KR101644058B1 - System for controlling active acoustic echo using single sensor and mothod thereof - Google Patents

Abstract

The present invention relates to an active acoustic echo cancellation sound control system capable of coping with an active acoustic echo sound and a method thereof and a single sensor based active echo sound control system according to an embodiment of the present invention is characterized in that an incident sound and a reflection sound are superimposed An acoustic receiving sensor for measuring a signal that has been received; An estimator for estimating the incident sound and the reflection sound using a previously measured acoustic path using white noise; A control signal for minimizing the echo from the estimated incident sound, and outputting the generated control signal to the control speaker.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an active acoustic echo cancellation control system using a single sensor,

The present invention relates to a system and method for active acoustic echo cancellation using a single sensor.

Acoustic noise problems are increasing in frequency due to the increase in industrial equipment. Much of these industrial equipment has periodic motion, which makes the noise concentrated in the low frequency band. Particularly, such noise causes not only discomfort and fatigue but also acts as a factor that hinders the efficiency of operation, so that noise control is required. Passive methods for noise control have drawbacks that are inefficient due to high cost and space constraints. Recently, an active noise control method for efficiently coping with various noise environments has been proposed.

Various methods for controlling acoustic reflections have been proposed. First, the reflection coefficient of the reflection surface is measured in advance by the method using the analog circuit, and then the reverse phase is applied to the signal measured by the sensor. However, it is limited to acoustic echo control for signals below 1kHz, and a near field effect problem arises. In addition, since it can not actively cope with noise, there is a disadvantage that an analog circuit must be manufactured every time when it is applied to a new sound system.

Next, a method is proposed that can actively cope with noise by controlling the impedance represented by the ratio of sound pressure and speed per unit area in an acoustic system. To estimate the incident sound on the reflection surface, the impedance of the reflection surface and the velocity of the medium are used. To minimize the error, the speedometer is placed on the reflection surface and the acoustic impedance of the reflection surface is measured in advance using two microphones. The control algorithm operates to minimize the estimated echo and the estimated echo using the difference in the signal measured by the sound pressure sensor located close to the reflective surface. This method improves the control efficiency of the full-band echo sound as well as the specific frequency band by combining with a passive module composed of a sound absorbing material for increasing the absorption rate. However, the sensors for measuring the incident sound measure both the incident sound as well as the control sound. At this time, the feedback sound signal measured by the sensor again acts as a factor that hinders the control performance. In order to avoid the influence of this feedback sound, a beamforming based active noise control is proposed which distinguishes between the incident sound and the reflected sound from two adjacent sound pressure sensors. At this time, beamforming is a technique of extracting a signal having a direction by compensating a time delay between adjacent sensors, and has an advantage of improving a signal ratio with respect to a specific direction. The acoustic reflections can be efficiently controlled by operating the adaptive algorithm to minimize the error by using the reference signal and the error signal of the active noise control algorithm, respectively, which are estimated through the beamforming.

As described above, the conventional technique uses a beam-forming technique to solve the acoustic feedback problem. However, the thickness of the tile module must be increased because the two sensors must maintain sufficient spacing to enable beamforming in an environment for controlling acoustic reflections from the surface, such as a tile. Since the beamforming performance is highly dependent on the position of the two sensors, it can act as an obstacle to the beamforming performance even with the fine movement of the sensor, which hinders the performance of the acoustic echo control.

In order to solve such a problem, in the present invention, an acoustic path previously estimated from a signal measured by a single sensor is applied, and then an acoustic incident sound and an echo sound are effectively distinguished from each other, and then each of them is used as a reference signal and an error signal for acoustic echo control And more particularly, to an active acoustic echo cancellation control system and a method thereof that actively control acoustical echo sound to cope with an active acoustic echo sound.

According to an aspect of the present invention, there is provided a single sensor-based active echo cancellation control system including: an acoustic sensor for measuring a signal in which an incident sound and a reflection sound are superimposed; An estimator for estimating the incident sound and the reflection sound using a previously measured acoustic path using white noise; A control signal for minimizing the echo from the estimated incident sound, and outputting the generated control signal to the control speaker.

를

를 통해 추정하며, 상기 반향음

을

식을 통해 추정하며,According to an embodiment of the present invention, the acoustic reception sensor measures a signal in which an acoustic incident sound, an echo sound, and a control signal are superimposed,

To

, And the echo

of

, Respectively,

은 제어 신호이고,

는 상기 제어 스피커와 상기 음향 수신 센서 사이의 음향 경로이고,

은 센서에서 집음되는 신호,

는 상기 음향 수신 센서와 반사면 사이의 궤환 경로이고,

는 미리 추정된 궤환 경로를 나타낼 수 있다.here,

And a control signal,

Is an acoustic path between the control speaker and the acoustic reception sensor,

A signal collected at the sensor,

Is a feedback path between the acoustic reception sensor and the reflective surface,

Can represent a previously estimated feedback path.

 A single sensor based active reverberation sound control method according to an embodiment of the present invention is a single sensor based active reverberation sound control method comprising: measuring a signal in which an incident sound and a reflection sound are superimposed; Estimating the incident sound and the echo sound using a previously measured acoustic path using white noise; Generating a control signal for minimizing the echo from the estimated incident sound, and outputting the generated control signal to the control speaker.

The active acoustic echo cancellation control system and method according to an embodiment of the present invention can be applied to an acoustic echo cancellation control system and method in which an acoustic path previously estimated from a single sensor is applied, It is possible to cope with the active acoustic incident sound robustly by actively controlling the acoustic echo sound.

FIG. 1 is a diagram illustrating an example of a signal measured by an acoustic receiving sensor according to the present invention.
2 is a diagram illustrating a single sensor based active acoustic echo cancellation control system in accordance with an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings so that those skilled in the art can easily carry out the technical idea of the present invention. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly explain the present invention, parts not related to the description are omitted, and like parts are denoted by similar reference numerals throughout the specification.

FIG. 1 is a diagram illustrating an example of a signal measured by an acoustic receiving sensor according to the present invention.

, 반사면에 의해 주 소음 신호가 반사된 반향음

와 반향음 제어신호

의 합으로 나타난다. 모든 음향 경로가 추정 가능하다고 가정하면, 음향 수신 센서(S)에 측정되는 신호

는 수학식 1과 같이 표현할 수 있다.1, the signal measured by the acoustic receiving sensor S is a main noise signal (incident sound)

, Reflections of the main noise signal reflected by the reflecting surface

And the reverberation control signal

. Assuming that all acoustic paths are presumable, the signal received at the acoustic receiving sensor S

Can be expressed by Equation (1).

는 주 소음 신호(입사음) ,

는 반사면에 의해 주 소음 신호가 반사된 반향음,

는 반향음 제어신호,

는 음향 수신 센서(S)와 반사면 사이의 궤환 경로를 나타낸다. here,

Is the main noise signal (incident sound),

Is a reflection sound in which the main noise signal is reflected by the reflection surface,

Is a reflection signal,

Represents a feedback path between the acoustic reception sensor S and the reflective surface.

과 반향음

에 대해서 정리하면 수학식 2-3과 같다. From Equation 1,

And reflections

The following equation (2-3) is obtained.

가 모듈화되어 있으므로 시불변한 특성을 보이므로 사전에 백색 잡음(white noise)를 이용하여 상기 입사음과 반향음을 추정(분리) 가능하다. 외부 스피커를 이용하여 모듈과 가까운 거리에서 백색 잡음을 출력하면 음향 수신 센서(S)에서 측정되는 신호는 백색 잡음이 음향 경로

와

를 거친 신호가 결합된 형태로 얻어진다. 이때 두 음향 경로

와

는 임펄스와 유사한 특성을 갖는다면, 두 음향 경로는 시간의 중첩이 발생하지 않으며 각각의 음향 경로의 분리가 용이하므로

를 추정할 수 있다. 수학식 4는 z-domain으로 표현한 반사파에 대한 수학식 3을 시간 영역으로 다시 표현한 것이다.A feedback path between the tone receiving sensor S and the reflecting surface

Since the time constant is modulated, the incident sound and the reflection sound can be estimated (separated) by using white noise in advance. If white noise is output at a distance from the module using an external speaker, the signal measured by the acoustic receiving sensor (S)

Wow

Is obtained as a combined signal. At this time,

Wow

Is similar to an impulse, the two acoustic paths do not overlap in time, and each acoustic path can be easily separated

Can be estimated. Equation (4) is a time domain representation of Equation (3) for the reflected wave in the z-domain.

시간에 대한 반향음

을 얻기 위해서는

과

의 컨벌루션 연산이 필요하므로 인과 시스템의 구현이 불가능하다. 이때,

는 시간 지연 시스템이므로 첫 번째 계수는 0이라고 가정하여

수학식 4를 다시 표현하면 수학식 5와 같다.In Equation 4,

Reflections on time

To get

and

It is impossible to implement a causal system. At this time,

Is a time delay system and therefore assumes that the first coefficient is zero

Equation (4) can be rewritten as Equation (5).

는 반향음 벡터,

는 시간 영역에서의 음형 수신 센서(S)와 반사면 사이의 궤환 경로를 나타낸다. here,

Is a reflection vector,

Represents a feedback path between the sound receiving sensor S and the reflective surface in the time domain.

와

는 각각

와

로 표현되며 음향 수신 센서(S)에서 측정되는 신호는 수학식 6과 같이 표현할 수 있다.From the impulse hypothesis of the preceding two acoustic paths

Wow

Respectively

Wow

And the signal measured by the acoustic receiving sensor S can be expressed by Equation (6).

는 음향 수신 센서(S)에서 측정되는 신호,

은 입력 백색 잡음,

와

는 외부 스피커와 음향 수신 센서(S)간의 음압 손실 계수 및 시간 지연,

와

는 궤환 경로의 음압 손실 계수 및 시간 지연을 나타낸다. here,

A signal measured by the acoustic receiving sensor S,

The input white noise,

Wow

A sound pressure loss coefficient and a time delay between the external speaker and the sound receiving sensor S,

Wow

Represents the sound pressure loss coefficient and the time delay of the feedback path.

를 만족하는

는

와

이 결합된 형태로 얻어지며 수학식 7과 같이 표현할 수 있다.Using the Wiener Filter

Satisfy

The

Wow

Can be expressed as Equation (7).

와

는 외부 스피커와 음향 수신 센서(S)간의 음압 손실 계수 및 시간 지연,

와

는 궤환 경로의 음압 손실 계수 및 시간 지연을 의미한다.

가 0 보다 큰 값이기 때문에 최적해

로 얻은 서로 다른 임펄스 위치에 0을 넣음으로써 두 개의 임펄스 응답을 분리 한 뒤 FFT를 이용하여 주파수 영역으로 변환하면 다음과 같이

와

의 음향 경로는 각각 수학식 8-9와 같다.In Equation (7)

Wow

A sound pressure loss coefficient and a time delay between the external speaker and the sound receiving sensor S,

Wow

Means the sound pressure loss factor and time delay of the feedback path.

Is greater than 0,

, The two impulse responses are separated by putting 0 at different impulse positions, and then converted to the frequency domain using FFT,

Wow

Are respectively expressed by Equation 8-9.

를 수행함으로써 반사 계수를 포함한 궤환

를 얻을 수 있다.At this time,

The feedback including the reflection coefficient

Can be obtained.

When a control signal for controlling the reverberation is generated, equation (1) representing a signal measured by a single sensor (acoustic receiving sensor) due to an incident sound and a reflection sound can be expressed in a form including a control signal as shown in equation (10).

이 되도록 동작하는 시스템과 반향음만을 제어하기 위한 능동 음향 반향음 제어의 관점에서는 음압의 크기를 입사파와 동일하게

하는 두 가지 관점이 있다.From the viewpoint of active noise control, the magnitude of the sound pressure measured by the acoustic receiving sensor is set to zero

And the active acoustic reverberation control for controlling only the reverberation, the magnitude of the sound pressure is equal to the incident wave

There are two points of view.

From the first point of view, Equation (10) can be expressed as Equation (11).

 From Equation (11), the control signal is summarized as Equation (12).

는 반사면의 반사계수,

는 wave number,

는 음향 수신 센서와 반사면사이의 시간 지연이다. In Equation (12)

Is the reflection coefficient of the reflecting surface,

Wave number,

Is the time delay between the acoustic receiving sensor and the reflective surface.

이 필요하므로 구현 불가능한 제어 시스템이 됨을 알 수 있다.At this time, in order to obtain an ideal control signal from Equation (12)

It becomes impossible to implement the control system.

을 이용하여 수학식 11을 제어 신호에 대해서 정리하면 수학식 13과 같이 표현할 수 있다.The second perspective

(11) can be summarized with respect to the control signal as shown in Equation (13).

로 첫 번째 관점과 다르게 제어 시스템의 구현이 가능하다. 따라서, 본 발명에서는 센서에서 집음되는 신호가 입사음와 같은 음압을 가지도록 구현함으로써 반향음를 제어한다.From equation (13), the ideal control signal is

The control system can be implemented differently from the first viewpoint. Accordingly, in the present invention, the echo can be controlled by realizing the signal collected by the sensor to have the same sound pressure as the incident sound.

Equation (11) can be expressed in terms of z-domain and time domain as shown in Equation (14-15).

를 거친 신호의 차이로 구할 수 있다. 따라서 반향음 제어를 위한 적응 필터의 참조 신호인 입사음은 제어 신호와 사전에 추정된 궤환 경로

를 이용하여 얻을 수 있으며 제어 신호

와 추정된 반향음

는 수학식 16-17과 같다.As can be seen from Equation (15), the incident sound is a time-delayed control signal in a signal picked up by the sensor,

Can be obtained by the difference of the rough signals. Therefore, the incident sound, which is the reference signal of the adaptive filter for controlling the reflection sound,

And the control signal

And estimated echo

(16-17).

는 제어 스피커와 음향 수신 센서 사이의 음향 경로이며,

는

에서 첫 번째 계수를 제거한 궤환 경로이다. 수학식 16과 17를 이용하여 추정된 입사음은 수학식 18과 같다.

Is the acoustic path between the control speaker and the acoustic reception sensor,

The

Is the feedback path from which the first coefficient is removed. The incident sound estimated using Equations (16) and (17) is shown in Equation (18).

과

는 각각 능동 음향 반향음 제어를 위한 적응 필터의 참조 신호와 오차 신호로 하여 계수 벡터를 적응적으로 갱신한다.For reverberation control

and

Adaptively updates the coefficient vector with the reference signal and the error signal of the adaptive filter for active acoustic echo cancellation control, respectively.

FIG. 2 illustrates an active acoustic echo cancellation control system based on a single sensor. As shown in FIG. 2, the active acoustic echo cancellation control system includes: an acoustic sensor 100 for measuring a signal in which an incident sound and a reflection sound are superimposed; An estimator (110) for estimating (separating) the incident sound and the reflection sound using the acoustic path measured in advance using the white noise; (For example, an adaptive filter) 120 for generating a control signal for minimizing the reverberation from the estimated (separated) incident sound and outputting the control signal to the control speaker 200 .

At this time, the coefficient update equation of the filtered-x LMS is expressed by Equation 19-20.

은 적응 필터,

은 적응필터의 수렴 계수,

은 참조 벡터,

은 filtered-x LMS알고리즘을 정규화하기 위해 사용하였으며, 상기 적응 필터의 갱신을 위한 참조 벡터

의 smoothing 인자

를 갖는 IIR 필터(infinite impulse response filter)를 통하여 파워를 추정한 값이다.

An adaptive filter,

Is the convergent coefficient of the adaptive filter,

Is a reference vector,

Is used to normalize the filtered-x LMS algorithm, and the reference vector for updating the adaptive filter

Smoothing factor

And an IIR filter (infinite impulse response filter).

It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be construed according to the following claims, and all technical ideas within the scope of equivalents should be construed as falling within the scope of the present invention.

Claims (6)

An acoustic receiving sensor for measuring a signal in which an incident sound and a reflection sound are superimposed;
An estimator for estimating the incident sound and the reflection sound using a previously measured acoustic path using white noise;
And a control unit for generating a control signal for minimizing the echo from the estimated incident sound and outputting the generated control signal to the control speaker,
Wherein the acoustic-
A signal in which an acoustic incident sound, a reflection sound and a control signal are superimposed is measured,
The signal

To

, Respectively,
The reflections

of

, Respectively,
here,

And a control signal,

Is an acoustic path between the control speaker and the acoustic reception sensor,

A signal collected at the sensor,

Is a feedback path between the acoustic reception sensor and the reflective surface,

The

Wherein the first and second coefficients represent a feedback path from which the first coefficient is removed. delete 2. The method according to claim 1,

silver

Wherein the first and second sensors are based on a single sensor-based active reverberation sound control system. delete delete A method of controlling a single sensor based active reverberation sound,
Measuring a signal in which an incident sound and a reflection sound are superimposed;
Estimating the incident sound and the echo sound using a previously measured acoustic path using white noise;
Generating a control signal for minimizing the echo from the estimated incident sound, and outputting the generated control signal to the control speaker,
Wherein the step of measuring a signal in which the incident sound and the reflection sound are overlapped,
Measuring a signal in which an acoustic incident sound, a reflection sound and a control signal are superimposed,
The signal

To

, Respectively,
The reflections

of

, Respectively,
here,

And a control signal,

Is an acoustic path between the control speaker and the acoustic reception sensor,

A signal collected at the sensor,

Is a feedback path between the acoustic reception sensor and the reflective surface,

The

And the first coefficient is removed from the feedback signal.

Images