JP3460229B2 - Noise reduction device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise reducing device, and more particularly to a noise reducing device capable of reducing noise components of a microphone output and obtaining a stereophonic voice output by using an omnidirectional microphone. .

[0002]

2. Description of the Related Art Many microphones have a structure in which a change in sound pressure of sound waves is converted into mechanical vibration of a diaphragm and an electroacoustic conversion system is operated based on the vibration. Therefore, when the sound is picked up by the microphone, if the diaphragm is affected by some factor, noise is generated.

If the above-mentioned factor is wind, noise due to wind [hereinafter referred to as wind noise] is generated, and if the above-mentioned factor is vibration externally applied, noise due to vibration [hereinafter, referred to as noise due to vibration] This is called vibration noise.

Examples of conventional techniques for reducing the above wind noise include the following. (1) Use of windscreen (windshield) (2) Use of electrical / acoustic high-pass filter (3) Adoption of configuration showing omnidirectionality in low range

Further, as a conventional technique for reducing the above-mentioned vibration noise, there is, for example, the following. (1) Adoption of anti-vibration mechanism (2) Adoption of omnidirectional microphone element (3) Analog noise cancellation method

[0006]

The conventional techniques for reducing the wind noise described above have the following problems, respectively. In contrast to (1) ... against the miniaturization of equipment. Generally, the larger the outer dimensions of the windshield and the larger the distance between the microphone and the inner wall of the windshield, the smaller the wind noise.

With respect to (2) ... The sound collection quality is deteriorated. Since the wind noise is mainly composed of low frequency components, it is effective to cut the low frequency components. However, in this case, not only the wind noise but also the low-frequency components of the voice are cut.

In contrast to (3) ... The level at which wind noise is reduced is insufficient. Wind noise levels are lower with omnidirectional microphones than with directional microphones. However,
Actually, due to the influence of the housing around the microphone,
Just by adopting "a configuration showing omnidirectionality in the low range",
It will not be low enough.

Therefore, in a current situation where a device equipped with a microphone is further downsized and higher sound collection quality is desired, wind noise can be further reduced only by the above-mentioned conventional technique. Is getting harder.
This also applies to vibration noise.

By the way, conventionally, it has been impossible to obtain a stereophonic sound output using only an omnidirectional microphone.

Therefore, a first object of the present invention is to provide a noise reduction device which can be miniaturized and can reliably remove noise. A second object of the present invention is to provide a noise reduction device that can obtain a stereoscopic sound output by using an omnidirectional microphone.

[0012]

According to a first aspect of the present invention, there is provided a pair of microphones arranged close to each other, a first subtraction means for subtracting outputs of the pair of microphones, and a pair of microphones.
One of the microphone outputs
Phone output as the main input and the first subtraction means
These outputs are used as the reference input and are based on the main input and the reference input.
Adaptive processing means for performing adaptive processing based on
Phase- shifting means for shifting the phase of one of the microphone outputs, and the output of the pair of microphones
The output of the other microphone and the output of the phase shifting means
Second subtraction means for subtraction and bass sound output from the adaptive processing means
A low-pass filter means for passing a band component, and a second reduction filter.
High pass filter that passes the high frequency component of the output of the computing means
Filter means and the output of the low pass filter means and the high pass filter.
A noise reduction device having a means for combining the output of
It is a place .

According to a third aspect of the present invention, a pair of microphones arranged in close proximity to each other, a pair of low-pass filter means for passing a low-frequency component of the output of the pair of microphones, and a pair of output of the pair of microphones. One of a pair of high-pass filter means for passing a high-frequency component, a first subtracting means for subtracting outputs of the pair of low-pass filter means, and an output of the pair of low-pass filter means. The output of the low-pass filter means is used as a main input, and the first
Of the output of the subtracting means as a reference input, the adaptive processing means for performing adaptive processing based on the main input and the reference input, and the output of the pair of high-pass filter means Phase shift means for shifting the phase of the output, second subtraction means for subtracting the output of the other high pass filter means and the output of the phase shift means of the outputs of the pair of high pass filter means, and It is a noise reduction device comprising means for combining the output of the processing means and the output of the second subtraction means.

[0014]

The operation of the noise reduction device according to claim 1 will be described. The outputs of the one and the other microphones include a voice signal component and a noise component. The first difference signal is formed by subtracting the output of one microphone from the output of the other microphone. This first difference signal has only a noise component, and the noise component is separated.

As a result, the output from one microphone contains a voice signal component and a noise component, and the output from the other microphone contains only a noise component. The output containing the above-mentioned voice signal component and noise component is the main input, and the output containing only the noise component is the reference input.

The reference input is adaptively processed to be equal to the noise component of the main input. Then, from the reference input that is processed adaptively is subtracted from the primary input, of the main input, only a noise component is minimized, that is, canceled is <br/>. Then, the first signal component, that is, the omnidirectional audio signal component in the low frequency band is separated from this audio signal.

The pair of microphones have a predetermined interval.
Further, the output of the other microphone of the pair of microphones is phase-shifted.

By subtracting the output of the other phase-shifted microphone from the output [main input] of the one microphone, a directional second difference signal is formed. Then, from this second difference signal, a second signal component,
That is, the directional and middle-range and high-range signal components are separated.

The first and second signal components are combined into an output audio signal. As a result, the output obtained from the noise reduction device is assumed to be unidirectional.

The operation of the noise reduction device according to claim 2 will be described. Two systems of the noise reduction device described in claim 1 are provided. Of the two systems, one system is used as the output of the L channel microphone, and the other system is used as the output of the R channel microphone. As a result, it is possible to obtain a stereophonic sound output.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. In the configuration of FIG. 1, a pair of microphones 1 and 2 arranged close to each other collects ambient sound, converts it into an electric signal, and outputs the electric signal. The microphones 1 and 2 are microphones of the same type, and
As shown in (1), since they are arranged close to each other, the same voice and noise are converted into electric signals and output.

The arrangement of the microphones 1 and 2 is shown in FIG. The main axes of the microphones 1 and 2 are on the same straight line, and the sound receiving surfaces 1a of the microphones 1 and 2 are
2a are arranged facing each other with a distance d therebetween with a separator 20 in between.

When the microphones 1 and 2 are arranged in the state as shown in the figure, a predetermined phase difference occurs between the audio signals input to the microphones 1 and 2, and the microphones 1 and 2 have a low phase difference. It is known that among the outputs in the range, the voice signal components have almost the same phase, the wind noise components have substantially no correlation, and the vibration noise components have mutually opposite phases.

In FIG. 1, the electric signal output from the microphone 1 is supplied to the A / D conversion circuit 3, and the electric signal output from the microphone 2 is supplied to the A / D conversion circuit 4.

In the A / D conversion circuits 3 and 4, electric signals supplied from the microphones 1 and 2 are converted into digital signals. The digital signal converted by the A / D conversion circuit 3 is the main input represented by (S + n). Also, A /
The digital signal converted by the D conversion circuit 4 is (S + (n
*)).

In the description of this embodiment, S represents a voice signal component, and n and (n *) represent a noise component composed of a vibration noise component and a wind noise component. The noise components n and (n *) have additivity, and the noise component (n *) has a correlation with the noise component n of the main input (S + n).

The above-mentioned main input (S + n) is the adder 5,
A delay circuit 7 provided in the adaptive noise canceller 6,
Further, it is supplied to the adder 22. Further, the output of the A / D conversion circuit 4 is supplied to the phase shift circuit 21, and the signal output from the A / D conversion circuit 4 and having a negative sign is supplied to the adder 5.

In the adder 5, a negative sign is added to A /
Output of D conversion circuit 4, that is, [-(S + (n *))]
Is added to the above-mentioned main input (S + n). As a result of this addition, the voice signal component S is removed and (n- (n *))
A reference input represented by is formed. The reference input (n
-(N *)) is supplied to the adaptive filter 9 of the adaptive noise canceller 6.

In the delay circuit 7 of the adaptive noise canceller 6, the main input (S + n) is delayed for a predetermined time and then output, and supplied to the adder 8. This delay is
It corresponds to the time delay required for the calculation for the adaptive processing or the time delay in the adaptive filter 9, and can be set appropriately according to the system configuration.

In the adder 8, the output from the delay circuit 7 and
The signal Y output from the adaptive filter 9 and having a negative sign is added. The signal Y is a component similar to the noise component n in the main input (S + n), as described later.

Therefore, in the adder 8, the main input (S +
n) the noise component n is subtracted from. In other words, the noise component n of the main input (S + n) is minimized and virtually cancelled. The audio signal component S is fed back to the adaptive filter 9 and supplied to the low pass filter 11.

The low-pass filter 11 extracts a low-frequency and omnidirectional voice signal component S from the output of the adaptive noise canceller 6 and supplies it to the adder 12.

On the other hand, in the phase shift circuit 21, in order to make the characteristic of the circuit block 26 of the configuration shown in FIG. 1 a unidirectional characteristic, specifically a cardioid shape characteristic, the A / D conversion circuit 4 A predetermined amount of phase shift is applied to the output of.

FIG. 4 shows an example of the case where the phase shift circuit 21 is configured in an analog manner. This phase shift circuit 21
Is a low-pass filter including a resistor Ro and a capacitor Co. In this low pass filter, the resistance Ro
Is large, omnidirectional characteristics are exhibited, and when the resistance Ro is small, bidirectional characteristics, that is, a so-called "8-shaped" characteristic is exhibited. With this low-pass filter, the A / D conversion circuit 4
A predetermined amount of phase θ2 is given to the output from The phase-shifted signal is supplied to the adder 22.

In the adder 22, the main input (S + n),
A signal [-(S + (n
*))] Is added. As described above, the microphones 1 and 2 are arranged so that a predetermined phase difference θ1 is generated when the microphones 1 and 2 are input, and in the phase shift circuit 15, the above-mentioned signal [-
(S + (n *))] is given a predetermined amount of phase θ2.

Therefore, the phase of the signal output from the adder 22 is within the range of [θ2 ± θ1]. The phase difference [± θ1] exists in the output between the microphones 1 and 2
This is due to the relative position of the sound source with respect to the microphones 1 and 2.

Then, for example, when θ2 = θ1,
[Θ2 + θ1 = 2 × θ2] and [θ2-θ1 = 0]
Becomes On the characteristic diagram, the above-mentioned [θ2 +
θ1 = 0] corresponds to the rear end portion of the cardioid, that is, the 0 ° direction on the characteristic diagram. Then, the level also changes in accordance with the value of [θ2 ± θ1] described above, whereby a unidirectional characteristic is obtained. The output of the adder 22 is supplied to the high pass filter 17.

The high-pass filter 17 separates the high frequency signal component from the output of the adder 22. Then, the high-frequency signal component is supplied to the adder 12.

The adder 12 adds the high-frequency signal component supplied from the high-pass filter 17 to the low-frequency audio signal component S supplied from the low-pass filter 11. As a result, the audio signal component S over the entire predetermined band from the low band to the high band is synthesized. As a result, omnidirectional signal components in the low frequency range and unidirectional signal components in the middle and high frequency ranges can be obtained. The signal component is supplied to the D / A conversion circuit 13.

In the D / A conversion circuit 13, the audio signal component S represented by a digital signal is converted into an analog signal, and the analog signal is taken out from the terminal 14.

The operation of the adaptive filter 9 of the adaptive noise canceller 6 will be described below. In the adaptive filter 9,
A signal Y is formed as a component similar to the noise component n of the main input (S + n). That is, the adaptive noise canceller 6
The filter characteristics are sequentially self-adjusted so that the output of s is similar to the audio signal component S of the main input (S + n).

As the adaptive filter 9, an FIR filter type adaptive linear combiner having the configuration shown in FIG. 2 is used. In the configuration of FIG. 2, DL1 to DLL represent delay circuits, and MP0 to MPL represent coefficient multipliers. Further, 16 is an adder, and 15 and 17 are terminals.

[Z ^-1 ] in the delay circuits DL1 to DLL described above represents a delay of the unit sampling time, and W _nk supplied to the coefficient multipliers MP0 to MPL represents a weighting coefficient. Normal FIR if the weighting factor W _nk is fixed
It is a digital filter.

Here, an algorithm for adaptively operating the adaptive filter 9 will be described. Although various algorithms can be used for the calculation in the adaptive filter 9, the LMS (Least Mean Square) algorithm, which is relatively practical, and has a relatively small amount of calculation, will be described below.

[0045] Expressed input vector Xk as X _k = _{[X k X k-1 X k} -2 ······ X kL ], the output Y _k of the adaptive filter 9, Given in.

If the output of the delay circuit 7 is d _k , the difference output [residual output] ε _k is ε _k = d ^T _KO −X _k W _k . In the above equation, "d ^T _KO " represents a high frequency signal of "d _k ". In the LMS (least mean square) method, the weight vector is updated according to the following equation. W _{k + 1} = W _k +2 με _k X _{k In the} above equation, μ is a gain factor that determines the speed and stability of adaptation, a so-called step gain.

By updating the weighting vector as described above, an action is taken to minimize the output power of the system. Hereinafter, this operation will be described by formulating it. For simplicity, when the delay circuit 7 is ignored, the differential output ε from the adder 8 is ε = S + n−Y.

The expected value of the square of ε is expressed by the following equation. E [ε ² ] = E [S ² ] + E [(n−Y) ² ] + 2E
[S (n−Y)] Here, since S has no correlation with n and Y, in the above equation, E [S (n−Y)] = 0. Therefore, the expected value E [ε ² ] of the square of (ε) is expressed by the following equation. E [ε ² ] = E [S ² ] + E [(n−Y) ² ]

The adaptive filter 9 is adjusted so that E [ε ² ] is minimized, but since E [S ² ] is not affected, the following equation is obtained. Emin [ε ² ] = E [S ² ] + Emin [(n−Y) ² ]

E [S ² ] is a weighted vector of the adaptive filter 9.
The fact that E [ε ² ] is minimized means that E [(n−Y) ² ] is minimized because it is not affected by the update of the rule. Therefore, the output Y of the adaptive filter 9
Is the best least squares estimate of [n].

When E [(n−Y) ² ] is minimized,
Since [ε-S = n−Y], E [(ε−
S) ² ] is also minimized. Therefore, adjusting the adaptive filter 9 to minimize the total output power is equivalent to that the differential output ε becomes the best least-squares estimated value of the audio signal component S.

The difference output ε is generally the speech signal component S with some noise component added, but the noise component to be output is given by (n−Y), so E [(n−Y) )
² ] is equivalent to maximizing the output signal-to-noise ratio.

In this embodiment, the circuit block 26 having the structure shown in FIG. 1 is provided in two systems as shown in FIG. 5, so that even the omnidirectional microphones 1 and 2 can output a stereo sound. You are able to get it.

In FIG. 5, each of the L and R channel circuit blocks 26 and 27 has a unidirectional characteristic.
By taking out the audio output from the circuit blocks 26 and 27 of the L and R channels from the terminals 14e and 14f, it is possible to give the audio output a stereo feeling.

The connection states of the circuit blocks 26 and 27 and the microphones 1 and 2 are not limited to the illustrated example. For example, as shown in FIG. 6, the microphones 1 and 2 may be commonly used, and the outputs of the microphones 1 and 2 may be supplied to the circuit blocks 26 and 27, respectively.

In this case, the channels connected to the microphones 1 and 2 are opposite to each other in the circuit blocks 26 and 27. That is, in the L-channel circuit block 26, the microphones 1 and 2 and the A / D conversion circuits 3 and 4 are shown in FIG.
In the R-channel circuit block 27, the microphone 1 and the A / D conversion circuit 4 are connected, and the microphone 2 and the A / D conversion circuit 3 are connected.

Further, in FIG. 5 and FIG. 6, for the low frequency band, for example, the output of the L channel circuit block 26 can be used.

FIG. 7 shows a first modification of the embodiment. This first modified example is different from the above-described first embodiment in that an adder 41 is arranged between the A / D conversion circuit 3 and the input side of the adaptive noise canceller 6, and the A / D conversion is performed.
That is, the adder 42 is arranged between the conversion circuit 4 and the input side of the adaptive noise canceller 6.

In this case, in the adder 41, both digital signals supplied from the A / D conversion circuits 3 and 4 are given a plus sign. However, in the adder 42, the digital signal supplied from the A / D conversion circuit 3 is given a positive sign, and the digital signal supplied from the A / D conversion circuit 4 is given a negative sign. There is.

Therefore, at the output of the adder 41, the voice signal component S doubled in level and the noise component (n + (n
*)) Is included, and these added values, that is, [2S
+ (N + (n *))] is the main input. The main input [2S + (n + (n *))] is taken out from the terminal 43.

Further, the output of the adder 42 is removed because the voice signal component S is in phase, so that the noise component (n- (n
*)) Only, and the noise component (n- (n *)) is used as a reference input. The reference input (n- (n *)) is the terminal 44.
Taken from. Further, among the noise components, the vibration noise components have opposite phases to each other and are therefore cancelled, and as a result, almost only the wind noise component is left. This makes it possible to further enhance the noise reduction effect.

When this first modification is used, A /
The output extracted from the point PA on the output side of the D conversion circuit 3 is supplied to the adder 22 shown in the embodiment, and the output extracted from the point PB on the output side of the A / D conversion circuit 4. Are supplied to the phase shift circuit 21 shown in one embodiment. The contents of other configurations, actions, effects, etc. are the same as those in the above-mentioned one embodiment.
The same parts as those in the first embodiment are designated by the same reference numerals, and overlapping description will be omitted.

FIG. 8 shows a second modification of the embodiment. This second modified example is different from the above-described first embodiment in that only the processing of the adaptive noise canceller 6 is performed with a digital signal, and other signal processing is performed with an analog signal. Therefore, the adders 5, 12, 22 shown in the embodiment are changed to analog adders 36, 39, 37, and the phase shift circuit 21 is changed to the analog phase shift circuit 3.
It has been changed to 5.

Further, by providing analog low-pass filters 31 and 32 on the output sides of the microphones 1 and 2, respectively, only low-frequency signal components are supplied to subsequent circuit blocks. Then, the above-mentioned low-pass filter 31,
Both the input and output signal components of 32 are extracted, and the adders 33, 3
That is, the high-frequency signal component is extracted by calculating the difference in 4, and the high-frequency signal component is supplied to the adder 37, the phase shift circuit 35, and the like. In addition, other configurations, actions,
The contents of the effects and the like are the same as those in the above-described first embodiment, and the same parts as those in the first embodiment are designated by the same reference numerals and the duplicate description will be omitted.

According to this embodiment, the microphone 1,
Since the two main axes are on the same straight line and the sound receiving surfaces 1a and 2a are arranged facing each other with the distance d interposed therebetween, a phase difference occurs between the input audio signals. Main input (S + n) at adaptive noise canceller 6
The noise component n is subtracted from. In other words, the noise component n of the main input (S + n) is minimized and substantially canceled, only the audio signal component S is taken out, and the low-pass filter 11 produces a low-frequency and omnidirectional audio signal component. S
Is taken out.

The main input (S + n) and the phase shift circuit 2
The output of the microphone 2 that is phase-shifted by 1 [-(S +
(N *))] is added by the adder 22 so that the signal component output from the adder 22 can have a directional characteristic, in this case, a unidirectional characteristic. From the output of the adder 22 having this unidirectional characteristic, a high-pass filter 23 extracts a high-frequency signal component.

In the adder 12, the low frequency audio signal component S and the high frequency signal component are added. As a result, an audio signal output having an omnidirectional characteristic in the low frequency range and a unidirectional characteristic in the high frequency range is formed.

Therefore, the normal pair of microphones 1, 2
By using, it is possible to cancel the noise component. Regarding the removal of this noise component, the windshield is not used, and since the microphones 1 and 2 are arranged close to each other, it is possible to contribute to the miniaturization of the device, and it is necessary to use an electrical / acoustic high-pass filter or the like. It is possible to prevent the sound quality from being deteriorated.

Further, the vibration noise component can be removed by using the normal pair of microphones 1 and 2. The circuit block having the configuration shown in FIG. 1 has two channels, that is, L and R channel circuit blocks 26 and 2.
7. By providing, omnidirectional microphone 1,
While using 2, it is possible to obtain a stereophonic sound output.

Since the adaptive noise canceller 6 is used, the characteristic of the adaptive filter 9 is automatically updated even if the characteristic of wind noise and / or vibration noise (for example, level or spectral distribution) changes. The wind noise component can be stably reduced.

Further, good sound collection quality can be realized with a single processing system without preparing a processing system for each type of noise.

According to the second modification of this embodiment, the output from the microphone 1 through the A / D conversion circuit 3 is supplied to the adders 41 and 42, and the microphone 2 performs A / D conversion. The output formed via the circuit 4 is supplied to the adder 41, and the signal formed by adding a negative sign to the output of the A / D conversion circuit 4 is supplied to the adder 43. Therefore,
In addition to the effect similar to that of the above-described embodiment, the sound signal component S
Can be doubled as compared with the above-mentioned embodiment.

The noise reduction device shown in this embodiment can be applied to a sound collecting system in various directions. For example, the present invention can be applied to a small portable video camera device or a single microphone. The pair of microphones 1 and 2 shown in this embodiment can be used with or without directivity.

[0074]

According to the invention of claim 1, there is an effect that a noise component can be canceled. As for the removal of noise components, a windshield is not used and a pair of microphones are placed close to each other, which contributes to downsizing of the device, and it is not necessary to use an electrical / acoustic high-pass filter or the like. There is an effect that it is possible to prevent deterioration of the sound collection quality. Further, there is an effect that good sound collection quality can be realized by a single processing system without preparing a processing system for each type of noise.

Further, according to the embodiment, since the adaptive noise canceller is used, the characteristic of noise [wind noise component + vibration noise component] [eg level or spectrum distribution]
Even if is changed, the characteristics of the adaptive filter are automatically updated, and the wind noise component can be stably reduced.

According to the invention of claim 2, in addition to the effect of claim 1, there is an effect that it is possible to obtain a stereoscopic sound output while using a pair of omnidirectional microphones.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an adaptive filter.

FIG. 3 is a diagram showing an example of arrangement of a pair of microphones.

FIG. 4 is a circuit diagram showing an example of an analog phase shift circuit.

FIG. 5 is a block diagram showing a configuration for obtaining a stereo output.

FIG. 6 is a block diagram showing a configuration in which outputs of a pair of microphones are used in L and R channel circuit blocks when a stereo output is obtained.

FIG. 7 is a block diagram showing a first modification example of the embodiment.

FIG. 8 is a block diagram showing a second modified example of the embodiment.

[Explanation of symbols]

One and two microphones 1a, 2a Sound receiving surface 5, 12, 22, 36, 37, 41, 42 Adder 6 Adaptive noise canceller 11 Low Pass Filter 20 separators 21 Phase shift circuit 23 High-pass filter 26 L channel circuit block 27 R channel circuit block d distance

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G10K 11/178

Claims (4)

(57) [Claims] 1. A pair of microphones arranged in close proximity to each other, a first subtraction unit for subtracting outputs of the pair of microphones, and an output of one of the outputs of the pair of microphones. In addition to the main input, the first
The output from the subtraction means is used as a reference input, and the phase of the output of one of the microphones of the pair of microphones and the adaptive processing means that performs adaptive processing based on the main input and the reference input is phase-shifted. Phase shifting means, a second subtracting means for subtracting the output of the other microphone from the output of the pair of microphones and the output of the phase shifting means, and the bass component of the output of the adaptive processing means is passed. Low-pass filter means, high-pass filter means for passing the high-frequency component of the output of the second subtraction means, output of the low-pass filter means and output of the high-pass filter means are combined. A noise reduction device comprising: 2. The noise reduction device according to claim 1, wherein the main input is supplied by adding outputs of the pair of microphones. 3. A pair of microphones arranged close to each other, a pair of low-pass filter means for passing a low-frequency component of the output of the pair of microphones, and a high-frequency component of the output of the pair of microphones. A pair of high-pass filter means to pass, and a first to subtract the output of the pair of low-pass filter means
And the output of the low-pass filter means of one of the outputs of the pair of low-pass filter means as a main input, and the output from the first subtraction means as a reference input, Adaptive processing means for performing adaptive processing based on the main input and reference input, and a phase shift for shifting the phase of the output of one of the outputs of the pair of high-pass filter means. Means, second subtraction means for subtracting the output of the other high-pass filter means of the pair of high-pass filter means and the output of the phase shift means, the output of the adaptive processing means, and the output of the adaptive processing means. A noise reduction device comprising means for combining the output of the second subtraction means. 4. The noise reduction device according to claim 3, wherein the main input is supplied by adding outputs of the pair of low-pass filter means.