JPH05313674A - Noise reducing device - Google Patents

Abstract

PURPOSE:To obtain sufficient noise reduction effect with small-sized constitution. CONSTITUTION:A desired voice with which a noise is mixed is collected by a 1st sound collecting means 11 to obtain a 1st voice signal. A voice which has large correlation with the noise is collected by a 2nd sound gathering means 21 to obtain a 2nd voice signal. A digital adaptive filter circuit 30 approximates the 2nd voice signal to the noise in the 1st voice signal. The signal of the approximation result is subtracted from the 1st voice signal and the digital adaptive filter is adjusted so that the residue of the subtraction result becomes minimum, thereby reducing the noise in the 1st voice signal. Further, the noise in the 1st voice signal is reduced in the state of the analog signal with the 2nd voice signal which is not passed through the adaptive filter circuit 30.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise reduction device using a digital adaptive filter.

[0002]

2. Description of the Related Art Conventionally, a circuit as shown in FIG. 9 has been known as an adaptive noise reduction circuit. In FIG. 9, 1
Is a main input terminal, 2 is a reference input terminal, and the signal input through the main input terminal 1 is supplied to the subtraction circuit 3. Further, the signal input through the reference input terminal 2 is supplied to the subtraction circuit 3 through the adaptive filter circuit 4. Then, the residual output of the subtraction circuit 3 is fed back to the adaptive filter circuit 4 and led to the output terminal 5.

In this noise reduction circuit, the main input terminal 1 receives the desired signal s and the noise n0 uncorrelated with the desired signal s. On the other hand, noise n1 is input to the reference input terminal 2. Noise n1 of this reference input
Is uncorrelated with the desired signal, but is correlated with noise n0.

The adaptive filter circuit 4 has a reference input noise n1.
Is filtered to output a signal y that approximates noise n0. The adaptive algorithm in the adaptive filter circuit 4 is the subtraction output (residual output) e which is the output of the subtraction circuit 3.
Works to minimize.

That is, assuming that s, n0, n1 and y are statistically stationary and the average value is 0, the residual output e is e = s + n0-y. The expected value of the square of this is that s is n0,
Also, because it is y ^{uncorrelated, E [e 2] = E} [s 2] + E [(n0 -y) 2] + 2E [s (n0 -y)] = E [s 2] + E [(n0 - y) ² ]. Assuming that the adaptive filter circuit 4 converges,
The adaptive filter circuit 4 is adjusted so that E [e ² ] is minimized. At this time, since E [s ² ] is not affected, Emin [e ² ] = E [s ² ] + Emin [(n 0 -y) ² ]. That, E E by [e ^2] is minimized [(n0 -y) ^2] is minimized, the output y of the adaptive filter circuit 4 becomes the estimated amount of noise n0. And
The expected value of the output from the subtraction circuit 3 is only the desired signal s. That is, adjusting the adaptive filter circuit 4 to minimize the total output power is equivalent to the subtraction output e becoming the least-squares estimated value of the desired speech signal s.

It is also possible to obtain a voice signal for canceling noise by the adaptive filter circuit 4, supply this signal to a speaker, an earphone, a headphone driver, etc., and acoustically add it to the main voice to reduce noise. it can.

[0007]

By the way, in the above-mentioned adaptive noise reduction circuit, the adaptive filter circuit 4 may be realized by an analog signal processing circuit or a digital signal processing circuit. However, when using an analog filter circuit, if the frequency of the noise signal is a single frequency, it is possible to achieve a sufficient noise reduction effect with a simple filter circuit. Since we have to filter out wide noise,
It is difficult to construct a filter circuit that obtains a sufficient noise reduction effect, and the noise reduction amount decreases when the surrounding environment such as the noise source changes.

Therefore, a digital filter is generally used as an adaptive filter, but if an attempt is made to obtain a sufficient reduction effect, the scale of the digital filter becomes large and
The device becomes expensive.

Since a digital filter for anti-aliasing is used as the digital filter, a delay time corresponding to several tens of taps is usually generated. Subtraction circuit 3
Adaptive filter circuit 4 from the main audio signal by signal processing
When subtracting the noise from the main audio signal, the main audio signal may be delayed by the delay in the adaptive filter circuit. However, in the device for acoustically reducing noise as described above, the main audio signal should be delayed. Therefore, the playback time of the noise canceling voice is delayed by the delay time in the adaptive filter circuit. Therefore, a sufficient noise reduction effect cannot be obtained.

In view of the above points, an object of the present invention is to provide a noise reduction device which is small in size and which can obtain a sufficient noise reduction amount.

[0011]

In order to solve the above-mentioned problems, the noise reduction device according to the present invention picks up a desired voice mixed with noise when the reference symbols of the embodiments described later are made to correspond, First sound collecting means 1 for converting into an audio signal
1, second sound collecting means 21 for collecting a voice having a large correlation with the noise and converting it into a second voice signal, and noise in the first voice signal for the second voice signal. A digital adaptive filter 30 for approximating to, and subtracting the signal of the approximation result from the first audio signal, and digital subtraction filter 3 so that the residual of the subtraction result is minimized.
0 for adjusting the first voice signal by digital processing means for reducing the noise in the first voice signal, and the second voice signal in the state of an analog signal not passing through the adaptive filter. And analog processing means 13 and 23 for reducing noise in the signal.

[0012]

In the present invention constructed as described above, in addition to the noise reduction by the digital signal processing by the digital adaptive filter 30, the noise reduction is performed in the state of the analog signal.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Some embodiments of the noise reducing apparatus according to the present invention will be described below with reference to the drawings. FIG. 1 shows an embodiment of a noise reduction device according to the present invention, in which reference numeral 11 denotes a main input microphone for picking up a desired voice s. The sound picked up by the main input microphone 11 is
It includes noise n0 which is not correlated with the desired voice s or is small. Reference numeral 21 is a reference input microphone for picking up noise n1 which has a correlation with the noise n0 but has little or no correlation with the desired voice s. For this purpose
The reference input microphone 21 is installed, for example, in the vicinity of a noise source, or is sound-insulated so as not to pick up a desired voice.

The voice signal picked up by the main input microphone 11 and converted into an electric signal is supplied to the A / D converter 14 via the amplifier 12 and the analog signal synthesizing circuit 13 to be converted into a digital signal. It is converted and supplied to the subtraction circuit 15.

The signal picked up by the reference input microphone 21 and converted into an electric signal is supplied to the A / D converter 24 via the amplifier 22, converted into a digital signal, and then subjected to an adaptive filter. It is supplied to the circuit 30. Then, the output signal of the adaptive filter circuit 30 is supplied to the subtraction circuit 15. The output signal of the subtraction circuit 15 is fed back to the adaptive filter circuit 30 and D / A
It is converted back into an analog signal by the converter 16 and led out to the output terminal 17.

In the adaptive filter circuit 30, as described below, the noise n0 contained in the main input speech is controlled so that the noise n1 as the reference input is approximated. Input microphone 11
Noise is subtracted and removed from the voice signal from, and the desired noise-free voice is obtained from the subtraction circuit 15. That is, noise is removed and reduced from the main voice by the adaptive digital signal processing described above.

An embodiment of an adaptive filter circuit 30 using a digital filter is shown in FIG. 2, in which a so-called LMS (Least Mean Square) method is used as an adaptive algorithm.

As shown in FIG. 2, an FIR filter type adaptive linear combiner 300 is used in this example. this is,
A plurality of delay circuits DL1, DL2, ... DLm (m is a positive integer) each having a delay time Z ⁻¹ of a unit sampling time, input noise n1 and each delay circuit DL1, DL
2, ... DLm is provided with weighting circuits MX0, MX1, MX2, ... MXm for multiplying the output signal of DLm by a weighting coefficient, and an adder circuit 310 for adding the outputs of the weighting circuits MX0 to MXm. The output of the adder circuit 310 is y.

The weighting coefficients supplied to the weighting circuits MX0 to MXm are formed on the basis of the residual signal e from the subtraction circuit 15 in the LMS arithmetic circuit 320 composed of, for example, a microcomputer. The algorithm executed by the LMS arithmetic circuit 320 is as follows.

Now, the input vector X _k at time k is
As shown in FIG. _{2, X k = [x 0k} x 1k x 2k ··· x mk] is ^T, the output y _k, the weighting coefficient _{w jk (j = 0,1,2, ...} m) and Then, the relationship of input and output is as shown in the following Equation 1.

[0021]

[Equation 1] Becomes

Then, the weight vector W _{k at} time _k
Is defined as W _k = [w _0k w _1k w _2k ... w _mk ] ^T , the input / output relationship is given by y _k = X _k ^T · W _k . If the desired response is d _k , the error e _{k from the} output is expressed as follows. e _k = d _k −y _k = d _k −X _k ^T · W _{k In the} LMS method, the weight vector is updated by the formula W _{k + 1} = W _k +2 μ · e _k · X _k . Here, μ is a gain factor (step gain) that determines the speed and stability of adaptation.

In this way, at the output terminal 17, a voice signal mainly consisting of the desired voice from which the noise n0 is removed can be obtained. However, in the configuration including only the adaptive digital filter circuit, a sufficient noise reduction effect cannot be expected unless the scale of the digital filter is increased. Therefore, in the present invention, an analog noise reduction circuit is added in addition to the adaptive digital signal processing circuit.

That is, the noise n1 through the amplifier 22
Is supplied to the gain and phase controller 23, adjusted so that the noise n0 in the signal from the amplifier 12 has the same gain and the phase is inverted, and then supplied to the synthesizing circuit 13. Therefore, the synthesizing circuit 13 works so that the noise n0 is removed from the main voice from the amplifier 12 by analog signal processing. Amplifiers 12 and 22
Since there is no delay between the output signals of, noise is effectively reduced.

In this case, the gain and phase controller 2
3 is adjusted so that the noise component in the output audio signal obtained at the output terminal 17 or the output audio signal of the synthesizing circuit 13 is most reduced. If the noise source is fixed, this adjustment can be performed in advance and fixed in the adjusted state. Of course, the so-called semi-fixed state may be set so that the adjustment can be performed again depending on the situation.

As described above, according to the present invention, in addition to the noise reduction by the adaptive digital processing, the noise reduction processing by the analog signal processing is performed, so that the scale of the digital filter is small and the noise reduction effect is obtained. Even when this is not enough, the analog signal processing can further reduce the noise, and a sufficient noise reduction effect can be obtained.

The gain and phase controller 23
Is the noise n0 in the main input when the gain and phase of the noise n1 are
The output of the gain and phase controller 23 may be subtracted from the main input audio signal in the synthesizing circuit 13 by adjusting so that

In the configuration of FIG. 1, the audio signal may be output as a digital signal without passing through the D / A converter 16. In addition, the adaptive filter circuit 30
A delay circuit is provided between the A / D converter 14 and the subtraction circuit 15 in order to compensate for the time delay required for the calculation for the adaptive processing in the above, the propagation time in the adaptive filter circuit and other time delays. Good.

In the example of FIG. 1, the analog signal processing is performed on the input side of the noise reducing device of the present invention, but the analog signal processing can also be performed on the output side of the noise reducing device. FIG. 3 shows the first embodiment in that case, and the main audio signal through the amplifier 12 is directly supplied to the A / D converter 14. Then, a combining circuit 26 is provided between the output terminal of the D / A converter 16 and the output terminal 17. Further, the noise n1 that has passed through the amplifier 22 is supplied to the gain and phase controller 25, and is adjusted so that the noise in the audio signal from the D / A converter 16 is equal to the gain and the phase is inverted. , To the synthesis circuit 26.

Therefore, the synthesizing circuit 26 works so as to remove noise from the main voice by analog signal processing. A / D converter 14 and D / A converter 16
Since the delay at 1 is smaller than the delay at the digital filter, if this is ignored, both analog signals synthesized by the synthesis circuit 26 have almost no delay, so that noise reduction is effectively performed.

In the case of this example, as the residual e input to the digital adaptive filter circuit 30, the output of the synthesizing circuit 26 is converted into a digital signal by the A / D converter 27 and is supplied. The weighting coefficient w _jk of the digital adaptive processing is updated according to e.

FIG. 4 shows a second embodiment of an example in which analog signal processing is performed on the output side of the noise reduction device. The difference from the example of FIG. Which signal to use? That is, in the case of this example, the output of the subtraction circuit 15 is input to the adaptive filter circuit 30 as the residual e, and the weighting coefficient w _jk of the digital adaptive processing is updated according to this residual e.

3 and 4, the gain and phase controller 25 is adjusted so that the gain and phase of the noise n1 match the noise in the output of the D / A converter 16, and the gain and phase are combined. The circuit 26 may subtract the output of the gain and phase controller 23 from the main input audio signal.

5 to 7 show data obtained by confirming the noise reduction effect of the configuration example of FIG. 3 by experiments. Figure 5
The frequency spectrum distribution of the output signal of the main input microphone 11 is shown, and noise is included. Figure 6
It shows the spectral distribution of the output signal of the device when analog signal processing is not performed, that is, when noise reduction is performed only by digital signal processing only by the adaptive filter circuit 30, and most noise is removed and noise is reduced. However, some noise remains. FIG. 7 shows FIG.
The spectrum distribution of the signal obtained at the output terminal 17 of the noise reduction device having the above configuration is shown, and it can be seen that a sufficient noise reduction effect is obtained.

Next, FIG. 8 shows an embodiment of a noise reduction device according to the present invention having a headphone structure. The headphones of this example are for both left and right ears, but FIG. 8 shows the headphones for one ear. The headphone portion for the other ear has the same structure.

In the case of this example, the desired voice is silent, and the main input voice does not exist. Reference numeral 41 is an acoustic-electrical converter, for example, a microphone for picking up external noise, whose output audio signal is A / D via an amplifier 42.
It is supplied to the converter 43 and converted into a digital signal,
It is supplied to the adaptive filter circuit 30. The output signal of the adaptive filter circuit 30 is converted into an analog signal by the D / A converter 44. Then, this analog signal is supplied to the headphone driver 46 via the synthesis circuit 45, and the headphone unit 47 as an electro-acoustic converter is supplied.
Is driven, and a noise canceling noise is emitted from the headphone unit 47 to the space 52 formed between the headphones when the headphones are attached to the ears. As a result, noise is reduced so that the space 52 becomes silent.

Then, the noise residual in this space 52 is picked up by an acoustic-electrical converter such as the microphone 48. The output signal of the microphone 48 is supplied to the A / D converter 50 via the amplifier 49 and converted into a digital signal, which is supplied to the adaptive filter circuit 30.
The adaptive filter circuit 30 performs the adaptive operation so that the noise residual becomes zero as described above.

In this example, analog signal processing is added in addition to the above digital adaptive processing. That is, the output signal of the amplifier 42 is supplied to the gain and phase controller 51, the gain and the phase are adjusted, and then supplied to the synthesis circuit 45, added to the noise canceling sound, and emitted from the headphone unit 47. .. The gain and phase controller 51 is adjusted so that the noise residual becomes zero.

[0039]

As described above, according to the present invention, noise reduction is performed by analog signal processing in addition to noise reduction by digital adaptive processing. Therefore, the scale of the digital filter is small and the noise reduction effect is reduced. Even if this is not enough, the noise reduction is further performed by the analog signal processing, and a sufficient noise reduction effect can be expected. Therefore, the scale of the digital filter may be small.

Even when the noise reduction effect of the adaptive digital signal processing does not appear, the noise reduction effect of the analog signal processing can be expected. Also, even in a system that acoustically reduces noise, there is no delay effect in analog signal processing, so a sufficient noise reduction effect can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of a noise reduction device according to the present invention.

FIG. 2 is a diagram showing an example of an adaptive filter circuit.

FIG. 3 is a block diagram of another embodiment of the noise reduction device according to the present invention.

FIG. 4 is a block diagram of still another embodiment of the noise reduction device according to the present invention.

5 is a diagram showing a spectral distribution of an output signal of a main input microphone in the example of FIG.

FIG. 6 is a diagram showing a noise-reduced output signal only by digital adaptive processing.

7 is a diagram showing a spectral distribution of an output signal obtained at the output terminal in the example of FIG.

FIG. 8 is a block diagram of another embodiment of the present invention.

FIG. 9 is a diagram for explaining an example of a conventional noise reduction device.

[Description of Reference Signs] 11,21 First and second microphones 13,26,45 Synthesizing circuit 14,24,27 A / D converter 15 Subtracting circuit 16 D / A converter 23,25,51 Gain and phase controller 30 Adaptation Filter circuit 41 Noise pickup microphone 47 Headphone unit 48 Residual pickup microphone

Claims (4)

[Claims] 1. A desired voice containing noise is picked up, and a first voice is collected.
First sound collecting means for converting into a sound signal, second sound collecting means for collecting a sound having a large correlation with the noise and converting into a second sound signal, and the second sound collecting means. A digital adaptive filter for approximating a voice signal to noise in the first voice signal is provided, and the signal of the approximation result is subtracted from the first voice signal so that the residual of the subtraction result is minimized. A digital processing system for adjusting the digital adaptive filter to reduce the noise in the first audio signal; and a second audio signal that has not passed through the adaptive filter in an analog signal state. 1. A noise reduction device comprising an analog processing system for reducing noise in the audio signal of 1. 2. A desired voice containing noise is picked up, and the first voice is collected.
First sound collecting means for converting into a sound signal, second sound collecting means for collecting sound having a high correlation with the noise and converting into a second sound signal, and state of analog signal Then, the second audio signal is transmitted to the first audio signal.
Calculating means for subtracting the audio signal from the first arithmetic means, a first A / D converter converting the audio signal from the first calculating means into a digital signal, and the second audio signal into a digital signal. Second A to convert
/ D converter, and the audio signal from the second A / D converter to the first
A digital adaptive filter for approximating the noise in the voice signal, and reducing the noise in the first voice signal by subtracting the approximated noise signal from the output voice signal of the first A / D converter. A noise reducing device comprising: two calculating means, and adjusting the digital adaptive filter so that the residual of the noise is minimized. 3. A desired voice mixed with noise is picked up, and the first voice is collected.
First sound collecting means for converting into a sound signal, second sound collecting means for collecting sound having a large correlation with the noise, and converting into a second sound signal, the first sound collecting means First A for converting audio signal to digital signal
/ D converter, and a second A for converting the second audio signal to a digital signal
/ D converter, and the audio signal from the second A / D converter to the first
A digital adaptive filter for approximating the noise in the voice signal, and reducing the noise in the first voice signal by subtracting the approximated noise signal from the output voice signal of the first A / D converter. A first arithmetic means, a D / A converter for converting an output signal of the first arithmetic means into an analog signal, and a second audio signal subtracting the second audio signal from the output signal of the D / A converter. A noise reduction device comprising a computing means and adjusting the digital adaptive filter so that the residual of the noise is minimized. 4. A first method for collecting noise and converting it into an audio signal.
Sound collecting means, an A / D converter for converting the first audio signal into a digital signal, a digital adaptive filter for approximating the audio signal from the A / D converter to the noise, and the digital adaptive filter D / A converter that converts the output of the analog output into an analog signal, an analog processing circuit that controls the gain and phase of the audio signal from the first sound collecting means, an output signal of this analog processing circuit, and the D / A A synthesizing means for synthesizing the output signal of the converter, an electro-acoustic converting means for reproducing a voice that cancels noise by being supplied with the output of the synthesizing means, and a noise residual in a reproduced acoustic space of the electro-acoustic converting means. A second sound collecting means for collecting the sound, and the digital adaptive filter is adjusted so that a noise residual collected by the second sound collecting means is minimized. Noise reduction apparatus comprising.