JP2000293806A - Automatic mechanical noise reduction device - Google Patents

Abstract

(57) [Problem] To automatically turn on / off a mechanical noise reduction circuit in a mechanical noise reduction device for extracting only an audio signal from an audio signal mixed with mechanical noise. SOLUTION: An adaptive filter for generating a pseudo noise signal, autocorrelation coefficient operation processing means for calculating an autocorrelation coefficient of the pseudo noise signal, and ON / OFF by an output of the autocorrelation coefficient operation processing means And a mechanical noise reduction circuit for which control is automatically performed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a camera-integrated VT.
The present invention relates to a device for reducing mechanical noise in a device having a built-in microphone such as R.

[0002]

2. Description of the Related Art In a camera-integrated VTR such as a digital camcorder (product name), mechanical noise (mechanical noise) caused by contact between a rotating magnetic head and a magnetic tape,
A mechanical noise reduction device that reduces only mechanical noise when a so-called head tapping sound and / or a rotating drum rotating sound is mixed into an audio signal collected from a microphone has already been proposed.

In general, this device turns on a reduction processing circuit only when mechanical noise is generated, and when the mechanical noise disappears and only an audio signal is generated, the reduction processing circuit does not affect the output audio signal. OFF
It is supposed to. At this time, the source of noise is monitored to switch ON / OFF the reduction processing circuit. For example, when the noise source is a rotating drum of a VTR, the presence / absence of rotation of the drum is monitored by a microcomputer or the like, and if it is rotating, the mechanical noise reduction processing circuit is turned on, and if not, it is turned off. Switching control is performed.

However, the cause of the mechanical noise mixing is not limited to the case described above, but may be various other cases. For example, when a signal captured and picked up by a single video camera is recorded on the spot on another VTR such as a stationary VTR, mechanical noise generated by a recording-side rotating drum, which is a noise source, is generated by the microphone of the single video camera. The microcomputer cannot monitor the rotation of the drum in this case. Furthermore, there has been proposed a reproduction mechanical noise reduction device that removes only mechanical noise from an audio signal obtained by reproducing a tape on which mechanical noise is recorded in tape reproduction.
In the tape reproduction when V images and the like are mixed on the tape, ON / OFF control of mechanical noise reduction could not be performed because the generation of mechanical noise was not monitored by the microcomputer.

[0005] Regarding the above-mentioned conventional mechanical noise reduction device,
This will be described in more detail below with reference to FIGS.
FIG. 5 is a system block diagram showing a mechanical noise reduction device at the time of recording. In the system shown in the figure, audio signals mixed with mechanical noise are input to the microphones MIC1A and MIC1B, and a left signal 1L and a right signal 1R thereof are amplified by amplifiers AMP2A and AMP2B, respectively.
It is input to the AGC circuit 3 as 2R. These audio signals 2L and 2R are gain-adjusted by an AGC circuit 3 so as to have an optimal signal level in a subsequent stage, and AD signals are output as signals 3L and 3R.
The signals are sent to converters ADC4A and ADC4B, where they are converted from analog to digital, and sent to mechanical noise reduction processing circuit 15 as signals 4L and 4R.

[0006] A drum reference signal 7 is supplied to the mechanical noise reduction processing circuit 15 from a microcomputer computer (hereinafter referred to as a microcomputer) 8. The microcomputer 8 further sends a servo signal 12 to the drum motor 13 to control the rotation frequency and phase of the drum motor 13. At this time, the drum reference signal 7 is also a reference signal of the servo signal 12.

A rotary drum 14 is mounted on the drum motor 13, and rotary magnetic heads 11 A and 11 B are mounted on the rotary drum 14. At the time of recording / reproduction, magnetic recording / reproduction is performed while the rotating magnetic heads 11A and 11B are in contact with the magnetic tape 9. Most of the mechanical noise generated here is generated at an integral multiple of the frequency including the frequency of the drum reference signal 7, and the mechanical noise is transmitted to the space or the cabinet of the device, and is transmitted to the microphones MIC1A and MIC1B together with the audio signal. Is input to

Here, the mechanical noise reduction processing circuit 15 will be described in more detail. The signals 4L and 4R digitized by the ADC circuits 4A and 4B are added to adders 6A and 6B, respectively.
And the pseudo noise signals 5L and 5R output from the adaptive filters 5A and 5B processed by the LMS algorithm are applied to the other inputs of the adders 6A and 6B, respectively, with negative polarity. Create the difference signals 6L and 6R,
The signals are sent to the recording system signal processing circuit as left channel Lch and right channel Rch outputs. Since the recording system signal processing circuit has no direct relation to the present invention, the description is omitted here. The difference signals 6L and 6R are output from the adaptive filter 5 respectively.
A and 5B are fed back as error signals. The adaptive filter 5A,
5B always operates to minimize this error signal.

Here, the mechanical noise reduction processing circuit 15 will be described in more detail with reference to FIG. Since the left channel Lch and the right channel Rch perform the same processing, only one channel is shown in FIG. In this circuit, an input signal 31 in which an audio signal and noise are mixed is input to an adder 30, and the adder 30 subtracts the pseudo noise signal 37 sent from the LMS processing circuit 15 to convert only the audio signal into a signal 38. Output as

A part of the signal 38 is fed back to the LMS processing circuit 15, and the mechanical noise component remaining in the signal 38 is processed as follows. That is, the signal 38 is sent to the limiter 26, where the large level portion of the audio signal is limited, and most of the mechanical noise components are set to the optimum level by the gain control circuit 27. Thereafter, the signal is sent to an adder 29 through a decimation circuit 28 to reduce the sampling rate to 1/2. RAM 21 in adder 29
Is added to the read data 39 read out of the RAM 21 and the sum is written to the RAM 21 as write data 40.

The address value of the RAM 21 is generated by a read / write address counter 22. In the circuit of FIG. 6, the read address 35 is directly stored in the RAM.
The write address 36 is applied to the RAM 21 via the delay circuit 23. The delay circuit 23 is for optimizing the time delay due to the sampling rate conversion processing. Then, the address of the read / write address counter 22 is reset by applying the impulse signal 33. This impulse signal 33 corresponds to the drum reference signal 7 in FIG. 5, and has a pitch period component of the generated noise. As a clock supplied to the read / write address counter 22, a clock obtained by subjecting a sampling clock 34 having an audio sampling frequency (32 kHz, 44.1 kHz, or 48 kHz) to a rate conversion by a frequency down-converter to 1/2 is used.

Next, the read data 39 is subjected to an interpolation process for rate conversion by an interpolation circuit 56, and passes through a switch SW25 to generate a pseudo noise signal 37.
Becomes Here, the switch SW25 is ON / OFF controlled by a cancel control signal 32 generated from a microcomputer or the like, that is, the switch SW25 is turned on when the rotating drum 14 in FIG. 5 is rotating, and is turned off when stopped. Is controlled.

The ON / OFF control of the switch SW25 has the following disadvantages. That is, for example, when trying to reduce only the mechanical noise sound from the reproduced signal at the time of tape reproduction, the microcomputer cannot know whether the mechanical noise sound is recorded on the tape at the time of recording. This is because the audio recording made on the tape is recorded in the microphone inside the camera, recorded outside the camera, recorded on TV (so-called air check), recorded without sound (only image),
The condition such as whether the level of the recorded mechanical noise sound is large or negligible (reduction processing is unnecessary) depends on the device that recorded them, and the tape playback It is because it is not judged for the first time.

[0014]

SUMMARY OF THE INVENTION One object of the present invention is to overcome the above-mentioned drawbacks of the conventional mechanical noise reduction device, detect a mechanical noise component from an input audio signal, and automatically reduce the mechanical noise according to the level of the component. Circuit ON /
An object of the present invention is to provide a mechanical noise reduction device that can be turned off. Another object of the present invention is to turn on a mechanical noise reduction circuit.
It is to improve the unnaturalness at the time of ON / OFF switching in the / OFF control. Still another object of the present invention is to improve the reduction efficiency of the mechanical noise reduction device.

[0015]

According to the present invention, there is provided a mechanical noise reducing apparatus provided with the following means for solving the above-mentioned problems. That is, in one aspect of the present invention, a mechanical noise reduction device is a mechanical noise reduction device for extracting only an audio signal from an audio signal mixed with mechanical noise, and an adaptive filter that generates a pseudo noise signal; A mechanical noise reduction device comprising: an autocorrelation coefficient operation processing means for calculating an autocorrelation coefficient of a signal; and a mechanical noise reduction circuit for automatically performing ON / OFF control by an output of the autocorrelation coefficient operation processing means. I will provide a.

According to a second aspect of the present invention, in the mechanical noise reduction device, the total number of samples when calculating the correlation coefficient by the autocorrelation coefficient calculation processing means is determined by the repetition pitch period of the generated mechanical noise signal. The present invention provides a mechanical noise reduction device that is adapted to:

According to a third aspect of the present invention, in the mechanical noise reduction device, the mechanical noise reduction circuit is ON / OFF controlled by cross-fade processing means to switch and output an original input signal and a signal with reduced mechanical noise. A mechanical noise reduction device is provided.

According to a fourth aspect of the present invention, the adaptive filter includes a noise reduction device that performs an LMS operation process in which a reduction band is extended to a Nyquist frequency without performing decimation and interpolation processes. provide. According to a fifth aspect of the present invention, there is provided the mechanical noise reduction device, wherein the LMS operation is performed by multiplying the signal read from the RAM by an attenuation coefficient.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A mechanical noise reduction device according to an embodiment of the present invention will be described below with reference to the accompanying drawings.
FIG. 1 is a system block diagram showing the entire mechanical noise reduction circuit during reproduction according to the present invention. In the apparatus shown in the figure, a reproducing device 80 such as a VTR
1 reproduces the RF signal recorded on the magnetic tape 9 and converts it into a reproduction system video / audio signal 91. At this time, the rotating drum 14 on which the rotating magnetic head 11 is mounted is driven by a drum motor 13 The motor 13 drives the magnetic tape 9 by the servo signal 12 from the microcomputer 8.
Is controlled to the same rotation frequency and phase as when the signal was recorded in the.

A reproduction video / audio signal 91 is input to a reproduction signal processing circuit 81, which performs deframing of video and audio signals, decompression of video, deshuffling of audio, and the like, and performs digital audio signals 81L and 81R. As
It is input to the mechanical noise reduction processing circuit 88. Further, a drum reference signal 7 which is a synchronization reference for the drum motor 13 is input from the microcomputer 8 to the mechanical noise reduction processing circuit 88. The signals 81L and 81R are input to the cross-fade signal switching processing circuits 84 and 85, respectively, and the adder 86 and 87 subtract the pseudo noise signals 92 and 93 to produce signals 94 and 95, respectively. And 85.

The signals 94 and 95 are inputted as error signals to adaptive filters 82 and 83 constituted by LMS processing circuits, and the adaptive filters 82 and 83 use the above-mentioned drum reference signal 7 as a reference input to generate a pseudo noise signal 92.
And 93 are generated and sent to the adders 86 and 87.
Correlation coefficient signals 96 and 97 are sent from the adaptive filters 82 and 83 to cross-fade signal switching processing circuits 84 and 85, and the cross-fade signal switching processing circuits 84 and 85 use these signals 96 and 97, The input two signals 81L and 81R and the noise-reduced signals 94 and 95 of these signals are switched to output the left channel Lch, respectively.
It is output from the mechanical noise reduction processing circuit 88 as the right channel Rch output. The circuit following this stage has no direct relation to the present invention, and a description thereof will be omitted. Further, the video signal processing is omitted.

Next, the mechanical noise reduction processing circuit 88 in the apparatus shown in FIG. 1 will be described in detail with reference to FIG. Here, FIG. 2 shows only one channel of the stereo signal. In the circuit shown in the figure, a signal in which noise is mixed in a voice or only a voice signal is input to a signal input 50, and this signal is directly input to the cross-fade signal switching processing circuit 43 on the one hand. On the other hand, the input signal is applied to one input of an adder 45 and the signal 51 obtained by subtracting the pseudo noise signal 59 applied to the other input is input to the cross-fade switching processing circuit 43. The two signals input to the cross-fade switching processing circuit 43 are switched by the correlation coefficient signal 58 sent from the auto-correlation coefficient operation processing circuit 46, and the audio signal is output to the output of the cross-fade switching processing circuit 43. It is selectively output as output 55.

For example, when the value of the correlation coefficient signal 58 is larger than a certain threshold value, the cross-fade signal switching processing circuit 43 outputs the signal 51 subjected to the mechanical noise reduction processing sent from the output of the adder 45. Is output, and when it is small, the switching operation is performed so as to output the input signal 50 as it is. This switching operation is performed so as to be gradually switched over a certain period of time. Further, as will be described later in detail, the correlation coefficient signal 58 may be used as it is as a control coefficient at the time of cross-fade and switched.

The signal 51 is a signal obtained by subtracting the pseudo noise signal 59 from the input signal 50. If the pseudo noise signal coincides with the actual noise signal, the signal 51 comprises only the audio signal component included in the input signal 50. It is. When the actual noise signal does not match the pseudo noise signal, the signal includes the error component. LMS processing circuit 4
0 performs a feedback control operation so as to minimize the error component of the noise.

The signal 51 fed back to the LMS processing circuit 40 is limited by a limiter 47 at a portion where the audio signal component of the signal is at a high level, set to an appropriate level by a gain control circuit 48, and added. Sent to one input of the device 49. The other input of the adder 49 includes the RAM 4
A signal obtained by multiplying the read data 56 read from 1 by the coefficient λ in the coefficient circuit 44 is input. The adder 49 adds the two signals, sends the sum to the RAM 41 as write data 57, writes it in the RAM 41, and sends it to the autocorrelation coefficient operation processing circuit 46.

The autocorrelation coefficient operation processing circuit 46 receives the read data 56 and the write data 57, calculates a correlation coefficient from the data, and outputs the correlation coefficient signal 5
8 is output. The read data 56 is sent to the adder 45 as a pseudo noise signal 59 as it is. RAM4
1 is controlled by the read / write address signal 54.
This is done using This read / write address signal 5
4 is generated by a read / write address counter 42 from an impulse signal 52 and a sampling clock 53. Here, the impulse signal 52 corresponds to the drum reference signal 7 in FIG. 1, is used for resetting the address of the read / write address counter 42, and uses the audio sampling frequency (32
kHz, 44.1 kHz, or 48 kHz).

An embodiment of the present invention has been described above. The mechanical noise reducing apparatus according to the present embodiment summarizes the autocorrelation coefficient of the pseudo noise signal generated by the adaptive filter. Is calculated, and the mechanical noise reduction circuit is turned on / off by using the calculated correlation coefficient.
The FF control is performed. As described above, the autocorrelation coefficient calculation processing circuit 46 of FIG.
The read data 56 and the write data 57 are input. In the present embodiment, the read address and the write address are the same, and the read data is data one noise pitch cycle before the write data, that is, the rotating drum. A time delay of one rotation has occurred. The autocorrelation coefficient is used to check the periodicity of a certain continuous signal waveform. In the case of the present embodiment, the noise component caused by the rotating drum has a periodicity corresponding to one rotation of the rotating drum. Is used to calculate the correlation coefficient.

The details of the circuit for calculating the autocorrelation coefficient will be described in more detail below with reference to FIG.
FIG. 14 is a circuit block diagram of the correlation coefficient calculation processing circuit 46. Here, first, in order to extract a noise band frequency of the read data 56 and the write data 57 as the input signals, the read data 56 and the write data 57 are passed through bandpass filters BPF61 and BPF62, respectively.
It is input to the correlation coefficient calculation circuit 63. Correlation coefficient calculation circuit 6
3 calculates the degree of correlation R between the read data and the write data according to the following formula.

The degree of correlation R generally takes a value in the range of -1 ≦ R ≦ 1, and the closer the value is to 1, the higher the correlation is. If the value is zero, the correlation is negative. It has the property of being high in nature. Therefore, the value approaches 1 as the phase correlation between the write data W and the read data R increases, and decreases as the phase correlation decreases.
Is calculated so as to approach. Next, the correlation R is amplified to an appropriate value by the amplifier AMP64, the sign is adjusted to a positive value by the level shift processing circuit 65, and output as the correlation coefficient signal 58 through the detection processing circuit 66 and the time constant adding circuit 67. I do. Here, the detection processing circuit 66 and the time constant adding circuit 67
Is a process for adjusting a coefficient change to human hearing. For example, an abrupt change in the degree of correlation R is provided with an attack / recovery time constant so as to change naturally.

Next, a second embodiment of the present invention will be described. In the present embodiment, the accuracy of the periodic noise extraction due to the rotation of the drum is further increased by making the total number N of calculation samples coincide with the repetition pitch period of the generated mechanical noise signal when calculating the correlation R. . Therefore, the number of samples N is obtained as follows. N = Audio sample frequency / Drum rotation frequency

As an example, if the audio sampling frequency is 48 kHz and the drum rotation frequency is 150 Hz, the total number of calculation samples N is 320, that is, the calculation of the correlation R may be performed every 320 samples.

Next, a third embodiment of the present invention will be described. The feature of this embodiment is that the mechanical noise reduction circuit
In performing the N / OFF control, the cross-fade signal switching processing circuit switches and outputs the signal subjected to the mechanical noise reduction processing and the signal not subjected to the mechanical noise reduction processing. In a conventional device of this type, for example, a pseudo noise signal to be subtracted from an input signal is switched on / off as in a switch SW25 shown in FIG. 6, but this method has the following problems. Was. 1. Since the loop of the LMS processing is broken at the time of the OFF, the pseudo noise signal component changes at the next ON, and the reduction effect is reduced until the adaptation is stabilized again. 2. If a pseudo noise signal component remains at the time of ON / OFF, noise is heard at the time of switching, and unnaturalness remains.

Therefore, the mechanical noise reduction device of the present embodiment solves the problem by gradually switching the signal for mechanical noise reduction and the completely original signal in the same phase without constantly cutting the LMS processing loop. ing. This will be described in more detail below with reference to FIG.

The circuit shown in FIG. 4 is a block diagram of a circuit corresponding to the cross-fade signal switching processing circuit 43 in FIG. The two input signals 50 and 51 shown in the figure are input to attenuators ATT71 and ATT72, respectively, where the respective signals are subjected to level control, and then sent to an adder 73 where they are added and the signal 55 is added. Is output as At this time, the input signals 50 and 5
1 basically indicates whether or not a mechanical noise signal component is included, the phases of the audio signals are aligned, and even if both are added, there is almost no sense of incongruity.

FIG. 4 shows input signals 50 and 51 input to attenuators ATT71 and ATT72, and these signals are converted into a coefficient inversion signal of a correlation coefficient 58 and a correlation coefficient signal sent from an autocorrelation coefficient calculation processing circuit. This is an example of a circuit for level control by 58 itself. Coefficient inversion circuit 70
When the correlation coefficient 58 is added to the coefficient processed in
The level adjustment processing of the input signals 50 and 51 is performed so as to be (ie, the whole). Therefore, the output signal 55 obtained by adding the two by the adder 73 is not affected by the addition even during the cross-fade, and is always kept at the same level as the input.

Next, a fourth embodiment of the present invention will be described. In the present embodiment, the signal processing in the decimation circuit 28 and the signal processing in the interpolation circuit 56 in the conventional LMS processing circuit 15 shown in FIG. The LMS calculation process is performed by extending the frequency up to the frequency (１ ／ of the audio sampling frequency).

In the conventional example, the processing band is set to one of the audio sampling frequencies in order to reduce the RAM capacity.
However, with the progress of the semiconductor miniaturization process, a large capacity, a low price, and a small area can be achieved, and the RAM capacity can be increased to increase the performance. .

Therefore, as an example, when the audio sampling frequency is 32 kHz, the noise that could only be reduced in the band up to 8 kHz in the past can be reduced to 16 kHz in the present embodiment, and almost covers the human audible band. As a result, the noise reduction effect is greatly improved. Further, the delay circuit DL2 provided in the write address supply path in the circuit of FIG.
3 can also be omitted, and the number of circuit components can be reduced.

Next, a fifth embodiment of the present invention will be described. The feature of this embodiment is that the LMS processing circuit 40 shown in FIG.
, The read data 5 read from the RAM 41
6 is multiplied by an attenuation coefficient λ in a coefficient circuit 44 to perform an LMS calculation process. This damping coefficient λ is set to take a value slightly smaller than 1, and generally, Leaky is smaller than that of the ordinary LMS method. It is called LMS method. Here Leaky The LMS method updates the data D in the RAM 41 of FIG. 2 by the following operation. D (n) = λ · D (n−1) + 2μ · E (n−1) · X
(N-1)

Here, n is a sample number, μ is an update step gain, which corresponds to the gain control circuit 48 in FIG. 2, E is a residual error signal, which corresponds to the audio signal 51 in FIG. 2, and X is a reference input signal. Corresponds to the impulse signal 52 in FIG.

The LMS algorithm includes, for example, audio sampling frequencies of 32 kHz, 44.1 kHz, and 4 kHz.
At each sampling of 8 kHz, the contents of the RAM 41 are updated in order to minimize the difference between the actual noise signal and the pseudo-noise signal that is subtracted by this equation. Here, the meaning of the attenuation coefficient λ is that when the residual error signal E suddenly becomes zero at the stage when the adaptation is sufficiently advanced, for example, when the audio signal becomes a non-audio recording portion in tape reproduction, or depending on the device. Considering the case where the sound is muted, for example, the second term on the right side disappears in the above arithmetic expression, so that D (n-1) remains forever and a sense of incongruity is generated. Since n) gradually decreases and approaches zero, the response in such a case is improved as compared with the conventional example.

[0042]

Advantages of the Invention In the reproduction mechanical noise reduction, even if the audio signal on which the mechanical noise is recorded and the audio signal on which the mechanical noise is not recorded are reproduced alternately, the mechanical noise reduction can be automatically turned ON only when there is mechanical noise, and the user is not aware of the mechanical noise, Can be prevented from occurring when mechanical noise reduction processing is performed during audio reproduction in which no is recorded. 2. Since the cross-fade processing is employed when the mechanical noise reduction processing is turned ON / OFF, the switching is natural and does not give a user a sense of discomfort. Further, the correlation coefficient can be directly used for the level control of the crossfade processing, and more natural switching is possible. 3. Also in the mechanical noise reduction processing at the time of recording, since the ON / OFF information of the mechanical noise reduction processing may not be provided from the microcomputer, the load on the microcomputer processing is reduced. 4. In the correlation coefficient calculation process, by setting the total calculation sample length to be processed to the mechanical noise generation cycle length, the extraction accuracy of the mechanical noise periodicity is increased, and reliable ON / OFF control can be performed. 5. Since the LMS processing band is doubled as compared with the conventional example, the effect of reducing mechanical noise is improved. 6. Leaky By adopting the LMS method, even if the input audio signal suddenly becomes zero, the responsiveness is improved because the pseudo noise signal becomes zero.

[Brief description of the drawings]

FIG. 1 is a circuit block diagram of an apparatus for reducing mechanical noise during reproduction according to an embodiment of the present invention.

FIG. 2 is a circuit block diagram showing in detail a mechanical noise reduction processing circuit portion in the circuit of FIG. 1;

FIG. 3 is a circuit block diagram showing in detail a part of a correlation coefficient operation processing circuit in the circuit of FIG. 2;

FIG. 4 is a circuit block diagram showing an example of a part of a cross-fade signal switching processing circuit of the circuit of FIG. 2;

FIG. 5 is a circuit block diagram of an example of a conventional mechanical noise reduction device.

FIG. 6 is a circuit block diagram of an example of a conventional mechanical noise reduction processing circuit.

[Explanation of symbols]

Reproduction apparatus such as 80 VTR, 81 reproduction signal processing circuit, 8 microcomputer, 88 mechanical noise reduction processing circuit, 82, 83 adaptive filter, 8
4,85 ° cross-fade signal switching processing circuit, 86,
87 ‥‥ adder

Claims (5)

[Claims] 1. A mechanical noise reduction device for extracting only an audio signal from an audio signal mixed with mechanical noise, comprising: an adaptive filter that generates a pseudo noise signal; and a self filter that calculates an autocorrelation coefficient of the pseudo noise signal. ON / OF by an output of the correlation coefficient operation processing means, and an output of the autocorrelation coefficient operation processing means
A mechanical noise reduction device comprising: a mechanical noise reduction circuit in which F control is automatically performed. 2. The mechanical noise reduction device according to claim 1, wherein the total number of samples when calculating the correlation coefficient by said autocorrelation coefficient calculation processing means matches the repetition pitch period of the generated mechanical noise signal. A mechanical noise reduction device. 3. The mechanical noise reduction device according to claim 1, wherein the mechanical noise reduction circuit is ON / OFF controlled by a cross-fade processing means to switch and output an original input signal and a signal with reduced mechanical noise. Mechanical noise reduction device. 4. The noise reduction device according to claim 1, wherein the adaptive filter performs an LMS operation process in which a reduction band is extended to a Nyquist frequency without performing decimation and interpolation processes. 5. The apparatus according to claim 4, wherein said L
A mechanical noise reduction device in which MS calculation processing is performed by multiplying a signal read from a RAM by an attenuation coefficient.