KR101903874B1 - Noise reduction method and apparatus based dual on microphone - Google Patents

Abstract

According to one embodiment, a dual microphone-based noise cancellation method is a method of removing two-channel input signals received from a time domain, wherein each of the dual microphones - each of the dual microphones is located at a symmetric location from a point of voice generation in the user's body - Converting into a frequency domain; Calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals; Determining whether a voice signal is included in the two-channel input signals based on a variance of the phase of the coherence; Updating the noise reduction filter based on the determination result; And filtering the two-channel input signals using the updated noise reduction filter.

Description

TECHNICAL FIELD [0001] The present invention relates to a dual microphone-based noise cancellation method and apparatus,

The following description relates to a dual microphone-based noise canceling apparatus and method thereof, and more particularly, to a dual microphone-based noise canceling apparatus and method thereof, and more particularly to a dual microphone- Lt; / RTI >

In an outdoor recording environment using a microphone, wind noise greatly degrades the sound quality of the recorded signal. To solve this problem, many studies have been conducted in both the short channel environment and the two channel environment. Wiener filters using speech signals and other frequency characteristics of noise are most commonly used. The Wiener filter exhibits good performance in the case of static noise, but has a disadvantage in that it exhibits limited performance for wind noise, which is an abnormal signal that most of the energy is distributed in a low frequency band.

Therefore, a technique for effectively eliminating or reducing wind noise needs to be proposed.

Embodiments propose a noise cancellation method and apparatus for minimizing the loss of voice signals of two channel input signals in a dual microphone environment and eliminating noise signals such as wind noise by using the coherence of two channel input signals.

In particular, one embodiment includes a noise cancellation method for adaptively updating a noise cancellation filter using the phase and intensity of the coherence of two channel input signals, filtering the two channel input signals based on the updated noise cancellation filter, Device.

In addition, one embodiment suggests a wearable device that uses a coherence of two-channel input signals to minimize the loss of voice signals of two-channel input signals in a dual-microphone environment and to remove noise signals .

According to one embodiment, a dual microphone-based noise cancellation method is a method of removing two-channel input signals received from a time domain, wherein each of the dual microphones - each of the dual microphones is located at a symmetric location from a point of voice generation in the user's body - Converting into a frequency domain; Calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals; Determining whether a voice signal is included in the two-channel input signals based on a variance of the phase of the coherence; Updating the noise reduction filter based on the determination result; And filtering the two-channel input signals using the updated noise reduction filter.

According to one aspect of the present invention, the step of updating the noise cancellation filter may include the steps of: when the voice signal is not included in the two-channel input signals, removing the noise signal included in the two- And adjusting the gain of the noise reduction filter based on the strength of the coherence.

According to another aspect of the present invention, the step of adjusting the gain of the noise canceling filter may include adjusting the gain of the noise canceling filter based on the intensity of the coherence at the previous time when the variance of the phase of the coherence is greater than the threshold, Adjusting a gain of the noise reduction filter at a current time according to a gain value of the noise reduction filter; Or adjusting the gain of the noise reduction filter based on the intensity of the coherence when the variance of the phase of the coherence is less than or equal to the threshold value.

According to another aspect of the present invention, the step of updating the noise cancellation filter includes a step of, when a voice signal is included in the two-channel input signals, adjusting the gain of the noise cancellation filter to prevent loss of the voice signal by the noise cancellation filter And the like.

According to another aspect of the present invention, the step of adjusting the gain of the noise cancellation filter may include adjusting the gain of the noise cancellation filter to a region where the intensity of the coherence is greater than a threshold value, ; And adjusting a gain of the noise reduction filter for an area where the intensity of the coherence is less than or equal to the threshold among the frequency areas of the two channel input signals based on the intensity of the coherence .

According to another aspect of the present invention, the step of determining whether or not a voice signal is included in the two-channel input signals includes: when the phase of the coherence is distributed to a specific value, ; And determining that the voice signal is not included in the two-channel input signals when the phase of the coherence is not distributed to the specific value.

According to another aspect of the present invention, there is provided a dual microphone-based noise canceling method, comprising: calculating a phase of the coherence; And calculating the intensity of the coherence.

According to another aspect, calculating the coherence of the 2-channel input signals comprises calculating a PSD of the 2-channel input signals using a recursive smoothed periodogram technique; And a normalized coherence of the 2-channel input signals at a ratio of APSD (Auto Power Spectral Density) and CPSD (Cross Power Spectral Density) of the 2-channel input signals based on the PSD of the calculated 2-channel input signals. May be defined.

According to another aspect, the step of filtering the 2-channel input signals may include performing Fixed beamforming on the 2-channel input signals, applying the updated noise reduction filter to the signal subjected to Fixed beamforming ; And outputting a signal to which the updated noise reduction filter is applied.

According to another aspect of the present invention, the dual microphone-based noise cancellation method may further include converting the output signal from the frequency domain to the time domain.

According to one embodiment, a dual microphone-based noise canceling apparatus includes two-channel input signals received from dual microphones - each of the dual microphones being located at a symmetrical position from a voice generating point of the user's body, To a frequency domain; A coherence calculator for calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals; A voice signal determination unit for determining whether a voice signal is included in the two-channel input signals based on a dispersion of the phase of the coherence; A filter updating unit updating the noise reduction filter based on the determination result; And a filtering unit for filtering the two-channel input signals using the updated noise reduction filter.

According to one aspect of the present invention, the noise canceling apparatus further includes a fixed beamforming performing unit for performing fixed beamforming with respect to the two-channel input signals, and the filtering unit may perform the fixed beamforming on the updated noise canceling filter And may output a signal to which the updated noise reduction filter is applied.

According to another aspect of the present invention, the noise canceller may further include a time domain transformer for transforming the output signal from the frequency domain to the time domain.

According to one embodiment, a wearable device worn on a wearer's body to provide a call function includes dual microphones disposed at symmetrical positions from the point of sound generation of the user's body; And a noise eliminator connected to the dual microphones, wherein the noise canceller comprises: a frequency domain converter for converting the two channel input signals received from the dual microphones into a frequency domain from a time domain; A coherence calculator for calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals; A voice signal determination unit for determining whether a voice signal is included in the two-channel input signals based on a dispersion of the phase of the coherence; A filter updating unit updating the noise reduction filter based on the determination result; And a filtering unit for filtering the two-channel input signals using the updated noise reduction filter.

One embodiment may propose a noise cancellation method and apparatus that minimizes loss of voice signals of two channel input signals in a dual microphone environment and eliminates noise signals such as wind noise by using coherence of two channel input signals .

In particular, one embodiment includes a noise cancellation method for adaptively updating a noise cancellation filter using the phase and intensity of the coherence of two channel input signals, filtering the two channel input signals based on the updated noise cancellation filter, Device can be proposed.

In addition, one embodiment suggests a wearable device that uses a coherence of two-channel input signals to minimize the loss of voice signals of two-channel input signals in a dual-microphone environment and to remove noise signals .

1 is a block diagram illustrating a configuration of a noise canceling apparatus according to an embodiment.
FIG. 2 is a diagram illustrating a phase distribution of coherence when two-channel input signals according to an embodiment include a voice signal and when not.
3 is a view for explaining dual microphones used in the noise canceling apparatus according to an embodiment.
4 is a flowchart illustrating a noise removal method according to an embodiment.
5 is a block diagram illustrating a wearable device including a noise canceller in accordance with one embodiment.

Hereinafter, embodiments according to the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to or limited by the embodiments. In addition, the same reference numerals shown in the drawings denote the same members.

Also, terminologies used herein are terms used to properly represent preferred embodiments of the present invention, which may vary depending on the viewer, the intention of the operator, or the custom in the field to which the present invention belongs. Therefore, the definitions of these terms should be based on the contents throughout this specification.

FIG. 1 is a block diagram illustrating a configuration of a noise canceling apparatus according to an embodiment. FIG. 2 is a diagram illustrating a case where two-channel input signals according to an embodiment include a voice signal, Fig.

1, the noise canceller 100 according to an embodiment includes a frequency domain transformer 110, a coherence calculator 120, a voice signal determiner 130, a filter updater 140, A fixed beamforming performing unit 150, a filtering unit 160, and a time domain transforming unit 170.

Hereinafter, the noise cancellation apparatus 100 may be implemented as a hardware module coupled to a wearable device having dual microphones, but is not limited thereto and may be a computer program installed in a processor included in the wearable device, And may be embodied in the form of a computer program stored in a recording medium readable by a device. Further, instead of being coupled with the wearable device, the noise canceller 100 may be implemented as a separate hardware module that communicates wired and wirelessly with the wearable device, and further, a computer installed in a processor included in a separate hardware module Program or a computer program stored on a readable medium in a separate hardware module.

Hereinafter, a case where the noise canceling apparatus 100 is used in a wearable apparatus will be described. However, the present invention is not limited thereto and can be used in combination with various electronic apparatuses equipped with dual microphones.

The frequency domain converter 110 converts the 2-channel input signals received from the dual microphones into the frequency domain from the time domain. Here, the dual microphones are disposed at symmetrical positions from the voice generation point in the user's body. A detailed description thereof will be described with reference to Fig.

For example, an FFT (Fast Fourier Transform) may be used as the frequency domain converter 110, and the time domain 2-channel input signals received from the dual microphones may be converted into 2-channel input signals of the frequency domain . Accordingly, the two-channel input signals are hereinafter referred to as two-channel input signals in the frequency domain.

The coherence calculation unit 120 calculates the coherence of the two channel input signals based on the PSD (Power Spectral Density) of the two channel input signals. Specifically, the coherence calculator 120 calculates the PSD of the 2-channel input signals in the frequency domain using the recursive smoothed periodogram technique as shown in Equation 1, Similarly, the normalized coherence of two channel input signals can be defined as a ratio of APSD (Auto Power Spectral Density) and CPSD (Cross Power Spectral Density) of two channel input signals.

<Formula 1>

<Formula 2>

는 2 채널 입력 신호들

,

의 PSD를 나타내고,

은 타임 인덱스를 나타내며,

는 주파수 인덱스를 나타내고,

는 쇼트 타임 스펙트럼을 나타내며,

는 smoothing constant를 나타낸다.In Equation 1,

Channel input signals &lt; RTI ID = 0.0 &

,

, &Lt; / RTI &gt;

Represents a time index,

Represents a frequency index,

Lt; / RTI &gt; represents a short-time spectrum,

Represents the smoothing constant.

는 2 채널 입력 신호들의 코히런스를 나타내고,

는 2 채널 입력 신호들

,

의 CPSD를 나타내며,

및

각각은 2 채널 입력 신호들

,

의 APSD를 나타낸다.In Equation 2,

Represents the coherence of the two channel input signals,

Channel input signals &lt; RTI ID = 0.0 &

,

Lt; / RTI &gt;

And

Each of the two-channel input signals

,

Lt; / RTI &gt;

Since the coherence of the 2-channel input signals calculated as described above is a useful feature for distinguishing the voice signal and the noise signal from the 2-channel input signals, the noise canceling apparatus 100 may include a voice signal determining unit 130, And the filter update unit 140 to use the coherence of the two-channel input signals.

The voice signal determination unit 130 determines whether or not the voice signal is included in the two-channel input signals based on the dispersion of the phase of the coherence.

More specifically, referring to FIG. 2, which illustrates the variance of the phase of each coherence when two-channel input signals include and not include a speech signal, if the two-channel input signals include speech signals 210, It can be seen that the coherence phase of the 2-channel input signals (phase difference of the 2-channel input signals) is mostly distributed to a value of 0, and the 2-channel input signals do not include speech signals but include only noise signals such as wind noise (220), it can be seen that the coherence phase of the two channel input signals is evenly distributed.

Therefore, when the phase of the coherence is distributed to a specific value (for example, a value of 0), the voice signal determination unit 130 can determine that the voice signal is included in the two-channel input signals, Is not distributed to the specific value, it can be determined that the voice signal is not included in the two-channel input signals.

The fact that the phase of the coherence is distributed to a particular value means that the variance of the coherence phase has a small value and the fact that the phase of the coherence is not distributed to a specific value but is evenly distributed means that the dispersion of the coherence phase The voice signal determination unit 130 can determine whether the voice signal is included in the two channel input signals by determining whether the variance of the coherence phase is larger or smaller than the threshold value have.

Here, the voice signal determination unit 130 determines the phase of the coherence of the 2-channel input signals before the determination process to determine whether the voice signal is included in the 2-channel input signals based on the dispersion of the phase of the coherence Can be calculated as shown in Equation 3 below.

<Formula 3>

는 2 채널 입력 신호들의 코히런스의 위상(2 채널 입력 신호들의 위상 차이)를 나타내고,

는 식 2와 같이 계산되는 2 채널 입력 신호들의 코히런스를 나타낸다.In Equation 3,

Represents the coherence phase of the two channel input signals (the phase difference of the two channel input signals)

Represents the coherence of the two-channel input signals calculated as Equation (2).

을 획득할 수 있다.Accordingly, the speech signal determination unit 130 determines the distribution of the coherence (the phase difference of the two-channel input signals) based on the coherence phase (the phase difference of the two-channel input signals) Dispersion)

Can be obtained.

<Formula 4>

However, the variance of the coherence phase and coherence of the two-channel input signals may be calculated in advance by the coherence calculator 120 instead of being calculated by the voice signal determiner 130, The voice signal determination unit 130 may use the dispersion of the phase of the coherence calculated by the coherence calculation unit 120. [

The filter update unit 140 updates the noise reduction filter based on the determination result of the voice signal determination unit 130. Hereinafter, a conventional Wiener filter may be used as the noise reduction filter, but not limited thereto, and various noise reduction filters may be used.

When the audio signal determination unit 130 determines that the audio signal is not included in the two-channel input signals, the filter update unit 140 updates the noise signal included in the two- The gain of the noise cancellation filter can be adjusted based on the intensity of the coherence (magnitude correlation between two channel input signals).

In this case, in order to use the intensity of the coherence in the process of adjusting the gain of the noise reduction filter, the filter update unit 140 calculates the coherence intensity of the two channel input signals before the gain adjustment process of the noise reduction filter, Can be calculated as follows.

&Lt; EMI ID =

는 2 채널 입력 신호들의 코히런스의 세기를 나타내고,

는 식 2와 같이 계산되는 2 채널 입력 신호들의 코히런스를 나타낸다.In Equation 5,

Represents the intensity of the coherence of the two channel input signals,

Represents the coherence of the two-channel input signals calculated as Equation (2).

However, the strength of the coherence of the two-channel input signals can be calculated in the course of calculating the variance of the coherence phase and the coherence phase of the two-channel input signals in the voice signal determination unit 130 And may be calculated in advance by the coherence calculation unit 120. [

Accordingly, when the voice signal is not included in the two-channel input signals, the filter update unit 140 can adjust the gain of the noise reduction filter to remove noise as shown in Equation 6 based on the intensity of the coherence. Hereinafter, the gain of the noise canceling filter adjusted for noise cancellation may be a gain applied to a frequency region in which noise signals exist in the frequency regions of the two channel input signals.

&Lt; EMI ID =

는 잡음 제거용으로 조절된 잡음 제거 필터의 게인(현재 타임에서의 조절된 잡음 제거 필터의 게인)을 나타내고,

는 이전 타임에서 잡음 제거용으로 조절된 잡음 제거 필터의 게인을 나타내며,

는 코히런스의 위상의 분산

의 임계값을 나타낸다.In Equation 6,

Represents the gain of the noise reduction filter adjusted for noise reduction (the gain of the adjusted noise reduction filter at the current time)

Represents the gain of the noise reduction filter adjusted for noise removal at the previous time,

The dispersion of the phase of the coherence

Lt; / RTI &gt;

That is, as shown in Equation (6), when the variance of the phase of the coherence is larger than the threshold value without the voice signal being included in the two channel input signals, the filter update unit 140 updates the coherence The gain of the noise canceling filter at the current time can be adjusted according to the gain value of the adjusted noise canceling filter and the variance of the phase of the coherence is smaller than the threshold value In the same case, the gain of the noise reduction filter can be adjusted based on the intensity of the coherence.

On the other hand, when the voice signal determination unit 130 determines that the voice signal is included in the two-channel input signals, the filter update unit 140 may perform a noise reduction You can adjust the gain.

For example, when the two-channel input signals include a voice signal, the filter update unit 140 updates the frequency regions of the two-channel input signals in a region where the intensity of the coherence is greater than a threshold value (The frequency region in which the signal exists) is set to a value of 1, and the gain of the noise canceling filter in the region where the coherence strength of the two-channel input signals is less than or equal to the threshold value Frequency domain) can be adjusted for noise cancellation (based on the intensity of the coherence).

Equation (7)

는 2 채널 입력 신호들에 음성 신호가 포함되는 경우에 조절된 잡음 제거 필터의 게인을 나타내고,

는 2 채널 입력 신호들의 코히런스의 세기

의 임계값을 나타내며,

는 식 6과 같이 코히런스의 세기에 기초하여 잡음 제거용으로 조절된 잡음 제거 필터의 게인을 나타낸다.In Equation 7,

Represents the gain of the noise reduction filter adjusted when the voice signal is included in the two channel input signals,

The coherence of two channel input signals

, &Lt; / RTI &gt;

Represents the gain of the noise canceling filter adjusted for noise removal based on the coherence strength as shown in Equation (6).

The fixed beamforming performing unit 150 performs fixed beamforming on the two-channel input signals. For example, the fixed beamforming unit 150 may perform fixed beamforming on the 2-channel input signals by performing a delay sum on the 2-channel input signals.

The filtering unit 160 filters the 2-channel input signals using the noise reduction filter updated by the filter updating unit 140 as described above. More specifically, the filtering unit 160 may output a signal to which the updated noise reduction filter is applied by applying the updated noise reduction filter to the signal subjected to the fixed beamforming.

Therefore, the output signal output from the filtering unit 160 may be a signal processed by the gain of the noise canceling filter whose fixed beamforming is performed as described above.

The time domain transform unit 170 transforms the signal output from the filtering unit 160 from the frequency domain to the time domain. For example, an IFFT (Inverse Fast Fourier Transform) can be used as the time domain transform unit 170, and the output signal of the frequency domain can be converted into the output signal of the time domain.

In this way, the noise canceller 100 minimizes the loss of the voice signal in the two-channel input signals and can remove only the noise signal by using the noise canceling filter adaptively adjusted in gain.

The noise canceling apparatus 100 includes a frequency domain transforming unit 110, a coherence calculating unit 120, a voice signal determining unit 130, a filter updating unit 140, a fixed beamforming performing unit 150, The filtering unit 160, and the time domain transform unit 170. However, the present invention is not limited to this, and the noise canceller 100 may be configured to include the two-channel input signals in the frequency domain Calculates the coherence of the two-channel input signals, determines whether or not the voice signal is included in the two-channel input signals based on the dispersion of the phase of the coherence, and adjusts the gain adaptively according to the determination result By updating the noise elimination filter, it is possible to provide more components or fewer components, while essentially including components that filter the two channel input signals through the updated noise elimination filter It may have a structure comprising a.

3 is a view for explaining dual microphones used in the noise canceling apparatus according to an embodiment.

Referring to FIG. 3, the dual microphones 310, 320, and 330 used in the noise canceling apparatus according to the embodiment are disposed at symmetrical positions from the sound generating point 340 in the user's body. For example, the dual microphones 310, 320, and 330 may be symmetrical in the left, right, top, bottom, or diagonal directions with respect to the mouth 340, (311, 321, 331).

4 is a flowchart illustrating a noise removal method according to an embodiment. Hereinafter, the noise canceling method will be described as being performed by the noise canceling apparatus described above with reference to FIG. Thus, the noise cancellation method may be implemented in the form of a computer program stored on a medium coupled to the computer.

Referring to FIG. 4, the noise canceling apparatus according to an exemplary embodiment converts 410-channel input signals received from dual microphones into a frequency domain from a time domain. Here, each of the dual microphones is disposed at a symmetrical position from the voice generation point in the user's body.

The noise canceller then calculates the coherence of the two channel input signals based on the Power Spectral Density (PSD) of the two channel input signals (420). For example, in step 420, the noise canceller computes the PSD of the 2-channel input signals using the recursive smoothed periodogram technique and calculates the APD of the 2-channel input signals based on the PSD of the computed 2-channel input signals Density) and CPSD (Cross Power Spectral Density) of the two-channel input signals.

The noise canceller then determines (430) whether the voice signal is included in the two channel input signals based on the variance of the phase of the coherence. For example, when the phase of the coherence is distributed to a specific value, the noise canceller may determine that the voice signal is included in the two channel input signals. On the other hand, when the phase of the coherence is not distributed to a specific value, the noise canceller can judge that the voice signal is not included in the two-channel input signals.

At this time, although not shown in the figure, the noise canceller can obtain the variance of the phase of the coherence by calculating the phase of the coherence before step 430. [

The noise removal apparatus updates the noise reduction filter based on the determination result (440).

More specifically, when the two-channel input signals do not include a speech signal, the noise canceller removes the noise signal included in the two-channel input signals, based on the strength of the coherence, The gain can be adjusted (441).

For example, in step 441, the noise canceller may adjust the gain of the noise reduction filter based on the intensity of the coherence when the variance of the phase of the coherence is less than or equal to the threshold value.

For example, in step 441, when the variance of the phase of the coherence is greater than the threshold value, the noise elimination apparatus calculates the noise level at the current time according to the gain value of the noise reduction filter adjusted based on the coherence strength at the previous time The gain of the reject filter can be adjusted.

On the other hand, when the two-channel input signals include a voice signal, the noise canceller may adjust the gain of the noise canceling filter to prevent loss of the voice signal by the noise canceling filter (442).

For example, in step 442, the noise canceller adjusts the gain of the noise cancellation filter to a value of 1 for the region where the coherence strength of the frequency regions of the 2-channel input signals is greater than the threshold, The gain of the noise canceling filter for the region where the intensity of the coherence in the frequency regions is less than or equal to the threshold value can be adjusted based on the intensity of the coherence.

At this time, although not shown in the figure, the noise canceller may calculate the intensity of the coherence before step 440. [

Thus, the noise canceller filters the two channel input signals using the updated noise cancellation filter (450). Specifically, in step 450, the noise canceller performs Fixed beamforming on the 2-channel input signals to apply the updated noise reduction filter to the signal subjected to the fixed beamforming, and then outputs the signal to which the updated noise reduction filter is applied can do.

The noise canceller may then convert the filtered signal (the signal to which the updated noise reduction filter is applied in step 450) to the time domain from the frequency domain (460).

5 is a block diagram illustrating a wearable device including a noise canceller in accordance with one embodiment.

5, a wearable device 500 according to one embodiment includes dual microphones 510 and a noise cancellation device 520, and includes more components that are worn on the user &apos; Elements. &Lt; / RTI &gt; For example, the wearable device 500 may further include a speaker, a communication module, and the like necessary for a call function.

The dual microphones 510 are disposed at symmetrical positions from the voice generation point in the user's body. For example, the dual microphones 510 may be symmetrical in the left, right, top, bottom, or diagonal direction with respect to the point of sound generation in the user's body, and may be disposed at the same distance.

The noise canceller 520 converts the two channel input signals received from the dual microphones into the frequency domain from the time domain and generates a coherence of the two channel input signals based on the PSD (Power Spectral Density) Determines whether or not a voice signal is included in the two channel input signals based on the variance of the phase of the coherence and updates the noise reduction filter based on the determination result, Filter two-channel input signals.

Since the noise eliminator 520 is composed of the same components as the noise eliminator described with reference to FIG. 1 and performs the noise elimination method described above with reference to FIG. 4, the noise eliminator 520 A detailed description thereof will be omitted.

The apparatus described above may be implemented as a hardware component, a software component, and / or a combination of hardware components and software components. For example, the apparatus and components described in the embodiments may be implemented within a computer system, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA) , A programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For ease of understanding, the processing apparatus may be described as being used singly, but those skilled in the art will recognize that the processing apparatus may have a plurality of processing elements and / As shown in FIG. For example, the processing unit may comprise a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of the foregoing, and may be configured to configure the processing device to operate as desired or to process it collectively or collectively Device can be commanded. The software and / or data may be in the form of any type of machine, component, physical device, virtual equipment, computer storage media, or device , Or may be permanently or temporarily embodied in a transmitted signal wave. The software may be distributed over a networked computer system and stored or executed in a distributed manner. The software and data may be stored on one or more computer readable recording media.

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions to be recorded on the medium may be those specially designed and configured for the embodiments or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. For example, it is to be understood that the techniques described may be performed in a different order than the described methods, and / or that components of the described systems, structures, devices, circuits, Lt; / RTI &gt; or equivalents, even if it is replaced or replaced.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

Claims (15)

In a dual microphone based noise cancellation method,
Converting the two-channel input signals received from the dual microphones, each of the dual microphones being located at a symmetrical position from a point of sound generation in the user's body, from the time domain to the frequency domain;
Calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals;
Determining whether a voice signal is included in the two-channel input signals based on a variance of the phase of the coherence;
Updating the noise reduction filter based on the determination result; And
Filtering the two-channel input signals using the updated noise reduction filter
Lt; / RTI &gt;
Wherein updating the noise reduction filter comprises:
Wherein the noise elimination filter removes a noise signal included in the two-channel input signals when the two-channel input signals do not contain a voice signal, the gain of the noise elimination filter based on the strength of the coherence Adjusting step
/ RTI &gt;
Wherein the step of adjusting the gain of the noise-
If the variance of the phase of the coherence is greater than a threshold value, the previous time-the previous time-refers to the time at which the performed noise cancellation method was performed immediately before the currently performed noise cancellation method-based on the strength of the coherence Adjusting a gain of the noise reduction filter in a current time according to a gain value of a noise reduction filter adjusted by the noise reduction filter, and the current time means a time at which a current noise reduction method is being performed; or
Adjusting the gain of the noise reduction filter based on the intensity of the coherence when the variance of the phase of the coherence is less than or equal to the threshold value
/ RTI &gt; wherein the noise reduction step comprises the steps of: delete delete The method according to claim 1,
Wherein updating the noise reduction filter comprises:
Adjusting a gain of the noise canceling filter to prevent loss of the voice signal by the noise canceling filter when the voice signal is included in the two channel input signals
&Lt; / RTI &gt; 5. The method of claim 4,
Wherein the step of adjusting the gain of the noise-
Adjusting a gain of the noise cancellation filter to a value of 1 for a region where the intensity of the coherence is greater than a threshold value among frequency regions of the 2-channel input signals; And
Adjusting the gain of the noise reduction filter for an area where the intensity of the coherence is less than or equal to the threshold among the frequency areas of the two channel input signals based on the intensity of the coherence
&Lt; / RTI &gt; The method according to claim 1,
Wherein the step of determining whether a voice signal is included in the two-
Determining that the voice signal is included in the two-channel input signals when the phase of the coherence is distributed to a specific value; or
Determining that the voice signal is not included in the two-channel input signals when the phase of the coherence is not distributed to the specific value
/ RTI &gt; wherein the noise reduction step comprises the steps of: The method according to claim 1,
Calculating a phase of the coherence; And
Calculating a strength of the coherence
&Lt; / RTI &gt; The method according to claim 1,
Wherein the step of calculating the coherence of the two-
Calculating a PSD of the two-channel input signals using a recursive smoothed periodogram technique; And
Channel input signals in a ratio of APSD (Auto Power Spectral Density) and CPSD (Cross Power Spectral Density) of the 2-channel input signals based on the PSD of the calculated 2-channel input signals. Steps to define
&Lt; / RTI &gt; The method according to claim 1,
Wherein the filtering of the two channel input signals comprises:
Performing fixed beamforming on the 2-channel input signals;
Applying the updated noise reduction filter to the signal subjected to the fixed beamforming; And
And outputting a signal to which the updated noise cancellation filter is applied
&Lt; / RTI &gt; 10. The method of claim 9,
Converting the output signal from the frequency domain to the time domain
&Lt; / RTI &gt; A computer program stored in a recording medium for executing a dual microphone-based noise canceling method in combination with a computer implementing an electronic device,
The noise cancellation method includes:
Converting the two-channel input signals received from the dual microphones, each of the dual microphones being located at a symmetrical position from a point of sound generation in the user's body, from the time domain to the frequency domain;
Calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals;
Determining whether a voice signal is included in the two-channel input signals based on a variance of the phase of the coherence;
Updating the noise reduction filter based on the determination result; And
Filtering the two-channel input signals using the updated noise reduction filter
Lt; / RTI &gt;
Wherein updating the noise reduction filter comprises:
Wherein the noise elimination filter removes a noise signal included in the two-channel input signals when the two-channel input signals do not contain a voice signal, the gain of the noise elimination filter based on the strength of the coherence Adjusting step
/ RTI &gt;
Wherein the step of adjusting the gain of the noise-
If the variance of the phase of the coherence is greater than a threshold value, the previous time-the previous time-refers to the time at which the performed noise cancellation method was performed immediately before the currently performed noise cancellation method-based on the strength of the coherence Adjusting a gain of the noise reduction filter in a current time according to a gain value of a noise reduction filter adjusted by the noise reduction filter, and the current time means a time at which a current noise reduction method is being performed; or
Adjusting the gain of the noise reduction filter based on the intensity of the coherence when the variance of the phase of the coherence is less than or equal to the threshold value
The computer program being stored on a recording medium. In a dual microphone based noise canceller,
A frequency domain conversion unit for converting the two channel input signals received from the time domain into the frequency domain from the dual microphones, each of the dual microphones being disposed at a symmetrical position from the voice generation point of the user's body;
A coherence calculator for calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals;
A voice signal determination unit for determining whether a voice signal is included in the two-channel input signals based on a dispersion of the phase of the coherence;
A filter updating unit updating the noise reduction filter based on the determination result; And
And a filtering unit for filtering the two-channel input signals using the updated noise elimination filter,
Lt; / RTI &gt;
Wherein the filter update unit comprises:
Wherein the noise elimination filter removes a noise signal included in the two-channel input signals when the two-channel input signals do not contain a voice signal, the gain of the noise elimination filter based on the strength of the coherence Regulating,
If the variance of the phase of the coherence is greater than a threshold value, the previous time-the previous time-refers to the time at which the performed noise cancellation method was performed immediately before the currently performed noise cancellation method-based on the strength of the coherence Wherein the current time is a time at which the current noise cancellation method is being performed, the gain of the noise cancellation filter being adjusted according to the gain value of the adjusted noise cancellation filter,
And adjusts the gain of the noise reduction filter based on the intensity of the coherence when the variance of the phase of the coherence is less than or equal to the threshold value. 13. The method of claim 12,
A fixed beamforming unit for performing fixed beamforming on the 2-channel input signals,
Further comprising:
Wherein the filtering unit comprises:
And applying the updated noise reduction filter to the signal subjected to the fixed beamforming and outputting a signal to which the updated noise reduction filter is applied. 14. The method of claim 13,
A time domain transformer for converting the output signal from the frequency domain to the time domain,
Further comprising a noise canceling unit. 1. A wearable device worn on a user's body to provide a call function,
Dual microphones positioned at symmetrical positions from the point of sound generation of the user's body; And
The noise eliminator connected to the dual microphones
Lt; / RTI &gt;
The noise cancellation device includes:
A frequency domain converter for converting the two channel input signals received from the dual microphones into a frequency domain from a time domain;
A coherence calculator for calculating a coherence of the two-channel input signals based on PSD (Power Spectral Density) of the two-channel input signals;
A voice signal determination unit for determining whether a voice signal is included in the two-channel input signals based on a dispersion of the phase of the coherence;
A filter updating unit updating the noise reduction filter based on the determination result; And
And a filtering unit for filtering the two-channel input signals using the updated noise elimination filter,
Lt; / RTI &gt;
Wherein the filter update unit comprises:
Wherein the noise elimination filter removes a noise signal included in the two-channel input signals when the two-channel input signals do not contain a voice signal, the gain of the noise elimination filter based on the strength of the coherence Regulating,
If the variance of the phase of the coherence is greater than a threshold value, the previous time-the previous time-refers to the time at which the performed noise cancellation method was performed immediately before the currently performed noise cancellation method - based on the strength of the coherence Wherein the current time is a time at which the current noise cancellation method is being performed, the gain of the noise cancellation filter being adjusted according to the gain value of the adjusted noise cancellation filter,
And adjusts the gain of the noise reduction filter based on the intensity of the coherence when the dispersion of the phase of the coherence is less than or equal to the threshold value.

Images