determine that the microphone channel is in an active state; otherwise, determine that the microphone channel is in an inactive state. Iterate through each microphone channel to obtain the operating status of each microphone channel.

8. The mixing calculation method for automatic threshold control according to claim 1, characterized in that, Obtaining the MFCC feature similarity between the secondary microphone channel and the primary microphone channel includes: ； in, Here, 'ch' represents the MFCC feature similarity, and 'ch' represents the secondary microphone channel whose MFCC feature similarity is calculated with the primary microphone channel. MFCC characteristics of the main microphone channel, This is the MFCC feature for the secondary microphone channel.

9. The mixing calculation method for automatic threshold control according to claim 8, characterized in that, Obtaining a similarity gain based on the MFCC feature similarity includes: ； in, For similarity gain, The RMS value of the main microphone channel in the m-th frame. This represents the RMS value of the secondary microphone channel in the m-th frame.

10. The mixing calculation method for automatic threshold control according to claim 9, characterized in that, Based on the similarity gain, speech energy value, and RMS value corresponding to the secondary microphone channel, the final gain corresponding to each microphone channel is obtained, including: ； in, This represents the final gain of the secondary microphone channel in frame m. This represents the speech energy value of the secondary microphone channel in the m-th frame. This represents the RMS value of the i-th microphone channel in the m-th frame. Let represent the speech energy value of the i-th microphone channel in the m-th frame, and NN represent the total number of active microphone channels.

Images