KR100873000B1 - Directional voice filtering system using microphone array and method thereof - Google Patents

Abstract

The present invention relates to a directional sound source filtering system using a microphone array and a method thereof, comprising: (a) inputting a sound source signal through a microphone sensor array having a plurality of microphone sensors installed at equal or non-equal intervals; (b) amplifying and A / D converting each signal input through the plurality of microphone sensors and storing the amplified signal; (c) decomposing the stored signal by frequency conversion (FFT) for each microphone sensor for each frequency; (d) applying a complex weight to each frequency signal decomposed for each frequency; (e) inverse frequency converting (IFFT) the frequency signal to which the complex weight is applied; And (f) outputting the inverse frequency transformed (IFFT) signal by D / A conversion. Accordingly, the microphone array may be used to selectively input only a signal in a desired direction and to remove a sound source in an undesired direction, and to remove noise incident from a different direction by using the directionality of the sound source. In addition, even if the microphone system is not mechanically moved, it is possible to steer in a desired direction by changing the weight of each microphone sensor electronically.

Microphone sensor, array, beamforming, sound source, voice, sound, signal, direction, filtering, rejection, frequency, noise, weight

Description

Directive voice filtering system using microphone array and method

1 is a block diagram of a directional sound filtering system using a microphone array according to a preferred embodiment of the present invention

2 and 3 are explanatory diagrams for explaining a directional sound source filtering method using a microphone array according to a preferred embodiment of the present invention.

4 is a configuration diagram of a broadband beamforming circuit according to a preferred embodiment of the present invention.

5 is a view for explaining a method for calculating an array coefficient in a wideband beamforming circuit of the present invention.

)=0°' 일 때의 배열 계수을 나타낸 도면6 shows the direction of the desired beam (

A diagram showing the array coefficient at) = 0 ° '

)=20°' 일 때의 배열 계수을 나타낸 도면7 shows the direction of the desired beam (

A diagram showing the array coefficient when) = 20 ° '

8 is a flowchart illustrating a method of filtering a directional sound source using a microphone array according to an exemplary embodiment of the present invention.

[Description of Code for Major Parts of Drawing]

10 microphone array 20 voice amplifier

100: wideband beamforming circuit

110: buffer 120: frequency converter (FFT)

130: first multiplier 140: second multiplier

150: summer 160: inverse frequency converter (IFFT)

The present invention relates to a directional sound source filtering system using a microphone array (or 'array'), and a method thereof. In particular, a microphone array capable of inputting sound in a desired direction using a conventional microphone array is provided. The present invention relates to a directional sound source filtering system used and a method thereof.

In general, a beam forming method receives a signal propagating from a desired direction and removes a signal propagating from the other direction. Signal processing is performed using an algorithm.

However, in the conventional sound source filtering system, there is an inconvenience of changing the microphone array in the direction in which the signal is received in order to receive the signal in the desired direction, and there is a problem that the algorithm for processing the signal is difficult to implement. In addition, there is a problem that the noise is badly generated when the sound source is not properly filtered.

Accordingly, the present invention has been made to solve the above problems, and a first object of the present invention is to provide a microphone array capable of steering in a desired direction by electronically changing the weight of each microphone sensor even without mechanically moving the microphone system. It is to provide a directional sound source filtering system used.

It is also a second object of the present invention to provide a directional sound source filtering system using a microphone array capable of selectively inputting only a signal in a desired direction using a microphone array and removing a sound source in an undesired direction.

In addition, a third object of the present invention is to provide a directional sound source filtering method using a microphone array capable of steering in a desired direction by electronically changing the weight of each microphone sensor even if the microphone system is not mechanically moved.

Further, a fourth object of the present invention is to provide a directional sound source filtering method using a microphone array capable of selectively inputting a signal in a desired direction using a microphone array and removing a sound source in an undesired direction.

Directional sound filtering method using a microphone array according to the present invention for achieving the above object, (a) the step of inputting a sound source signal through a microphone sensor array consisting of a plurality of microphone sensors are installed at equal intervals or boiling intervals; ; (b) amplifying and A / D converting each signal input through the plurality of microphone sensors and storing the amplified signal; (c) decomposing the stored signal by frequency conversion (FFT) for each microphone sensor for each frequency; (d) applying a complex weight to each frequency signal decomposed for each frequency; (e) inverse frequency converting (IFFT) the frequency signal to which the complex weight is applied; And (f) outputting the inverse frequency transformed (IFFT) signal by D / A conversion.

The complex weighting value is calculated by the following equation.

[Equation]

는 제 1 곱셈기(130)에 곱해지는 가중치 값, k 번째 마이크의

번째 주파수의 가중치, 즉,

에서

는

번째 파장,

는 각속도,

는 신호의 속도,

는 크기를 가지는 상수,

는 자연로그의 밑,

은 허수,

는 마이크 어레이의 인덱스,

는

번째 어레이에 들어오는 신호,

는 원하는 빔의 방향,

는 각 센서의 간격을 나타낸다.From here

Is a weight value multiplied by the first multiplier 130, k k of the microphone

Weight of the first frequency, i.e.

in

Is

First wavelength,

Is the angular velocity,

Is the speed of the signal,

Is a constant with size,

Is the base of the natural logarithm,

Imaginary,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the desired beam direction,

Represents the interval of each sensor.

In the directional sound source filtering method using the microphone array, an array coefficient for outputting a signal in a desired direction is obtained by Equation 1 to Equation 3 below.

[Equation 1]

[Equation 2]

[Equation 3]

는 각 어레이의 복소가중치,

는 파장,

는 크기를 가지는 상수,

는 자연로그의 밑,

는 마이크 어레이의 인덱스,

는

번째 어레이에 들어오는 신호,

는 일반 복소수 신호의 크기,

는 배열계수이며, 빔의 모양,

는 신호의 입사각,

는 원하는 빔의 방향,

는 신호의 입사각,

는 각 센서의 간격을 나타낸다.here,

Is the complex weight of each array,

The wavelength,

Is a constant with size,

Is the base of the natural logarithm,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the magnitude of a common complex signal,

Is the array coefficient, the shape of the beam,

Is the angle of incidence of the signal,

Is the desired beam direction,

Is the angle of incidence of the signal,

Represents the interval of each sensor.

In addition, the directional sound source filtering system using the microphone array according to the present invention for achieving the above object, the microphone sensor array is composed of a plurality of microphone sensors are installed at equal intervals or non-equal intervals to input a sound source signal, respectively; A plurality of voice amplifiers for amplifying each signal input through the plurality of microphone sensors; And a wideband beamforming circuit for inputting a signal received through the microphone sensor array from the plurality of voice amplifiers to change a weight of each microphone sensor to selectively output only a signal in a desired direction and to remove a signal in an undesired direction; Characterized in that configured to include.

The wideband beamforming circuit includes: a plurality of buffers configured to input and store signals received through the microphone sensor array from the plurality of voice amplifiers; A plurality of frequency converters which receive signals from the plurality of buffers and perform frequency conversion (FFT) for each microphone sensor and decompose them for each frequency; A plurality of first and second multipliers for receiving signals from the plurality of frequency converters and multiplying complex weights corresponding to respective frequencies; A plurality of adders for adding and outputting two signals from among signals output from the plurality of second multipliers; And an inverse frequency converter (IFFT) for receiving signals from the plurality of summers and outputting only a signal in a desired direction by performing inverse frequency conversion (IFFT).

Accordingly, the microphone array may be used to selectively input only a signal in a desired direction and to remove a sound source in an undesired direction, and to remove noise incident from a different direction by using the directionality of the sound source.

In addition, even if the microphone system is not mechanically moved, steering in the desired direction is possible by electronically changing the weight of each microphone sensor.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Directional sound source using microphone array Filter  system

1 is a block diagram of a directional sound filtering system using a microphone array according to a preferred embodiment of the present invention.

Directional sound filtering system using a microphone array according to a preferred embodiment of the present invention, as shown in Figure 1, a microphone sensor array consisting of a plurality of microphone sensors are installed at equal or non-equal intervals to input a sound source signal, respectively ( 10) and a plurality of voice amplifiers 20 for amplifying respective signals input through the plurality of microphone sensors, and signals received through the microphone sensor array 10 from the plurality of voice amplifiers 20. It is configured to include a wideband beamforming circuit 100 for inputting and varying the weight of each microphone sensor to selectively output only a signal in a desired direction and to remove a signal in an undesired direction.

The plurality of microphone sensors are sensors having almost no directivity of about 120 degrees in all directions (360 degrees), and the microphone sensor array 10 using the same may be configured at equal intervals or boiling intervals.

Each voice signal received through the microphone sensor array 10 is appropriately amplified by the plurality of voice amplifiers 20 configured to correspond to the microphone sensors, respectively, and then input to the wideband beamforming circuit 100. .

The wideband beamforming circuit 100 receives a plurality of voice signals received through the microphone sensor from the plurality of voice amplifiers 20, selects and outputs only a sound source in a predetermined direction (or a desired direction), and The sound source in the direction not set is removed.

Since the voice signal is a wideband signal, a wideband frequency independent beamforming method is used instead of a narrowband beamforming method. This method accumulates a predetermined time data for each sensor and converts it into a frequency domain to narrowband beam for each frequency. After forming, inverse frequency conversion is performed again to perform wideband beamforming. Therefore, in the present invention, noise incident from different directions may be removed by using the directivity of the sound source.

2 and 3 are explanatory diagrams for explaining a directional sound source filtering method using a microphone array according to a preferred embodiment of the present invention.

As shown in FIG. 2, when an English speaker and a Chinese speaker are present at 90 degrees and 30 degrees and speak at the same time, when the microphone system (microphone sensor array) is steered at 90 degrees, the English speaker Only the speaker's voice is heard, and when steered at 30 degrees, only the Chinese speaker's voice is heard.

In the directional sound source filtering system using the microphone array according to the present invention, it is not necessary to move the microphone system mechanically, and the steering in the desired direction is possible by electronically changing the weight of each microphone sensor.

In addition, when using the voice input home network device in the home when it is not possible to recognize the directional noise, such as the surrounding noise, radio sound, etc., when using the system according to the present invention can be steered in the direction of the speaker to filter such noises can do.

In addition, when a voice input is required by hands-free or voice recognition when a phone call is made in a car, when noise is input due to a car engine sound or a wind noise of a window, the system according to the present invention can remove directional noise. High quality voice input is possible.

Broadband Beamforming  Circuit (100)

4 is a configuration diagram of a broadband beamforming circuit 100 according to a preferred embodiment of the present invention.

)를 곱하는 복수 개의 제 1 및 제 2 곱셈기(130)(140)와, 상기 복수 개의 제 2 곱셈기(140)로부터 출력되는 신호 중에서 각각 2개의 신호를 더하여 출력하는 복수 개의 합산기(150)와, 상기 복수 개의 합산기(150)로부터 신호를 수신받아 역주파수 변환(IFFT)하여 원하는 방향의 신호만을 출력하는 역주파수 변환부(IFFT)(160)을 포함하여 구성한다.As shown in FIG. 4, the wideband beamforming circuit according to an exemplary embodiment of the present invention receives a signal X _K (t) received from the plurality of voice amplifiers 20 through the microphone sensor array 10. A plurality of frequencies for decomposing the signal for each frequency by receiving a signal (X _{K , M} ) from each of the plurality of buffers 110 and the plurality of buffers 110 to store and input the frequency conversion (FFT) for each microphone sensor A complex weight corresponding to each frequency by receiving signals from the converter (FFT) 120 and the plurality of frequency converters 120 (

A plurality of first and second multipliers 130 and 140 to multiply), a plurality of adders 150 for adding and outputting two signals from the signals output from the plurality of second multipliers 140, and It includes an inverse frequency transform unit (IFFT) 160 for receiving signals from the plurality of summers 150 and outputting only a signal in a desired direction by performing inverse frequency transform (IFFT).

The operation of the broadband beamforming circuit of the present invention having the above configuration will be described below with reference to FIGS. 5 to 7.

)를 곱하여 더한 값(배열 계수(array factor: F))은 다음의 식과 같다.First, when the K equally spaced linear microphone sensor arrays 10 are shown in FIG. 5, the complex signal is obtained by comparing the input signal x (t) and the values output from the respective sensors.

) And multiplying them (array factor (F)) is as follows.

에서 배열 계수(Array factor)

가 최대의 응답일 때,

는 원하는 빔의 방향이다. 즉, 안테나 빔이 입사하는 신호의 방향으로 조정되었다 할 수 있다. If the

Array factor in

Is the maximum response,

Is the direction of the desired beam. That is, the antenna beam may be adjusted in the direction of the incident signal.

If Equation 2 is developed using the sum of the equal sequences, an array factor may be expressed in the form of the expression of the “sinc” function.

는 제1 곱셈기에 곱해지는 가중치 값, k 번째 마이크의

번째 주파수의 가중치, 즉,

에서

는

번째 파장,

는 각속도,

는 신호의 속도,

는 파장,

는 크기를 가지는 상수,

는 자연로그의 밑,

은 허수,

는 마이크 어레이의 인덱스,

는

번째 어레이에 들어오는 신호,

는 원하는 빔의 방향,

는 신호의 입사각,

는 각 센서의 간격을 나타낸다.
참고로, 도 6은 '원하는 빔의 방향(

)=0°' 일 때의 배열 계수을 나타낸 도면이고, 도 7은 '원하는 빔의 방향(

)=20°' 일 때의 배열 계수(array factor)를 나타낸 도면이다.From here

Is the weight value multiplied by the first multiplier,

Weight of the first frequency, i.e.

in

Is

First wavelength,

Is the angular velocity,

Is the speed of the signal,

The wavelength,

Is a constant with size,

Is the base of the natural logarithm,

Imaginary,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the desired beam direction,

Is the angle of incidence of the signal,

Represents the interval of each sensor.
For reference, FIG. 6 shows the direction of the desired beam (

) Is a diagram showing the arrangement coefficient when 0 ° ', and FIG. 7 shows the direction of the desired beam (

It is an illustration of an array factor when) = 20 ° '.

Broadband Beamforming Wideband beamforming )

가 변화하게 되므로 배열 계수(array factor)은 주파수에 따라 변화하게 된다. In general, the voice signal is a wideband signal distributed around 200 Hz to 3000 Hz. In Equation 3

Since is changed, the array factor changes with frequency.

If you do not want the beam width to be a function of frequency, you can divide the signal by each frequency, apply weights, and add them together to create an independent beam of frequency. In other words, wideband beamforming stores signals input from each microphone sensor in a buffer for a certain period, decomposes them by frequency by performing a frequency conversion (FFT) for each sensor, and obtains a weight corresponding to each frequency. After the application, the filtered voice signal is output again through an IFFT process. This process is shown in Figure 4. Complex weights are given as a function of the wavelength corresponding to each frequency. That is, the weight in the i-th frequency domain of the k-th microphone sensor is expressed by Equation 4.

Directional sound source using microphone array Filter  Way

8 is a flowchart illustrating a directional sound source filtering method using a microphone array according to an exemplary embodiment of the present invention.

1, 4 and 8, the sound source signal is input through the microphone sensor array 10 which is installed at equal intervals or non-equal intervals and is configured of a plurality of microphone sensors (step S10).

Then, after amplifying and A / D converting each of the signals input through the plurality of microphone sensors through the voice amplifier 20 and storing them in the buffer 110 of the broadband beamforming circuit 100 ( Step S20).

Next, the plurality of frequency converters (FFT) 120 receives the signals stored from the plurality of buffers 110 and performs frequency conversion (FFT) on each microphone sensor to decompose them by frequency (step S30).

Then, a complex weight value is applied to each frequency signal decomposed for each frequency through the first and second multipliers 130 and 140 and the summer 150 (step S40). In this case, the complex weight may be calculated by the following equation.

는 제1 곱셈기(130 또는 140)에 곱해지는 가중치 값, k 번째 마이크의

번째 주파수의 가중치, 즉,

에서

는

번째 파장,

는 각속도,

는 신호의 속도,

는 크기를 가지는 상수,

는 자연로그의 밑,

은 허수,

는 마이크 어레이의 인덱스,

는

번째 어레이에 들어오는 신호,

는 원하는 빔의 방향,

는 각 센서의 간격을 나타낸다.From here

Is a weight value that is multiplied by the first multiplier 130 or 140,

Weight of the first frequency, i.e.

in

Is

First wavelength,

Is the angular velocity,

Is the speed of the signal,

Is a constant with size,

Is the base of the natural logarithm,

Imaginary,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the desired beam direction,

Represents the interval of each sensor.

Then, by inverting the frequency signal to which the complex weight value is applied, through the inverse frequency converter (IFFT) 160 again, only a signal in a desired direction is selectively input and a sound source in an unwanted direction is removed. It may be (step S50).

Finally, the inverse frequency transformed (IFFT) signal is subjected to D / A conversion and output (step S60).

The present invention is not limited to the above-described embodiments, and various modifications and changes can be made by those skilled in the art, which should be regarded as included in the spirit and scope of the present invention as defined in the appended claims. will be.

As described above, according to the directional sound source filtering system using the microphone array according to the present invention and the method, it is possible to selectively input only a signal in a desired direction and to remove a sound source in an undesired direction by using the microphone array. .

In addition, the directionality of the sound source may be used to remove noise incident from different directions.

In addition, even if the microphone system is not mechanically moved, it is possible to steer in a desired direction by changing the weight of each microphone sensor electronically.

Claims (7)

In the directional sound source filtering method using a microphone array, (a) inputting a sound source signal through a microphone sensor array having a plurality of microphone sensors installed at equal or non-equal intervals; (b) amplifying and A / D converting each signal input through the plurality of microphone sensors and storing the amplified signal; (c) decomposing the stored signal by frequency conversion (FFT) for each microphone sensor for each frequency; (d) applying a complex weight to each frequency signal decomposed for each frequency; (e) inverse frequency converting (IFFT) the frequency signal to which the complex weight is applied; And and (f) outputting the inverse frequency transformed (IFFT) signal by D / A conversion. The method of claim 1, The complex weighting method is a directional sound source filtering method using a microphone array, characterized in that calculated by the following equation. [Equation]

From here

Is a weight value that is multiplied by the first multiplier 130 or 140,

Weight of the first frequency, i.e.

in

Is

First wavelength,

Is the angular velocity,

Is the speed of the signal,

Is a constant with size,

Is the base of the natural logarithm,

Imaginary,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the desired beam direction,

Represents the interval of each sensor. The method according to claim 1 or 2, Directional sound filtering method using the microphone array is: An array coefficient for outputting a signal in a desired direction is obtained by the following equations (1) to (3) below. [Equation 1]

[Equation 2]

[Equation 3]

here,

Is the complex weight of each array,

The wavelength,

Is a constant with size,

Is the base of the natural logarithm,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the magnitude of a common complex signal,

Is the array coefficient, the shape of the beam,

Is the angle of incidence of the signal,

Is the desired beam direction,

Is the angle of incidence of the signal,

Represents the interval of each sensor. In the directional sound filtering system using a microphone array, A microphone sensor array having a plurality of microphone sensors installed at equal intervals or at non-equal intervals to respectively input sound source signals; A plurality of voice amplifiers for amplifying each signal input through the plurality of microphone sensors; And A wideband beamforming circuit for inputting a signal received through the microphone sensor array from the plurality of voice amplifiers to vary the weight of each microphone sensor to selectively output only a signal in a desired direction and to remove a signal in an undesired direction; Directional sound filtering system using a microphone array, characterized in that configured to include. 5. The broadband beamforming circuit of claim 4 wherein: A plurality of buffers for receiving and storing the signals received through the microphone sensor array from the plurality of voice amplifiers; A plurality of frequency converters which receive signals from the plurality of buffers and perform frequency conversion (FFT) for each microphone sensor and decompose them for each frequency; A plurality of first and second multipliers for receiving signals from the plurality of frequency converters and multiplying complex weights corresponding to respective frequencies; A plurality of adders for adding and outputting two signals from among signals output from the plurality of second multipliers; And And an inverse frequency transform unit (IFFT) for receiving signals from the plurality of summers and outputting only a signal in a desired direction by performing inverse frequency transform (IFFT) on the directional sound source filtering system using a microphone array. The method of claim 5, wherein The complex weight value is a directional sound source filtering system using a microphone array, characterized in that calculated by the following equation. [Equation]

From here

Is a weight value that is multiplied by the first multiplier 130 or 140,

Weight of the first frequency, i.e.

in

Is

First wavelength,

Is the angular velocity,

Is the speed of the signal,

Is a constant with size,

Is the base of the natural logarithm,

Imaginary,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the desired beam direction,

Represents the interval of each sensor. The method according to claim 4 or 6, Directional sound filtering system using the microphone array is: Directional sound source filtering system using a microphone array, characterized by obtaining the array coefficient for outputting a signal in a desired direction by the following equations (1) to (3). [Equation 1]

[Equation 2]

[Equation 3]

here,

Is the complex weight of each array,

The wavelength,

Is a constant with size,

Is the base of the natural logarithm,

Is the index of the microphone array,

Is

Incoming signal to the first array,

Is the magnitude of a common complex signal,

Is the array coefficient, the shape of the beam,

Is the angle of incidence of the signal,

Is the desired beam direction,

Is the angle of incidence of the signal,

Represents the interval of each sensor.

Images