CN115776626A

CN115776626A - Frequency response calibration method and system of microphone array

Info

Publication number: CN115776626A
Application number: CN202310093147.9A
Authority: CN
Inventors: 曹祖杨; 曹睿颖; 张凯强; 杜子哲
Original assignee: Hangzhou Crysound Electronics Co Ltd
Current assignee: Hangzhou Crysound Electronics Co Ltd
Priority date: 2023-02-10
Filing date: 2023-02-10
Publication date: 2023-03-10
Anticipated expiration: 2043-02-10
Also published as: CN115776626B

Abstract

The invention relates to the technical field of microphone frequency response calibration, in particular to a method and a system for calibrating the frequency response of a microphone array. A frequency response calibration method of a microphone array comprises the following steps that an L1.N value setting module sets an initial value of N to be 1; l2, selecting the Nth microphone to be calibrated; selecting a reference microphone; l3, playing calibration audio, and simultaneously collecting the calibration audio by the microphone to be calibrated and the reference microphone; l4, converting the calibration audio collected by the microphone to be calibrated into a frequency curve to be calibrated by the frequency curve forming module to be calibrated; the reference frequency curve forming module converts the calibration audio collected by the reference microphone into a reference frequency curve; and L5, the frequency response calibration parameter determining module acquires the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated through the frequency curve to be calibrated and the reference frequency curve. The frequency response calibration method and the frequency response calibration system can calibrate the frequency response of the microphone array, so that the microphones in the microphone array have good frequency response consistency.

Description

Frequency response calibration method and system of microphone array

Technical Field

The invention relates to the technical field of microphone frequency response calibration, in particular to a method and a system for calibrating the frequency response of a microphone array.

Background

Microphones are usually arranged in various smart devices in the form of microphone arrays, and due to the influence of materials, assembly and other factors, the frequency response characteristics of each microphone in the same microphone array are quite discrete (i.e. when playing the same audio, the frequency responses generated by each microphone in the same microphone array are different). The frequency response dispersion of the microphone array may cause the signal processing capability of the microphone array to be reduced, and for this reason, frequency response calibration needs to be performed on each microphone in the microphone array. However, in the prior art, no suitable frequency response calibration method is available at present, which can perform frequency response calibration on a microphone array, so that microphones in the microphone array have good frequency response consistency.

Disclosure of Invention

Aiming at the problems in the prior art, the invention provides a method and a system for calibrating the frequency response of a microphone array, which can calibrate the frequency response of the microphone array efficiently and accurately, so that the microphones in the microphone array have good frequency response consistency.

The technical scheme adopted by the invention for solving the technical problem is as follows: a frequency response calibration method of a microphone array comprises the following steps

The L1.N value setting module sets the initial value of N to 1;

l2, selecting the Nth microphone to be calibrated; selecting a reference microphone;

l3, playing calibration audio, and simultaneously acquiring the calibration audio by the microphone to be calibrated and the reference microphone;

l4, converting the calibration audio collected by the microphone to be calibrated into a frequency curve to be calibrated by the frequency curve forming module to be calibrated; a reference frequency curve forming module converts the calibration audio collected by the reference microphone into a reference frequency curve;

l5, the frequency response calibration parameter determining module acquires the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated through the frequency curve to be calibrated and the reference frequency curve, and stores the frequency response calibration parameters of the frequency response calibration filter into the corresponding microphone to be calibrated; the N value setting module adds one to the value of N;

the L6.N value judging module compares the value of N with the total number of the microphones to be calibrated, and when the value of N is greater than the total number of the microphones to be calibrated, the operation is ended; otherwise, L2 is returned.

Preferably, the L5 specifically comprises the following steps

An l51.M value setting unit sets an initial value of M to 1;

l52, the first frequency response value data forming unit discretizes the frequency curve to be calibrated at equal frequency intervals to obtain a group of first frequency response value data, the second frequency response value data forming unit discretizes the reference frequency curve at equal frequency intervals to obtain a group of second frequency response value data, and the frequency intervals of the first frequency response value data are the same as those of the second frequency response value data;

l53, the frequency response difference value data calculation unit calculates a group of frequency response difference value data through the first frequency response value data and the second frequency response value data;

l54, a frequency response difference data d obtaining unit obtains frequency response difference data d with the largest absolute value in the frequency response difference data, and determines a frequency point f corresponding to the frequency response difference data d;

l55, a frequency response difference data d judging unit compares the absolute value of the frequency response difference data d with a first preset threshold, when the absolute value of the frequency response difference data d is smaller than the first preset threshold, the judgment is finished, and an N value setting module adds one to the value of N; entering L56 when the absolute value of the frequency response difference data d is greater than or equal to a first preset threshold;

l56, the central frequency value determining unit takes the frequency point f as the central frequency value of the Mth frequency response calibration filter; the gain value determining unit takes the absolute value of the frequency response difference data d as the gain value of the Mth frequency response calibration filter; the filter type determining unit determines the type of the Mth frequency response calibration filter according to the positive and negative values of the frequency response difference data d; the frequency coverage range value determining unit determines the frequency coverage range value of the Mth frequency response calibration filter;

l57, the latest frequency curve to be calibrated forming unit fits the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the latest frequency curve to be calibrated; the M value setting unit increments the value of M by one and returns to L52.

Preferably, the step of determining the frequency coverage value of the mth frequency response calibration filter by the frequency coverage value determining unit in L56 specifically includes

L561, the initial frequency coverage range value setting subunit sets an initial value for the frequency coverage range value of the Mth frequency response calibration filter;

l562, fitting the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain a frequency curve to be determined by a frequency curve forming subunit;

l563. A difference value operator unit calculates the difference value between the to-be-determined frequency curve and the reference frequency curve; a comparison difference value setting subunit sets the corresponding difference value as a comparison difference value;

the S value setting subunit sets the count value S to 1;

l565, the frequency coverage range value adjusting subunit adjusts the frequency coverage range value of the Mth frequency response calibration filter;

l566. Undetermined frequency curve forming subunit fits the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain a new undetermined frequency curve;

l567, a difference value operator unit calculates the difference value between the new undetermined frequency curve and the reference frequency curve; the difference value judging subunit compares the difference value obtained by the latest calculation of the difference value operator unit with the comparison difference value, when the difference value is smaller than the comparison difference value, the comparison difference value setting subunit sets the corresponding difference value as the comparison difference value, the frequency coverage range value recording subunit records the corresponding frequency coverage range value, and the L564 is returned; otherwise, the S value setting subunit adds one to the count value S and enters L568;

the L568.S value judgment subunit compares the count value S with a second preset threshold value, and returns to L565 when the count value S is smaller than the second preset threshold value; and when the counting value S is equal to a second preset threshold value, taking the frequency coverage range value recorded by the frequency coverage range value recording subunit as the frequency coverage range value of the Mth frequency response calibration filter.

Preferably, the step of calculating the difference value between L563 and L567 specifically includes

L371, the undetermined frequency response data forming subunit discretizes the undetermined frequency curve at equal frequency intervals to obtain a group of undetermined frequency response data, the reference frequency response data forming subunit discretizes the reference frequency curve at equal frequency intervals to obtain a group of reference frequency response data, and the frequency intervals of the undetermined frequency response data are the same as the frequency intervals of the reference frequency response data;

l372, the initial difference value calculation data acquisition subunit performs one-to-one corresponding subtraction on data point values in the undetermined frequency response data and corresponding data point values in the reference frequency response data to obtain a group of initial difference value calculation data;

l373. A final difference value calculation data acquisition subunit multiplies the data point values in the initial difference value calculation data and the data point values in the weight data in a one-to-one correspondence manner to obtain a group of final difference value calculation data;

and L374, performing root mean square calculation on the data point values in the final difference value calculation data by the difference value calculation secondary unit to obtain the difference values.

Preferably, the step of obtaining weight data in L373 includes

L3731. The weight curve forming subunit forms a weight curve through a normal distribution formula 1);

1）

wherein ,

is a normal distribution function, parameter

Is 1, parameter

Is the value of the frequency point f, and

the minimum value of the value range is the minimum frequency point value in the reference frequency response data,

the maximum value of the value range is the maximum frequency point value in the reference frequency response data;

l3732, the weight data obtaining subunit discretizes the equal frequency interval of the weight curve to obtain a set of weight data, and the frequency interval of the weight data is the same as the frequency interval of the reference frequency response data.

A frequency response calibration system for a microphone array includes

The calibration audio processing device comprises a to-be-calibrated frequency curve forming module, a calibration processing module and a calibration processing module, wherein the to-be-calibrated frequency curve forming module is used for converting calibration audio collected by a to-be-calibrated microphone into a to-be-calibrated frequency curve;

a reference frequency curve forming module for converting the calibration audio collected by the reference microphone into a reference frequency curve;

the frequency response calibration parameter determining module is used for acquiring the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated through the frequency curve to be calibrated and the reference frequency curve, and storing the frequency response calibration parameters of the frequency response calibration filter into the corresponding microphone to be calibrated;

the N value setting module is used for setting the initial value of N to be 1, and adding one to the value of N after the frequency response calibration parameter of the frequency response calibration filter is stored in the corresponding microphone to be calibrated by the frequency response calibration parameter determining module;

and the N value judging module is used for comparing the value of N with the total number of the microphones to be calibrated.

Preferably, the frequency response calibration parameter determination module comprises

The first frequency response value data forming unit is used for discretizing the frequency curve to be calibrated at equal frequency intervals to obtain a group of first frequency response value data;

the second frequency response value data forming unit is used for discretizing the reference frequency curve at equal frequency intervals to obtain a group of second frequency response value data;

the frequency response difference value data calculation unit is used for calculating a group of frequency response difference value data through the first frequency response value data and the second frequency response value data;

the device comprises a frequency response difference data d obtaining unit, a frequency response difference data determining unit and a frequency response difference data calculating unit, wherein the frequency response difference data d obtaining unit is used for obtaining the frequency response difference data d with the largest absolute value in the frequency response difference data and determining a frequency point f corresponding to the frequency response difference data d;

the frequency response difference data d judging unit is used for comparing the absolute value of the frequency response difference data d with a first preset threshold;

the center frequency value determining unit is used for taking the frequency point f as the center frequency value of the Mth frequency response calibration filter;

a gain value determining unit, for using the absolute value of the frequency response difference data d as the gain value of the Mth frequency response calibration filter;

the filter type determining unit is used for determining the type of the Mth frequency response calibration filter according to the positive and negative values of the frequency response difference data d;

a frequency coverage value determination unit for determining a frequency coverage value of the mth frequency response calibration filter;

the latest frequency curve forming unit to be calibrated is used for fitting the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the latest frequency curve to be calibrated;

and the M value setting unit is used for setting the initial value of M to be 1 and adding one to the value of M after the latest frequency curve to be calibrated is obtained by the latest frequency curve to be calibrated forming unit.

Preferably, the frequency coverage value determination unit includes

The initial frequency coverage range value setting subunit is used for setting an initial value for the frequency coverage range value of the Mth frequency response calibration filter;

the frequency curve to be determined forming subunit is used for fitting the frequency curve to be calibrated and the frequency response generated by the Mth frequency response calibration filter to obtain a frequency curve to be determined;

the difference value operator unit is used for calculating the difference value between the undetermined frequency curve and the reference frequency curve;

the difference value judgment subunit is used for comparing the difference value obtained by the calculation of the difference value operator unit with the comparison difference value;

a comparison difference value setting subunit, configured to set the corresponding difference value as the comparison difference value when the difference value is obtained by the first calculation or when the difference value is smaller than the comparison difference value;

an S value setting subunit, configured to set a value of the count value S;

a frequency coverage range value adjusting subunit, configured to adjust a frequency coverage range value of the mth frequency response calibration filter;

the frequency coverage range value recording subunit is used for recording the corresponding frequency coverage range value in the frequency coverage range value adjusting subunit when the difference value is smaller than the comparison difference value;

and the S value judging subunit is used for comparing the counting value S with a second preset threshold value.

Preferably, the disparity value operator unit comprises

The undetermined frequency response data forming subunit is used for discretizing the undetermined frequency curve at equal frequency intervals to obtain a group of undetermined frequency response data;

the reference frequency response data forming subunit is used for discretizing the reference frequency curve at equal frequency intervals to obtain a group of reference frequency response data;

the initial difference value calculation data acquisition subunit is used for carrying out one-to-one corresponding subtraction on data point values in the undetermined frequency response data and corresponding data point values in the reference frequency response data so as to obtain a group of initial difference value calculation data;

a final difference value calculation data acquisition subunit, configured to perform one-to-one corresponding multiplication on the data point values in the initial difference value calculation data and the data point values in the weight data to obtain a set of final difference value calculation data;

and the difference value calculation subunit is used for performing root-mean-square calculation on the data point values in the final difference value calculation data to obtain the difference values.

Preferably, the difference value operator unit further includes

A weight curve forming subunit for forming a weight curve by a normal distribution formula 1);

1）

wherein ,

is a normal distribution function, parameter

Is 1, parameter

Is the value of the frequency point f, and

the weight data acquisition subunit is used for discretizing the equal frequency intervals of the weight curve to obtain a group of weight data, and the frequency intervals of the weight data are the same as the frequency intervals of the reference frequency response data.

Advantageous effects

According to the frequency response calibration method and system provided by the embodiment of the invention, the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated can be obtained through the frequency curve to be calibrated and the reference frequency curve, so that each microphone in the microphone array can be calibrated to be the same as the reference microphone one by one, and when all the microphones are calibrated to be the same as the reference microphone, the microphones in the corresponding microphone array have good frequency response consistency;

the number of the frequency response calibration filters needed by the microphone to be calibrated and the frequency response calibration parameter corresponding to each frequency response calibration filter can be determined through the frequency response difference data d, and the frequency response of the microphone to be calibrated can be calibrated to be very close to the frequency response of the reference microphone through the frequency response calibration filters corresponding to the number and the frequency response calibration parameters.

Drawings

FIG. 1 is a schematic diagram of a frequency response calibration system according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of a frequency response calibration parameter determination module according to an embodiment of the present invention;

FIG. 3 is a diagram of a frequency coverage value determination unit according to an embodiment of the present invention;

FIG. 4 is a diagram illustrating a difference value operator unit according to an embodiment of the present invention;

FIG. 5 is a diagram illustrating a weight curve according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of an embodiment of a microphone array without frequency response calibration;

fig. 7 is a schematic diagram of a microphone array with frequency response calibration according to an embodiment of the invention.

Detailed Description

The technical scheme of the invention is further explained by the specific implementation mode in combination with the attached drawings.

Example 1: a frequency response calibration method of a microphone array comprises the following steps

The l1.N value setting module sets an initial value of N to 1.

And L2, selecting the Nth microphone to be calibrated, selecting the 1 st microphone to be calibrated when the value of N is 1, selecting the 2 nd microphone to be calibrated when the value of N is 2, and selecting one microphone to be calibrated each time in the step L2, wherein the N value of the step L2 is gradually increased, wherein N can be understood as the number of the microphone to be calibrated. A reference microphone is selected, the reference microphone being a fixed microphone, and step L2 selects the same microphone as the reference microphone each time. When the frequency response characteristics of all the microphones to be calibrated are calibrated to be the same as the frequency response characteristics of the reference microphone, the aim of calibrating the frequency response characteristics of all the microphones to be calibrated to be consistent is fulfilled.

In this embodiment, step L2 may be to select the 1 st microphone to be calibrated.

And L3, playing calibration audio, and simultaneously acquiring the calibration audio by the microphone to be calibrated and the reference microphone. The calibration audio played each time in step L3 is fixed.

L4, converting the calibration audio collected by the microphone to be calibrated into a frequency curve to be calibrated by the frequency curve forming module to be calibrated; the reference frequency curve forming module converts the calibration audio collected by the reference microphone into a reference frequency curve.

And L5, the frequency response calibration parameter determining module acquires the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated through the frequency curve to be calibrated and the reference frequency curve, and stores the frequency response calibration parameters of the frequency response calibration filter into the corresponding microphone to be calibrated. A frequency response calibration filter has a set of frequency response calibration parameters that specifically include a type of filter, a center frequency value of the filter, a gain value of the filter, and a frequency coverage value of the filter. One microphone to be calibrated may not need to be provided with a frequency response calibration filter, may only need to be provided with one frequency response calibration filter, and may also need to be provided with a plurality of frequency response calibration filters.

For example, case 1: when the frequency curve to be calibrated obtained in step L4 is very close to the reference frequency curve, the frequency response characteristic of the microphone to be calibrated does not need to be calibrated, and the number of the frequency response calibration filters used by the microphone to be calibrated is 0, and correspondingly, the frequency response calibration parameters of the frequency response calibration filters are not available, so that the frequency response calibration parameters of the frequency response calibration filters do not need to be stored in the corresponding microphone to be calibrated. It can also be understood here that the frequency response calibration parameter determination module stores blank data packets (filter type 0, filter center frequency value 0, filter gain value 0, and filter frequency coverage value 0) into the corresponding microphones to be calibrated.

As another example, case 2: when the frequency curve to be calibrated acquired in step L4 has a small difference from the reference frequency curve, the frequency curve to be calibrated of the microphone to be calibrated may be calibrated to be the same as the reference frequency curve of the reference microphone by a small number of frequency response calibration filters (e.g., 1). Then the number of the frequency response calibration filters used by the microphone to be calibrated is 1, and the frequency response calibration filters correspond to 1 set of frequency response calibration parameters. And the frequency response calibration parameter determining module stores the corresponding frequency response calibration parameters of the corresponding frequency response calibration filter into the corresponding microphone to be calibrated in the form of data packets.

As another example, case 3: when the frequency curve to be calibrated acquired in step L4 is different from the reference frequency curve, the frequency curve to be calibrated of the microphone to be calibrated may be calibrated to be the same as the reference frequency curve of the reference microphone by a plurality of frequency response calibration filters (e.g., 4). The number of frequency response calibration filters used by the microphone to be calibrated is 4, and each frequency response calibration filter has 1 set of frequency response calibration parameters. The frequency response calibration parameter determining module sequentially stores the corresponding frequency response calibration parameter of the 1 st frequency response calibration filter, the corresponding frequency response calibration parameter of the 2 nd frequency response calibration filter, the corresponding frequency response calibration parameter of the 3 rd frequency response calibration filter and the corresponding frequency response calibration parameter of the 4 th frequency response calibration filter into the corresponding microphones to be calibrated in the form of four data packets.

After the frequency response calibration parameter determining module stores the frequency response calibration parameter of the frequency response calibration filter in the corresponding microphone to be calibrated, the current frequency response calibration operation of the microphone to be calibrated is completed, at this time, the N value setting module adds one to the value of N, and when the value of N in the step L2 is 1, the value of N at this time becomes 2.

Assuming that the microphone array of this embodiment has 150 microphones to be calibrated, since the value of N is 2 and is less than 150, it is necessary to return to step L2 to perform the frequency response calibration operation of the 2 nd microphone to be calibrated. If the value of N at this time is 151, the frequency response calibration operation of the entire microphone array is finished because the value of N is greater than 150.

According to the frequency response calibration method, the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated can be acquired through the frequency curve to be calibrated and the reference frequency curve, so that the microphones in the microphone array can be calibrated to be the same as the reference microphone one by one, and when all the microphones are calibrated to be the same as the reference microphone, the microphones in the corresponding microphone array have good frequency response consistency.

Further, the L5 specifically includes the following steps

L51.M value setting unit sets the initial value of M to 1.

And L52, discretizing the equal frequency intervals of the frequency curve to be calibrated by the first frequency response value data forming unit to obtain a group of first frequency response value data, discretizing the equal frequency intervals of the reference frequency curve by the second frequency response value data forming unit to obtain a group of second frequency response value data, wherein the frequency intervals of the first frequency response value data are the same as the frequency intervals of the second frequency response value data.

Assume that the frequency range of the frequency curve to be calibrated and the frequency curve of the reference frequency curve in this embodiment is 100Hz to 10000Hz.

The first frequency response value data forming unit discretizes the frequency curve to be calibrated at a frequency interval of 10Hz, and the first frequency response value data comprise a frequency response value a1 at a frequency point 100Hz, a frequency response value a2 at a frequency point 110Hz, a frequency response value a3 at a frequency point 120Hz, a frequency response value a4 \8230ata frequency point 130Hz, and a frequency response value an at a frequency point 10000Hz. Namely, the first frequency response value data are a1, a2, a3 and a4 \8230, an.

The second frequency response value data forming unit discretizes the reference frequency curve at a frequency interval of 10Hz, and the second frequency response value data comprise a frequency response value b1 of the reference frequency curve at a frequency point of 100Hz, a frequency response value b2 at a frequency point of 110Hz, a frequency response value b3 at a frequency point of 120Hz, a frequency response value b4 at a frequency point of 130Hz, 8230, and a frequency response value bn at a frequency point of 10000Hz. Namely, the second frequency response value data are b1, b2, b3, b4 \8230, 8230bn.

And L53, the frequency response difference value data calculation unit calculates a group of frequency response difference value data through the first frequency response value data and the second frequency response value data. Specifically, the frequency response value of each frequency point in the first frequency response value data is subtracted from the frequency response value of the corresponding frequency point in the second frequency response value data. For example, the frequency response difference data in the embodiment are a1-b1, a2-b2, a3-b3, a4-b4 \8230; an-bn.

And L54, acquiring the frequency response difference data d with the maximum absolute value in the frequency response difference data by a frequency response difference data d acquisition unit, and determining a frequency point f corresponding to the frequency response difference data d.

In this embodiment, the frequency response difference data points with the largest absolute value in the frequency response difference data are a57-b57, and a57-b57 are the frequency response difference data d, and the frequency point f corresponding to the frequency response difference data d is 660Hz. And the frequency point f is the place with the maximum frequency response difference between the frequency curve to be calibrated and the reference frequency curve. In most cases, there is only one frequency response difference data d, and if there are two or more frequency response difference data d, the first frequency response difference data d is selected (i.e., the frequency response difference data d with the smallest frequency point f is selected).

L55, comparing the absolute value of the frequency response difference data d with a first preset threshold by a frequency response difference data d judging unit, finishing when the absolute value of the frequency response difference data d is smaller than the first preset threshold, and adding one to the value of N by an N value setting module; and entering L56 when the absolute value of the frequency response difference data d is greater than or equal to a first preset threshold value.

Assume that the first preset threshold is 5dB in this embodiment.

When the frequency response difference data d (i.e. a57-b 57) is 2dB, which is less than 5dB, it means that the frequency response difference between the frequency curve to be calibrated and the reference frequency curve is not more than 2dB, i.e. it means that the frequency curve to be calibrated and the reference frequency curve are very close. At this time, the corresponding microphone to be calibrated does not need to be subjected to frequency response calibration, and therefore, the frequency response calibration parameters of the frequency response calibration filter do not need to be determined, and the process is ended directly (corresponding to case 1).

When the frequency response difference data d (i.e., a57-b 57) is 8dB, which is greater than 5dB, it indicates that the frequency curve to be calibrated of the microphone to be calibrated needs to be calibrated, and then step L56 is performed.

L56, the central frequency value determining unit takes the frequency point f as the central frequency value of the Mth frequency response calibration filter; the gain value determining unit takes the absolute value of the frequency response difference data d as the gain value of the Mth frequency response calibration filter; the filter type determining unit determines the type of the Mth frequency response calibration filter according to the positive and negative values of the frequency response difference data d; a frequency coverage value determination unit determines a frequency coverage value of the Mth frequency response calibration filter.

As shown in step L51, the value M is 1, that is, the frequency response calibration parameter of the 1 st frequency response calibration filter (specifically, the value of the center frequency of the filter, the value of the gain of the filter, the type of the filter, and the value of the frequency coverage of the filter) needs to be determined.

The central frequency value of the filter can be determined through the frequency point f, and the frequency point f is 660Hz in this embodiment, so the central frequency value of the filter is 660Hz. The gain value of the filter is determined by the absolute value of the frequency response difference data d, which is 8dB in this embodiment, so the gain value of the filter is 8dB. The type of the filter can be directly determined by the positive and negative values of the frequency response difference data d, and when the frequency response difference data d is a positive value, a Notch filter is adopted (the Notch filter can generate a downward concave waveform); when the frequency response difference data d has a negative value, a Peak filter is used (the Peak filter can generate an upwardly convex waveform). In this embodiment, the frequency response difference data d is 8dB and is a positive value, so the filter is a Notch filter.

When the frequency response calibration parameters of the 1 st frequency response calibration filter are all determined, the process proceeds to step L57.

The 1 st frequency response calibration filter can generate a frequency response through the frequency response calibration parameters. At this time, the frequency response generated by the 1 st frequency response calibration filter is fitted to the current frequency curve to be calibrated, so as to obtain a latest frequency curve to be calibrated. The latest frequency curve to be calibrated is used as the frequency curve to be calibrated in the next step L52. And the M value setting unit increments the value of M by one, at which time the M value becomes 2.

Step L52 is re-entered, at which time the frequency range of the frequency curve to be calibrated and the frequency curve of the reference is still 100Hz to 10000Hz.

The first frequency response value data forming unit discretizes the frequency curve to be calibrated (the latest frequency curve to be calibrated in the step L57) at frequency intervals of 10Hz, and the first frequency response value data comprises a frequency response value c1 of the frequency curve to be calibrated at a frequency point of 100Hz, a frequency response value c2 at a frequency point of 110Hz, a frequency response value c3 at a frequency point of 120Hz, and a frequency response value c4 at a frequency point of 130Hz, 8230, a frequency response value cn at a frequency point of 10000Hz. Namely, the first frequency response value data are c1, c2, c3, c4 \8230;. Cn.

The second frequency response value data forming unit discretizes the reference frequency curve at frequency intervals of 10Hz, and the second frequency response value data are b1, b2, b3 and b4 \8230 \8230andbn.

And L53, the frequency response difference value data calculation unit calculates a group of frequency response difference value data through the first frequency response value data and the second frequency response value data. Specifically, the frequency response value of each frequency point in the first frequency response value data is subtracted from the frequency response value of the corresponding frequency point in the second frequency response value data. At this time, the frequency response difference data in this embodiment are c1-b1, c2-b2, c3-b3, c4-b4 \8230, cn-bn.

In this embodiment, if the frequency response difference data point with the largest absolute value in the frequency response difference data is c83-b83, c83-b83 is the frequency response difference data d, and the frequency point f corresponding to the frequency response difference data d is 920Hz.

In this embodiment, the first predetermined threshold is still 5dB.

When the frequency response difference data d (i.e., c83-b 83) is-4 dB, the absolute value of the frequency response difference data d is less than 5dB, which indicates that the frequency curve to be calibrated is very close to the reference frequency curve. And the corresponding microphone to be calibrated does not need to be subjected to further frequency response calibration, and the calibration is directly finished at the moment. I.e. the corresponding microphone to be calibrated only needs one frequency response calibration filter for frequency response calibration (corresponding to case 2).

When the frequency response difference data d (i.e., c83-b 83) is-7 dB, the absolute value of the frequency response difference data d is greater than 5dB, which indicates that the current frequency curve to be calibrated of the microphone to be calibrated needs to be further calibrated, and then the process proceeds to step L56.

From the last step L57, the value M at this time is 2, that is, the frequency response calibration parameter of the 2 nd frequency response calibration filter needs to be determined at this time.

The central frequency value of the filter can be determined through a frequency point f, and at the moment, the frequency point f is 920Hz, so that the central frequency value of the filter is 920Hz. The gain value of the filter is determined by the absolute value of the frequency response difference data d, which is 7dB, so the gain value of the filter is 7dB. The type of the filter can be directly determined by the positive and negative values of the frequency response difference data d, and the frequency response difference data d is-7 dB and is a negative value, so that the type of the filter is a Peak filter.

When the frequency response calibration parameters of the 2 nd frequency response calibration filter are all determined, the process proceeds to step L57.

The 2 nd frequency response calibration filter can generate a frequency response through the frequency response calibration parameters. At this time, the frequency response generated by the 2 nd frequency response calibration filter is fitted to the current frequency curve to be calibrated, so that a latest frequency curve to be calibrated can be obtained again. The latest frequency curve to be calibrated is used as the frequency curve to be calibrated in the next step L52. And the M value setting unit increments the value of M by one, at which time the M value becomes 3.

And step L52 is re-entered, and the above steps are repeated until the absolute value of the frequency response difference data d is smaller than the first preset threshold in the latest step L55, and the process is ended. At this time, if the value of M is large, M-1 frequency response calibration filters are adopted by the corresponding microphone to be calibrated. For example, if the value of M is 4, then 3 frequency response calibration filters are actually used for the corresponding microphone to be calibrated (corresponding to case 3).

The frequency response calibration method of the embodiment can determine the number of the frequency response calibration filters required to be used by the microphone to be calibrated and the frequency response calibration parameters (including the center frequency value, the gain value and the type) corresponding to each frequency response calibration filter through the frequency response difference data d, and can calibrate the frequency response corresponding to the microphone to be calibrated to be very close to the frequency response of the reference microphone through the corresponding number of the frequency response calibration filters and the corresponding frequency response calibration parameters. The method comprises the steps of firstly calibrating the position with the largest difference between a frequency curve to be calibrated and a reference frequency curve through a1 st frequency response calibration filter, then calibrating the position with the second largest difference between the frequency curve to be calibrated and the reference frequency curve through a2 nd frequency response calibration filter, \8230, and finally calibrating the position with the smallest difference between the frequency curve to be calibrated and the reference frequency curve through an nth frequency response calibration filter.

Further, the step of the L56 frequency coverage value determining unit determining the frequency coverage value of the mth frequency response calibration filter specifically includes

L561, the initial frequency coverage value setting subunit sets an initial value for the frequency coverage value of the mth frequency response calibration filter.

Assuming that the value M is 2, the initial frequency coverage value setting subunit sets an initial value to the frequency coverage value of the 2 nd frequency response calibration filter, and the initial value of the frequency coverage value may be 0.5.

And L562, the undetermined frequency curve forming subunit fits the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the undetermined frequency curve.

At this time, the frequency response calibration parameters of the 2 nd frequency response calibration filter are all determined. For example, the filter is of the Peak type, the center frequency value of the filter is 920Hz, the gain value of the filter is 7dB, and the frequency coverage value of the filter is 0.5. The 2 nd frequency response calibration filter can generate a frequency response according to the frequency response calibration parameters, and a frequency curve to be determined can be obtained by fitting the current frequency curve to be calibrated with the frequency response generated by the 2 nd frequency response calibration filter.

L563. A difference value operator unit calculates the difference value between the frequency curve to be determined and the reference frequency curve; the contrast difference value setting subunit sets the corresponding difference value as a contrast difference value.

The difference value operator unit is used for calculating the difference between the to-be-determined frequency curve and the reference frequency curve, and when the calculated difference value is larger, the difference between the to-be-determined frequency curve and the reference frequency curve is larger, and when the calculated difference value is smaller, the difference between the to-be-determined frequency curve and the reference frequency curve is smaller. Step L563 is configured to calculate a difference between the undetermined frequency curve and the reference frequency curve when the frequency coverage value is 0.5 as the initial value, and use the calculated difference value as a comparison difference value.

The l564.S value setting subunit sets the count value S to 1.

And L565, the frequency coverage range value adjusting subunit adjusts the frequency coverage range value of the Mth frequency response calibration filter.

Step L565 is used to adjust the frequency coverage value of the 2 nd frequency response calibration filter, for example, if the amplitude of each adjustment is 0.1, then the frequency coverage value of the filter at this time becomes 0.6.

And L566. The undetermined frequency curve forming subunit fits the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain a new undetermined frequency curve.

At this time, the frequency response calibration parameter of the 2 nd frequency response calibration filter is that the type of the filter is a Peak filter, the central frequency value of the filter is 920Hz, the gain value of the filter is 7dB, and the frequency coverage value of the filter is 0.6. The 2 nd frequency response calibration filter can generate a new frequency response according to the frequency response calibration parameter, and a new frequency curve to be determined can be obtained by fitting the current frequency curve to be calibrated with the new frequency response generated by the 2 nd frequency response calibration filter.

L567, a difference value operator unit calculates the difference value between the new undetermined frequency curve and the reference frequency curve; the difference value judging subunit compares the difference value obtained by the latest calculation of the difference value operator unit with the comparison difference value, when the difference value is smaller than the comparison difference value, the comparison difference value setting subunit sets the corresponding difference value as the comparison difference value, the frequency coverage range value recording subunit records the corresponding frequency coverage range value, and the L564 is returned; otherwise, the S value setting subunit increments the count value S by one and proceeds to L568.

Step L567 is configured to calculate a difference between the undetermined frequency curve and the reference frequency curve when the frequency coverage value is 0.6, and obtain a new difference value.

When the new difference value is smaller than the original comparison difference value, it indicates that the frequency coverage value is better at 0.6 (i.e. when the frequency coverage value is 0.6, the difference between the undetermined frequency curve and the reference frequency curve obtained by fitting becomes smaller). Then the new difference value is taken as the control difference value at this point and the frequency coverage value corresponding to the control difference value is recorded (i.e., 0.6) and returned to L564.

Re-entering the l564.S value setting sub-unit sets the count value S to 1.

And L565, the frequency coverage range value adjusting subunit adjusts the frequency coverage range value of the 2 nd frequency response calibration filter, wherein the frequency coverage range value of the filter becomes 0.7.

And L566. The undetermined frequency curve forming subunit fits the frequency curve to be calibrated with the frequency response generated by the 2 nd frequency response calibration filter to obtain a new undetermined frequency curve.

And L567, calculating the difference value of the new undetermined frequency curve and the reference frequency curve by a difference value operator unit.

Step L567 is configured to calculate a difference between the new undetermined frequency curve and the reference frequency curve when the frequency coverage value is 0.7, and obtain a new difference value.

When the new difference value is greater than or equal to the comparison difference value, which indicates that the frequency coverage value is 0.7, the difference between the undetermined frequency curve and the reference frequency curve obtained by fitting becomes larger. At this time, the value of the count value S is incremented by one, the count value S becomes 2, and the process proceeds to step L568.

The L568.S value judgment subunit compares the count value S with a second preset threshold value, and returns to the L565 when the count value S is smaller than the second preset threshold value; and when the counting value S is equal to a second preset threshold value, taking the frequency coverage range value recorded by the frequency coverage range value recording subunit as the frequency coverage range value of the Mth frequency response calibration filter.

Assume that the second preset threshold is 3 in this embodiment. The count value S at this time is 2 and is smaller than the second preset threshold value, and therefore, the routine returns to L565.

And re-entering the step L565. The frequency coverage range value adjusting subunit adjusts the frequency coverage range value of the 2 nd frequency response calibration filter, wherein the frequency coverage range value of the filter becomes 0.8.

Step L567 is configured to calculate a difference between the new undetermined frequency curve and the reference frequency curve when the frequency coverage value is 0.8, and obtain a new difference value.

When the new difference value is greater than or equal to the comparison difference value, which indicates that the frequency coverage value is 0.8, the difference between the undetermined frequency curve and the reference frequency curve obtained by fitting becomes larger. At this time, the value of the count value S is incremented by one, the count value S becomes 3, and the process proceeds to step L568.

At this time, if the count value S is 3 and is equal to the second preset threshold, the frequency coverage value (i.e., 0.6) recorded by the frequency coverage value recording subunit is used as the frequency coverage value of the 2 nd frequency response calibration filter, that is, if the frequency coverage value is continuously adjusted for many times (the number of times is equal to the second preset threshold), and the difference values between the calculated undetermined frequency curve (obtained by fitting the current frequency curve to be calibrated and the frequency response generated by the 2 nd frequency response calibration filter) and the reference frequency curve are all greater than or equal to the comparison difference value, the frequency coverage value currently recorded by the frequency coverage value recording subunit is used as the frequency coverage value of the 2 nd frequency response calibration filter.

The frequency response calibration method of the embodiment can automatically determine the frequency coverage value of the corresponding frequency response calibration filter, and the determined frequency coverage value can prevent a difference value between a latest frequency curve to be calibrated and a reference frequency curve, which are formed by fitting the frequency response generated by the corresponding frequency response calibration filter and the frequency curve to be calibrated, from being smaller.

Further, the step of calculating the difference between the L563 and L567 specifically includes

And L371, performing equal-frequency interval discretization on the undetermined frequency curve by the undetermined frequency response data forming subunit to obtain a group of undetermined frequency response data, performing equal-frequency interval discretization on the reference frequency curve by the reference frequency response data forming subunit to obtain a group of reference frequency response data, wherein the frequency interval of the undetermined frequency response data is the same as that of the reference frequency response data.

The frequency ranges of the undetermined frequency curve and the reference frequency curve of the embodiment are 100Hz to 10000Hz.

The undetermined frequency response data forming subunit discretizes the undetermined frequency curve at a frequency interval of 20Hz, and the undetermined frequency response data comprise a frequency response value e1 of the undetermined frequency curve at a frequency point of 100Hz, a frequency response value e2 at a frequency point of 120Hz, a frequency response value e3 at a frequency point of 140Hz, a frequency response value e4 at a frequency point of 160Hz, 8230and a frequency response value en at a frequency point of 10000Hz. Namely undetermined frequency response value data are e1, e2, e3 and e4 \8230, and en 8230.

The reference frequency response data forming subunit discretizes the reference frequency curve at 20Hz frequency interval, and the reference frequency response data comprises a frequency response value f1 of the reference frequency curve at 100Hz, a frequency response value f2 at 120Hz, a frequency response value f3 at 140Hz, a frequency response value f4 at 160Hz, 8230, a frequency response value fn at 10000Hz. Namely, the reference frequency response data are f1, f2, f3 and f4 \8230, 8230and fn.

And L372, the initial difference value calculation data acquisition subunit performs one-to-one corresponding subtraction on the data point values in the undetermined frequency response data and the corresponding data point values in the reference frequency response data to obtain a group of initial difference value calculation data.

Specifically, the frequency response value of each frequency point in the undetermined frequency response data is subtracted from the frequency response value of the corresponding frequency point in the reference frequency response data. For example, the initial difference value calculation data in this embodiment are e1-f1, e2-f2, e3-f3, e4-f4 \8230, en-fn.

And L373. The final difference value calculation data acquisition subunit multiplies the data point values in the initial difference value calculation data and the data point values in the weight data in a one-to-one correspondence manner to obtain a group of final difference value calculation data.

Wherein the step of obtaining the weight data in L373 comprises

L3731. A weight curve forming subunit forms a weight curve through a normal distribution formula 1);

1）

wherein ,

is a normal distribution function, parameter

Is 1, parameter

Is the value of the frequency point f, and

the maximum value of the value range is the maximum frequency point value in the reference frequency response data. In this embodiment, the frequency point f may be 660Hz, so the parameter

And may be 660Hz. The minimum frequency point value in the reference frequency response data is 100Hz, and the maximum frequency point value in the reference frequency response data is 10000Hz, so the method has the advantages of high accuracy, low cost and high reliability

The minimum value of the value range is 100Hz, and the maximum value is 10000Hz. Specifically, the weighting curve of the present embodiment may be as shown in fig. 5.

The weight data acquisition subunit discretizes the weight curve at 20Hz frequency interval, and the weight data includes weight coefficient value g1 of the weight curve at 100Hz, weight coefficient value g2 at 120Hz, weight coefficient value g3 at 140Hz, weight coefficient value g4 at 160Hz, 82308270 and weight coefficient value gn at 10000Hz. Namely, the weight data are g1, g2, g3, g4 \8230, 8230gng.

Step L373 is specifically to multiply the frequency response value of each frequency point in the initial difference value calculation data by the weight coefficient value of the corresponding frequency point in the weight data. For example, the final difference value calculation data in this example are (e 1-f 1) · g1, (e 2-f 2) · g2, (e 3-f 3) · g3, (e 4-f 4) · g4 \8230; (en-fn) · g5.

And L374, performing root mean square calculation on the data point values in the final difference value calculation data by the difference value calculation secondary unit to obtain the difference values. Specifically, each data point value in the final difference value calculation data is squared, then all squared values are averaged, and finally the average value is rooted to obtain the difference value.

In the frequency response calibration method of the embodiment, when the difference value between the undetermined frequency curve and the reference frequency curve is calculated, the weight curve is introduced, the setting of the weight curve enables the difference value between the undetermined frequency curve and the reference frequency curve at a place with a large difference to be larger, and the difference value at a place with a small difference to be smaller, so that the calculation result of the difference value is more reasonable (namely, the difference of the frequency response values on the frequency band where the frequency response calibration filter is arranged is strengthened, and the frequency response difference on other frequency bands is weakened), and finally, the parameter (namely, the frequency coverage range value) of the frequency response calibration filter is set more accurately.

As shown in fig. 6, when the microphone array is not frequency-response-calibrated, the frequency response values of the respective microphones at the frequency point of 1kHz are very discrete. As shown in fig. 7, after the frequency response of the microphone array is calibrated by the frequency response calibration method of this embodiment, the frequency response values of the microphones at the frequency point of 1kHz are very consistent. The frequency response calibration method of the microphone array in the embodiment of the invention can efficiently and accurately calibrate the frequency response of the microphone array, so that the microphones in the microphone array have good frequency response consistency.

Example 2: a frequency response calibration system of a microphone array is applied to the frequency response calibration method in embodiment 1, and specifically includes, as shown in fig. 1, a frequency curve to be calibrated forming module, a reference frequency curve forming module, a frequency response calibration parameter determining module, an N value setting module, and an N value determining module.

The frequency curve forming module to be calibrated is used for converting the calibration audio collected by the microphone to be calibrated into the frequency curve to be calibrated. The reference frequency curve forming module is used for converting the calibration audio collected by the reference microphone into a reference frequency curve. The frequency response calibration parameter determining module is used for acquiring the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated through the frequency curve to be calibrated and the reference frequency curve, and storing the frequency response calibration parameters of the frequency response calibration filter into the corresponding microphone to be calibrated. The N value setting module is used for setting the initial value of N to be 1, and adding one to the value of N after the frequency response calibration parameter determining module stores the frequency response calibration parameter of the frequency response calibration filter into the corresponding microphone to be calibrated. The N value judging module is used for comparing the value of N with the total number of the microphones to be calibrated.

The frequency response calibration system of the embodiment can acquire the frequency response calibration parameters of the frequency response calibration filter of the microphone to be calibrated through the frequency curve to be calibrated and the reference frequency curve, and further can calibrate each microphone in the microphone array one by one to be the same as the reference microphone, and when all the microphones are calibrated to be the same as the reference microphone, the microphones in the corresponding microphone array have good frequency response consistency.

Further, as shown in fig. 2, the frequency response calibration parameter determining module includes a first frequency response value data forming unit, a second frequency response value data forming unit, a frequency response difference value data calculating unit, a frequency response difference value data d obtaining unit, a frequency response difference value data d determining unit, a center frequency value determining unit, a gain value determining unit, a filter type determining unit, a frequency coverage value determining unit, a latest frequency curve to be calibrated forming unit, and an M value setting unit.

The first frequency response value data forming unit is used for discretizing the frequency curve to be calibrated at equal frequency intervals to obtain a group of first frequency response value data. The second frequency response value data forming unit is used for discretizing the reference frequency curve at equal frequency intervals to obtain a group of second frequency response value data. The frequency response difference value data calculation unit is used for calculating a group of frequency response difference value data through the first frequency response value data and the second frequency response value data. The frequency response difference data d obtaining unit is used for obtaining the frequency response difference data d with the largest absolute value in the frequency response difference data and determining the frequency point f corresponding to the frequency response difference data d. The frequency response difference data d judging unit is used for comparing the absolute value of the frequency response difference data d with a first preset threshold value. And the central frequency value determining unit is used for taking the frequency point f as the central frequency value of the Mth frequency response calibration filter. The gain value determination unit is used for taking the absolute value of the frequency response difference value data d as the gain value of the Mth frequency response calibration filter. The filter type determining unit is used for determining the type of the Mth frequency response calibration filter according to the positive and negative values of the frequency response difference data d. The frequency coverage range value determining unit is used for determining the frequency coverage range value of the Mth frequency response calibration filter. The latest frequency curve forming unit to be calibrated is used for fitting the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the latest frequency curve to be calibrated. The M value setting unit is used for setting the initial value of M to be 1, and adding one to the value of M after the latest frequency curve to be calibrated is obtained by the latest frequency curve to be calibrated forming unit.

The frequency response calibration system of the embodiment can determine the number of the frequency response calibration filters required to be used by the microphone to be calibrated and the frequency response calibration parameters (including the center frequency value, the gain value and the type) corresponding to each frequency response calibration filter through the frequency response difference data d, and can calibrate the frequency response corresponding to the microphone to be calibrated to be very close to the frequency response of the reference microphone through the frequency response calibration filters corresponding to the number and the frequency response calibration parameters. The method comprises the steps of firstly calibrating the position with the largest difference between a frequency curve to be calibrated and a reference frequency curve through a1 st frequency response calibration filter, then calibrating the position with the second largest difference between the frequency curve to be calibrated and the reference frequency curve through a2 nd frequency response calibration filter, \8230, and finally calibrating the position with the smallest difference between the frequency curve to be calibrated and the reference frequency curve through an nth frequency response calibration filter.

Further, as shown in fig. 3, the frequency coverage value determining unit includes an initial frequency coverage value setting subunit, an undetermined frequency curve forming subunit, a difference value operator unit, a difference value determining subunit, a contrast difference value setting subunit, an S value setting subunit, a frequency coverage value adjusting subunit, a frequency coverage value recording subunit, and an S value determining subunit.

The initial frequency coverage range value setting subunit is configured to set an initial value for the frequency coverage range value of the mth frequency response calibration filter. And the undetermined frequency curve forming subunit is used for fitting the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the undetermined frequency curve. And the difference value operator unit is used for calculating the difference value between the to-be-determined frequency curve and the reference frequency curve. The difference value judging subunit is used for comparing the difference value obtained by the calculation of the difference value operator unit with the comparison difference value. The control difference value setting subunit is configured to set the corresponding difference value as the control difference value when the difference value is calculated for the first time or when the difference value is smaller than the control difference value. The S-value setting subunit is configured to set a value of the count value S. The frequency coverage range value adjusting subunit is configured to adjust a frequency coverage range value of the mth frequency response calibration filter. The frequency coverage range value recording subunit is used for recording the corresponding frequency coverage range value in the frequency coverage range value adjusting subunit when the difference value is smaller than the comparison difference value. The S value judging subunit is used for comparing the count value S with a second preset threshold value.

The frequency response calibration system of the embodiment can automatically determine the frequency coverage value of the corresponding frequency response calibration filter, and the determined frequency coverage value can prevent the difference value between the latest frequency curve to be calibrated and the reference frequency curve, which is formed by fitting the frequency response generated by the corresponding frequency response calibration filter and the frequency curve to be calibrated, from being smaller.

Further, as shown in fig. 4, the difference value calculation subunit includes an undetermined frequency response data forming subunit, a reference frequency response data forming subunit, an initial difference value calculation data acquisition subunit, a final difference value calculation data acquisition subunit, a difference value calculation subunit, a weight curve forming subunit, and a weight data acquisition subunit.

The undetermined frequency response data forming subunit is used for discretizing the undetermined frequency curve at equal frequency intervals to obtain a group of undetermined frequency response data. The reference frequency response data forming subunit is used for discretizing the reference frequency curve at equal frequency intervals to obtain a group of reference frequency response data. The initial difference value calculation data acquisition subunit is used for carrying out one-to-one corresponding subtraction on data point values in the undetermined frequency response data and corresponding data point values in the reference frequency response data so as to obtain a group of initial difference value calculation data. And the final difference value calculation data acquisition subunit is used for multiplying the data point values in the initial difference value calculation data and the data point values in the weight data in a one-to-one correspondence manner to obtain a group of final difference value calculation data. The difference value calculating subunit is configured to perform root-mean-square calculation on the data point values in the final difference value calculating data to obtain the difference value.

The weight curve forming subunit is used for forming a weight curve through a normal distribution formula 1);

1）

wherein ,

is a normal distribution function, parameter

Is 1, parameter

Is the value of the frequency point f, and

the maximum value of the value range is the maximum frequency point value in the reference frequency response data. The weight data acquisition subunit is used for discretizing the equal frequency intervals of the weight curve to obtain a group of weight data, and the frequency intervals of the weight data are the same as those of the reference frequency response data.

When the frequency response calibration system of this embodiment calculates the difference between the undetermined frequency curve and the reference frequency curve, a weight curve is introduced, and the setting of the weight curve makes the difference between the undetermined frequency curve and the reference frequency curve in a place with a large difference larger and the difference between the undetermined frequency curve and the reference frequency curve in a place with a small difference smaller, so that the calculation result of the difference is more reasonable (i.e., the difference of the frequency response values in the frequency band in which the frequency response calibration filter is set is strengthened, and the frequency response difference in other frequency bands is weakened), and finally the setting of the parameters (i.e., the frequency coverage range value) of the frequency response calibration filter is more accurate.

The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the spirit and scope of the present invention. Various modifications and improvements of the technical solutions of the present invention may be made by those skilled in the art without departing from the design concept of the present invention, and the technical contents of the present invention are all described in the claims.

Claims

1. A frequency response calibration method of a microphone array is characterized by comprising the following steps: comprises the following steps

The L1.N value setting module sets the initial value of N to 1;

l4, converting the calibration audio collected by the microphone to be calibrated into a frequency curve to be calibrated by the frequency curve forming module to be calibrated; the reference frequency curve forming module converts the calibration audio collected by the reference microphone into a reference frequency curve;

the L6.N value judging module compares the value of N with the total number of the microphones to be calibrated, and ends when the value of N is larger than the total number of the microphones to be calibrated; otherwise, L2 is returned.

2. The method of claim 1, wherein the method comprises: the L5 specifically comprises the following steps

An l51.M value setting unit sets an initial value of M to 1;

l52, discretizing the equal frequency interval of the frequency curve to be calibrated by the first frequency response value data forming unit to obtain a group of first frequency response value data, discretizing the equal frequency interval of the reference frequency curve by the second frequency response value data forming unit to obtain a group of second frequency response value data, wherein the frequency interval of the first frequency response value data is the same as that of the second frequency response value data;

l54, a frequency response difference data d obtaining unit obtains frequency response difference data d with the maximum absolute value in the frequency response difference data, and determines a frequency point f corresponding to the frequency response difference data d;

l55, comparing the absolute value of the frequency response difference data d with a first preset threshold by a frequency response difference data d judging unit, finishing when the absolute value of the frequency response difference data d is smaller than the first preset threshold, and adding one to the value of N by an N value setting module; entering L56 when the absolute value of the frequency response difference data d is greater than or equal to a first preset threshold;

l57, the latest to-be-calibrated frequency curve forming unit fits the to-be-calibrated frequency curve with the frequency response generated by the Mth frequency response calibration filter to obtain the latest to-be-calibrated frequency curve; the M value setting unit increments the value of M by one and returns to L52.

3. The method of claim 2, wherein the method comprises: the step of determining the frequency coverage value of the mth frequency response calibration filter by the frequency coverage value determining unit in the L56 specifically includes

l562, the undetermined frequency curve forming subunit fits a frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain an undetermined frequency curve;

l563. A difference value operator unit calculates the difference value between the frequency curve to be determined and the reference frequency curve; a comparison difference value setting subunit sets the corresponding difference value as a comparison difference value;

the S value setting subunit sets the count value S to 1;

l566. The frequency curve to be calibrated is fitted with the frequency response generated by the Mth frequency response calibration filter by the frequency curve to be determined forming subunit to obtain a new frequency curve to be determined;

4. The method of claim 3, wherein the method comprises: the step of calculating the difference between L563 and L567 specifically comprises

l372, the initial difference value calculation data acquisition sub-unit performs one-to-one corresponding subtraction on data point values in the frequency response data to be determined and corresponding data point values in the reference frequency response data to obtain a group of initial difference value calculation data;

5. The method of claim 4, wherein the frequency response of the microphone array is calibrated by: the step of acquiring the weight data in L373 comprises

1）

wherein ,

is a normal distribution function, parameter

Is 1, parameter

Is the value of the frequency point f, and

6. A frequency response calibration system of a microphone array is characterized in that: comprises that

The calibration audio processing device comprises a to-be-calibrated frequency curve forming module, a to-be-calibrated microphone and a to-be-calibrated frequency curve generating module, wherein the to-be-calibrated frequency curve forming module is used for converting calibration audio collected by the to-be-calibrated microphone into a to-be-calibrated frequency curve;

the N value setting module is used for setting the initial value of N to be 1, and adding one to the value of N after the frequency response calibration parameter determining module stores the frequency response calibration parameter of the frequency response calibration filter into the corresponding microphone to be calibrated;

7. The system of claim 6, wherein: the frequency response calibration parameter determination module comprises

the frequency response difference data d acquisition unit is used for acquiring the frequency response difference data d with the maximum absolute value in the frequency response difference data and determining the frequency point f corresponding to the frequency response difference data d;

a gain value determination unit, configured to use an absolute value of the frequency response difference data d as a gain value of the mth frequency response calibration filter;

the latest frequency curve forming unit is used for fitting the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the latest frequency curve to be calibrated;

8. The system of claim 7, wherein: the frequency coverage value determination unit comprises

the undetermined frequency curve forming subunit is used for fitting the frequency curve to be calibrated with the frequency response generated by the Mth frequency response calibration filter to obtain the undetermined frequency curve;

the difference value judging subunit is used for comparing the difference value obtained by the calculation of the difference value operator unit with the comparison difference value;

a comparison difference value setting subunit, configured to set the corresponding difference value as a comparison difference value when the difference value is obtained by the first calculation or when the difference value is smaller than the comparison difference value;

an S value setting subunit, configured to set a value of the count value S;

the frequency coverage range value adjusting subunit is used for adjusting the frequency coverage range value of the Mth frequency response calibration filter;

and the S value judging subunit is used for comparing the count value S with a second preset threshold value.

9. The system of claim 8, wherein the frequency response of the microphone array is calibrated by: the difference value operator unit comprises

The undetermined frequency response data forming subunit is used for carrying out equal-frequency interval discretization on the undetermined frequency curve to obtain a group of undetermined frequency response data;

10. The system of claim 8, wherein: the difference value operator unit further comprises

A weight curve forming subunit for forming a weight curve by the normal distribution formula 1);

1）

wherein ,

is a normal distribution function, parameter

Is 1, parameter

Is the value of frequency point f, an