JP2014026002A - Sound recording device and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sound recording device that maintains pre noise-reduced localized perception of a sound or sense of presence thereof even after the noise is reduced.SOLUTION: Sound recording devices 1 and 10 according to the present invention comprise: a plurality of sound collecting parts 41; a first noise estimation part 52L that performs a first noise estimation of a first sound collected by a first sound collecting part 41L of sound collecting parts 41; a first noise reduction part 53L that reduces the noise estimated by the first noise estimation part 52L from the first sound; and a second adjustment part 54R that calculates a first relation R serving as a relation between a second sound collected by a second sound collecting part 41R of the plurality of sound collecting parts 41 and the pre noise-reduced first sound, and adjusts the second sound so as to make a relation between the second sound and the post noise-reduced first sound equal to the first relation R.

Description

  The present invention relates to a recording apparatus and a program.

  Some cameras and the like that can be recorded include a plurality of microphones and drive units for recording external sound. In such an apparatus, a technique for reducing noise generated from a drive unit from recorded voice has been proposed (see Patent Document 1).

JP 2011-259319 A

However, in the case of stereo audio, if noise reduction processing is performed independently for the left ear audio and the right ear audio, the ratio of the noise component remaining in the left ear audio to the noise component remaining in the right ear audio is It changes every moment, and the localization of the residual noise also changes every moment.
In addition, since the ratio between the target sound deterioration component of the left-ear sound and the target sound deterioration component of the right-ear sound changes every moment, the target sound's localization also changes every moment from the original localization.
For this reason, the processed sound is a sound in which the original localization and the sense of presence are impaired.

  An object of the present invention is to provide a recording apparatus and a program in which a sense of orientation and a sense of presence before noise reduction are maintained even after noise reduction.

  The present invention solves the above problems by the following means. In addition, in order to make an understanding easy, although the code | symbol corresponding to embodiment of this invention is attached | subjected and demonstrated, it is not limited to this.

According to the first aspect of the present invention, noise estimation of a first sound collected by a plurality of sound collection units (41) and a first sound collection unit (41L) of the plurality of sound collection units (41) is provided. A first noise estimator (52L) that performs noise, a first noise reducer (53L) that reduces noise estimated by the first noise estimator (52L) from the first sound, and the plurality of sound collections The first relationship (R), which is the relationship between the second sound collected by the second sound collecting unit (41R) of the units (41) and the first sound before noise reduction, is obtained, and after noise reduction And a second adjusting unit (54R) that adjusts the second sound so that the relationship with the first sound is the same as the first relationship (R). , 10).
The invention according to claim 2 is the recording apparatus (1, 10) according to claim 1, wherein the first relation (R) is a ratio for each amplitude spectrum. Device (1,10).
A third aspect of the present invention is the recording apparatus (1, 10) according to the first or second aspect, wherein a second noise estimation is performed for the sound collected by the second sound collecting unit (41R). A noise estimation unit (52R), a second noise reduction unit (53R) for reducing the noise estimated by the second noise estimation unit (52R) from the second sound, and the second after noise reduction The noise from the first adjustment unit (54L) for adjusting the first sound and the first noise reduction unit (53L) so that the relationship with the sound is the same as the first relationship (R). A first average unit (56L) that averages the reduced first noise after noise reduction and the adjusted first sound adjusted by the first adjustment unit (54L), and the second noise reduction unit (53R). ) In which the noise is reduced, the second sound after noise reduction, and the second sound Further comprising second averaging unit for averaging the second tone after adjustment adjusted by integer unit (54R) and (56R), and a recording apparatus according to claim (1, 10).
The invention according to claim 4 is the recording apparatus (1, 10) according to any one of claims 1 to 3, wherein the first sound is subjected to spectrum conversion before the noise estimation or the adjustment. The first conversion unit (51L), the first inverse conversion unit (55L) that performs spectrum inverse conversion on the first sound after the noise reduction or the adjustment, and the second sound before the noise estimation or the adjustment A recording apparatus comprising: a second conversion unit (51R) that performs spectrum conversion; and a second inverse conversion unit (55R) that performs spectrum inverse conversion on the second sound after noise reduction or adjustment. 1,10).
A fifth aspect of the present invention is the recording apparatus (1, 10) according to any one of the first to fourth aspects, wherein the first sound collection unit (41L) or the first The recording device (1, 10) is characterized in that one of the two sound collection units (41R) is selected and a selection unit for selecting the other for the left ear is provided.
The invention according to claim 6 performs the noise estimation of the first sound collected by the first sound collection unit (41L) of the plurality of sound collection units (41), and the estimated noise is determined by the first sound collection unit (41L). A second sound that is reduced from one sound and collected by the second sound collecting section (41R) of the plurality of sound collecting sections (41) and the first sound before noise reduction is the first relation. 1 is calculated, and the computer device is caused to execute processing for adjusting the second sound so that the relationship with the first sound after noise reduction is the same as the first relationship (R). It is a program.
Note that the configuration described with reference numerals may be modified as appropriate, and at least a part of the configuration may be replaced with another component.

  According to the present invention, it is possible to provide a recording apparatus and a program in which a sense of orientation and a sense of presence before noise reduction are maintained even after noise reduction.

It is the external appearance schematic perspective view seen from the front side of the camera to which one Embodiment of the recording device in this invention is applied. It is a block block diagram of a camera. It is a functional block diagram of a sound information processing part. It is a figure for demonstrating an example of the 1st frequency spectrum and 2nd frequency spectrum which a shock noise reduction process part acquires. A method for obtaining the frequency spectrum SSR of the right channel from the left channel will be described. It is a flowchart which shows operation | movement of a sound information processing part. It is a functional block diagram of the sound information processing part in 2nd Embodiment.

(First embodiment)
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic external perspective view seen from the front side of a camera 1 to which an embodiment of a recording apparatus according to the present invention is applied. FIG. 2 is a block configuration diagram of the camera 1.

As shown in FIGS. 1 and 2, the camera 1 includes a camera body 10 and a lens barrel 20. The camera 1 has an autofocus (hereinafter abbreviated as AF) function for automatically focusing. In addition, the camera 1 can shoot both still images and moving images, and can record sound in stereo simultaneously with images during moving image shooting.
Note that the camera 1 may be such that the lens barrel 20 can be replaced with respect to the camera body 10, and the camera body 10 and the lens barrel 20 may be integrated.

The camera body 10 includes an imaging unit 30, a recording unit 40, a recording unit 11, and a control unit 12.
The image capturing unit 30 includes an image sensor 31, an A / D conversion unit 32, and an image processing unit 33.
The image sensor 31 is configured by a photoelectric conversion element such as a CCD. The image sensor 31 converts the subject image light imaged by the imaging optical system of the lens barrel 20 into an analog electric signal and outputs it to the A / D conversion unit 32.

The A / D converter 32 converts the analog image signal input from the image sensor 31 into a digital image signal and outputs the digital image signal to the image processor 33.
The image processing unit 33 performs image processing on the digital image signal input from the A / D conversion unit 32 to generate image data, and outputs the image data to the recording unit 11.

The recording unit 40 includes a stereo sound collecting device 41, an A / D conversion unit 42, and a sound information processing unit 50.
As shown in FIG. 1, the stereo sound collecting device 41 includes a pair of left and right microphones (a left microphone 41L and a right microphone 41R). Each of the microphones 41L and 41R collects external sounds and outputs them to the A / D converter 42 as analog signals.
The A / D conversion unit 42 converts the analog sound signal input from the stereo sound collector 41 into a digital sound signal and outputs the digital sound signal to the sound information processing unit 50.

The sound information processing unit 50 reduces the operating noise included in the sound signals of the left microphone 41L and the right microphone 41R in the stereo sound collecting device 41 input from the A / D conversion unit 42, creates sound data, and records the sound data. To the unit 11.
Note that the operation noise to be reduced in this embodiment is drive noise (AF drive sound) related to AF control generated from the lens barrel 20. The sound information processing unit 50 will be described in detail later.

The recording unit 11 records the image data output by the imaging unit 30 (image processing unit 33) and the sound data output by the recording unit 40 (sound information processing unit 50) on a recording medium (not shown) such as a memory card. .
The control unit 12 includes a CPU and the like, and each of the camera 1 including each component to be described later of the lens barrel 20 according to set imaging conditions (for example, an aperture value, an exposure value, etc.). Centrally control the components. For example, the control unit 12 generates a drive control signal for driving an AF drive motor 23 in a lens barrel 20 described later, and outputs the drive control signal to the AF drive motor 23.
Moreover, the control part 12 is provided with the timing signal detection part 12a. The timing signal detector 12a detects the timing of AF control generated from the lens barrel 20, as will be described later.

The lens barrel 20 includes an imaging optical system including a focusing lens 21, a camera shake correction lens and a zooming lens (not shown), an AF encoder 22, and an AF drive motor 23.
The AF encoder 22 detects the position of the focusing lens 21 and outputs the detected position to the control unit 12 of the camera body 10 as drive control information for the AF drive motor 23.
The AF drive motor 23 moves and drives the focusing lens 21 in accordance with a drive control signal for controlling the position of the focusing lens 21 input from the control unit 16. The driving sound of the AF driving motor 23 is a main AF driving sound that is targeted for reduction processing by the sound information processing unit 50 of the recording unit 40 in the present embodiment.

  The camera 1 is controlled by the control unit 12 to perform a shooting operation when a shooting command is issued by a user pressing a shutter button (not shown). That is, the imaging unit 30 converts subject image light into an electrical signal and performs image processing to form image data, which is recorded (photographed) on a recording medium via the recording unit 11. The control unit 12 performs AF control for moving the focusing lens 21 via the AF drive motor 23 during shooting.

At the time of moving image shooting, the imaging unit 30 converts the subject image light into an electrical signal with a rolling shutter and sequentially captures it, and records an image of a predetermined frame (frame number) on a recording medium via the recording unit 11 per second. .
Further, as described above, the sound data collected by the recording unit 40 and subjected to sound information processing is recorded (recorded) on the recording medium via the recording unit 11 for each frame together with the image data.
During moving image shooting, AF control is performed in response to changes in the position of the subject throughout the shooting period, and AF drive sound is generated irregularly.

Next, the sound information processing unit 50 in the recording unit 40 will be described in detail. FIG. 3 is a functional block diagram of the sound information processing unit 50.
As described above, the sound information processing unit 50 reduces the AF driving noise included in the sound signals of the left microphone 41L and the right microphone 41R in the stereo sound collecting device 41 input from the A / D conversion unit 42.
Hereinafter, the flow of sound information collected by the left microphone 41L will be described as a left channel (abbreviated as Lch in the drawing), and the flow of sound information collected by the right microphone 41R will be described as a right channel (Rch). The left and right are the left and right viewed from the photographer side.

  The sound information processing unit 50 in the present embodiment performs noise reduction processing on the sound information of one channel (left channel in the present embodiment) by the spectral subtraction method, and uses this information for the other channel (right channel). The spectrum adjustment processing is performed on the sound information.

As shown in FIG. 3, the sound information processing unit 50 includes left and right channel spectrum conversion units 51L and 51R, a left channel noise spectrum estimation unit 52L, a left channel noise spectrum reduction unit 53L, and a right channel. A spectrum adjustment unit 54R and left and right channel inverse conversion units 55L and 55R are provided.
The noise spectrum reduction unit 53L further includes an impact sound noise reduction processing unit 53A and a driving sound noise reduction processing unit 53B.

Hereinafter, each part of the sound information processing unit 50 will be described in more detail.
(Spectrum converter)
The spectrum converters 51L and 51R weight the sound signals of both the left and right channels input from the A / D converter 42 with a window function for each predetermined section, and the sound signal for each section. Are converted into spectra SL and SR (see FIG. 5 to be described later) representing the amplitude for each frequency domain.

FIG. 4 is an example of a microphone sound signal collected by a microphone when the AF lens is driven. In the graph of FIG. 4, the vertical axis represents the microphone signal collected by the microphone 230, and the horizontal axis represents time. As shown in FIG. 4, when AF driving is performed, impact sounds are generated at times t10 to t11 and t20 to t21 shown in the drawing.
For convenience of explanation, FIG. 4 shows only the sound signal of the operation sound among the microphone sound signals, and the illustration of the sound signal of the target sound is omitted.

  The spectrum converters 51L and 51R weight the microphone sound signal output from the A / D converter 42 with a window function for each predetermined section. Then, the microphone sound signal for each section is subjected to, for example, Fourier transform or fast Fourier transform (FFT), converted into the frequency domain, and a frequency spectrum corresponding to each section of the window function is calculated. To do.

  Here, the predetermined section of the window function is a unit (frame) of signal processing, and is a section repeated at a constant interval. Each section of these window functions has an autofocus overlap with each section of other window functions. As the window function, for example, a Hanning window (Hanning window) function can be used.

An example of a frequency spectrum corresponding to each section of the window function calculated by the spectrum conversion units 51L and 51R will be described with reference to FIG.
As described above, the spectrum conversion units 51L and 51R weight the microphone sound signal output from the A / D conversion unit 42 with a window function that overlaps with other sections in half. Thus, the microphone sound signal is divided into window function sizes.
The spectrum converters 51L and 51R perform, for example, a Fourier transform for each microphone signal in each section weighted with the window function, and as shown in FIG. ~ S14 is calculated.

(Noise spectrum estimation unit)
The estimation of the noise spectrum is performed based on the timing detected from the timing signal detection unit 12a for the sound signal input from the left channel and converted to the frequency spectrum.

The timing signal detection unit 12a detects timing (operation change timing) at which the operation state of the lens barrel 20 changes.
Examples of the operation change timing include an operation start timing at which the focusing lens 21 starts operating and an operation stop timing at which the operation of the focusing lens 21 stops.

The timing signal detector 12 a can detect the operation change timing based on the drive control signal input to the AF drive motor 23.
Further, the timing signal detection unit 12a may detect the operation change timing based on a process or a command executed in the control unit 12 when generating the drive control signal.
Furthermore, the timing signal detection unit 12a may detect the operation change timing based on the operation signal input from the operation unit.
The timing signal detection unit 12a may detect the operation change timing based on an output based on a pulse signal output from the encoder 22, for example.

Then, the timing signal detection unit 12a outputs information indicating time t10 to t11 when the impact sound shown in FIG. 4 is generated as an operation start timing signal.
Further, the timing signal detection unit 12a outputs information indicating the time t20 to t21 when the impact sound shown in FIG. 4 is generated as an operation stop timing signal.

Based on the operation start timing signal and the operation stop timing signal from the timing signal detection unit 12a, the noise spectrum estimation unit 52L causes the frequency spectrum S2 to S4 in FIG. 4 to have an impact corresponding to the operation start timing t10 of the AF lens 21. It is estimated that the sound information includes sound generation periods t10 to t11.
Further, the noise spectrum estimation unit 52L estimates that the frequency spectrum S9 to S12 is sound information including the impact sound generation period t20 to t21 corresponding to the operation stop timing t20 of the AF lens 21.
Then, the noise spectrum estimation unit 52L estimates that the frequency spectra S5 to S8 are sound information corresponding to the generation period of the driving sound by the AF lens 21.

The noise spectrum estimation unit 52L compares the sum of the frequency components of the frequency spectrum corresponding to the impact sound generation period with a predetermined threshold value.
The predetermined threshold is the sum of the frequency components of the frequency spectrum of the target sound, which is assumed to have little voice deterioration due to the impact sound because the target sound is larger than the impact sound.

  When it is determined that the sum of the frequency components of the frequency spectrum corresponding to the impact sound generation period is less than a predetermined threshold, the noise spectrum estimation unit 52L uses the calculated frequency spectrums S1 to S14 of the noise spectrum reduction unit 53L. Output to the impact noise reduction processing unit 53A.

  On the other hand, when it is determined that the sum of the frequency components of the frequency spectrum corresponding to the impact sound generation period is equal to or greater than a predetermined threshold, the noise spectrum estimation unit 52L uses the calculated frequency spectrums S1 to S14 as the noise spectrum reduction unit. This is output to the 53L driving sound noise reduction processing unit 53B.

(Shock noise reduction processing part)
For example, the impact sound noise reduction processing unit 53A acquires a frequency spectrum (first frequency spectrum) corresponding to a period during which the impact sound is highly likely to be generated from the frequency spectra S1 to S14.
For example, the shock noise reduction processing unit 53A acquires the frequency spectrums S2 to S4 shown in the shock sound generation periods t10 to t11 corresponding to the operation start timing t10 as the first frequency spectrum. And the impact sound noise reduction process part 53A acquires the frequency spectrum S9-S12 shown in the impact sound generation period t20-t21 corresponding to the operation stop timing t20 as the first frequency spectrum.

Further, the shock noise reduction processing unit 53A acquires a frequency spectrum (second frequency spectrum) corresponding to a period in which there is a high possibility that the shock sound is not generated from the frequency spectra S1 to S14.
In the present embodiment, the impact noise reduction processing unit 53A acquires the first frequency spectrum and the frequency spectrum closest to the time axis direction as the second frequency spectrum.

For example, as shown in FIG. 4, the impact noise reduction processing unit 53A sets the second frequency spectrum corresponding to the frequency spectra S2 and S3, which are the first frequency spectrum, in the past direction of the time axis of the frequency spectra S2 and S3. The closest frequency spectrum S1 is acquired.
Further, the frequency spectrum S5 closest to the future direction on the time axis of the frequency spectrum S4 is acquired as the second frequency spectrum corresponding to the frequency spectrum S4 that is the first frequency spectrum.
Moreover, the frequency spectrum S8 closest to the past direction of the time axis of the frequency spectra S9, 10 is acquired as the second frequency spectrum corresponding to the frequency spectra S9, S10 that are the first frequency spectrum.
Further, the frequency spectrum S13 closest to the future direction on the time axis of the frequency spectrum S11, 12 is acquired as the second frequency spectrum corresponding to the frequency spectrum S11, 12 that is the first frequency spectrum.

Then, the shock noise reduction processing unit 53A replaces at least a part of the first frequency spectrum with a corresponding part of the second frequency spectrum.
As an example of this replacement, for example, the impact noise reduction processing unit 53A has a frequency spectrum equal to or higher than a predetermined threshold frequency in the first frequency spectrum and a frequency equal to or higher than a predetermined threshold frequency in the second frequency spectrum. When the spectrum is compared for each frequency component and it is determined that the second frequency spectrum is smaller than the first frequency spectrum, the frequency component in the first frequency spectrum is replaced with the frequency component of the second frequency spectrum. .

(Drive noise reduction processing part)
On the other hand, the drive sound noise reduction processing unit 53B is based on the timing signal input from the timing signal detection unit 12a, for example, a frequency spectrum corresponding to a period in which the drive sound is likely to be generated from the frequency spectra S1 to S14. (Third frequency spectrum) is acquired.
For example, the drive sound noise reduction processing unit 53B generates a drive sound based on an impact sound generation period t10 to t11 corresponding to the operation start timing t10 and an impact sound generation period t20 to t21 corresponding to the operation stop timing t20. The frequency spectrums S2 to S12 corresponding to the possible period are acquired as the third frequency spectrum.

The drive sound noise reduction processing unit 53B performs drive sound noise reduction processing for reducing noise that is predetermined according to the drive pattern, with respect to the acquired third frequency spectrum.
For example, the drive sound noise reduction processing unit 53B uses a frequency spectrum subtraction method that subtracts a frequency component of a frequency spectrum representing noise determined in advance according to a drive pattern from a frequency component of the third frequency spectrum.
Note that the frequency spectrum of noise that is predetermined according to the drive pattern is preset in the sound information processing unit 50 as a set value. However, the present invention is not limited to this, and the drive sound noise reduction processing unit 53B subtracts the frequency spectrum of the section where the drive sound is not generated from the frequency spectrum of the section where the drive sound is generated based on the past microphone sound signal. Thus, the frequency spectrum of the estimated noise of the driving sound may be calculated as a driving pattern.

  As described above, the impact sound noise reduction processing unit 53A and the driving sound noise reduction processing unit 53B of the noise spectrum reduction unit 53L have the frequency of the left channel where the occurrence of AF driving noise is detected for each frequency spectrum of each frame. From the spectrum SL, the frequency spectrum SSL of the left channel obtained by reducing the AF driving noise component is calculated.

(Spectrum adjustment unit)
FIG. 5 illustrates a method of obtaining the frequency spectrum SSR of the right channel from the left channel.
As shown in the figure, the spectrum adjustment unit 54R of the sound information processing unit 50 has the frequency of the right channel calculated by the spectrum conversion unit 51R with respect to the frequency spectrum SL of the left channel (before noise reduction processing) calculated by the spectrum conversion unit 51L. The right channel frequency spectrum SSR is obtained by adjusting the amplitude of the frequency spectrum SSL of the left channel frequency spectrum SSL after the noise reduction processing by the noise spectrum reducing unit 53L by the ratio of the amplitude of each spectrum SR in the frequency domain.

That is,
SL (k): Spectrum before noise reduction of left channel SR (k): Spectrum before noise reduction of right channel SSL (k): Spectrum after noise reduction of left channel SSR (k): After adjustment of right channel Spectrum where k is the frequency component number
The spectrum SSR (k) after adjustment of the right channel is
SSR (k) = SSL (k) * SR (k) / SL (k)
Ask for.

  In other words, the spectrum adjustment unit 54R obtains the ratio of the amplitude of the frequency spectrum SR of the right channel to the frequency spectrum SL of the right channel before the noise reduction processing for each frequency region (the left-right amplitude ratio: R in FIG. 5). For the frequency spectrum SSL of the left channel after the noise reduction processing, a frequency spectrum SSR of the right channel having the left / right amplitude ratio: R is calculated for each frequency region.

For example, as illustrated in FIG.
SL (f3) = 1.0
SR (f3) = 0.5
SSL (f3) = 0.6
In Case of,
R (f3) = 0.5 / 1.0 = 0.5
Because
SSR (f3) = 0.6 × 0.5 = 0.3
It becomes.

(Inverse conversion part)
The inverse transform units 55L and 55R respectively perform, for example, an inverse Fourier transform, or the frequency spectrum SSL that has been subjected to noise reduction processing by the noise spectrum reduction unit 53L, or the frequency spectrum SSR that has been adjusted by the spectrum adjustment unit 54R, respectively. By performing inverse fast Fourier transform (IFFT: Inverse Fast Fourier Transform), it transforms into the time domain. The inverse conversion unit 55L performs reverse conversion using the phase information of the Lch input sound, and the inverse conversion unit 55R performs reverse conversion using the phase information of the Rch input sound. Then, the inverse conversion units 55L and 55R output the sound signal converted into the time domain to the recording unit 140.

(Operation flowchart)
Next, the above-described operation in the sound information processing unit 50 will be described based on the flowchart of FIG.

  First, the sound signals of both the left and right channels input from the left and right microphones 41L and 41R are A / D converted by the A / D conversion unit 42 and input to the sound information processing unit 50 (step S10).

  The spectrum conversion units 51L and 51R of the sound information processing unit 50 calculate a frequency spectrum corresponding to each section of the window function, for example, by performing a Fourier transform on the microphone sound signal (step S11).

  The noise spectrum estimation unit 52L of the sound information processing unit 50 estimates the impact sound generation period from the operation change timing signal, and estimates the spectrum in which the impact sound is generated (step S12).

  The noise spectrum estimation unit 52L of the sound information processing unit 50 compares the sum of the frequency components of the frequency spectrum corresponding to the impact sound generation period with the predetermined threshold for the sound information of the left channel (step S13). ).

When the sum of the frequency components of the frequency spectrum corresponding to the impact sound generation period is smaller than the threshold value (step S13, YES), the impact sound noise reduction processing unit 53A of the sound information processing unit 50 may generate the impact sound. A frequency spectrum (first frequency spectrum) corresponding to a high period of time is acquired.
Further, the impact noise reduction processing unit 53A acquires a frequency spectrum (second frequency spectrum) corresponding to a period during which there is a high possibility that no impact sound is generated from the frequency spectra S1 to S14 (step S14).

  The shock noise reduction processing unit 53A obtains a shock noise reduction spectrum by replacing at least a part of the first frequency spectrum with a corresponding part of the second frequency spectrum (step S15).

The drive sound noise reduction processing unit 53B acquires a frequency spectrum corresponding to a period during which drive sound may be generated as a third frequency spectrum. Then, the driving sound noise reduction processing unit 53B reduces the noise determined in advance according to the driving pattern based on the acquired third frequency spectrum with respect to the impact sound noise reduction spectrum. Is performed (step S16).
In addition, also when the sum total of the frequency component of the frequency spectrum corresponding to an impact sound generation period is more than a threshold value (step S13, NO), it progresses to step S16.

  The spectrum adjustment unit 54R of the sound information processing unit 50 is a frequency region of the frequency spectrum SR of the right channel calculated by the spectrum conversion unit 51R with respect to the frequency spectrum SL of the left channel (before the noise reduction process) calculated by the spectrum conversion unit 51L. The right channel frequency spectrum SSR is obtained by adjusting the amplitude of the frequency spectrum SSL of the left channel frequency spectrum SSL after the noise reduction processing by the noise spectrum reduction unit 53L by the ratio of the amplitude of each (step S17).

  The spectrum adjustment unit 54R performs, for example, inverse Fourier transform on the frequency spectrum SSL that has been subjected to noise reduction processing by the noise spectrum reduction unit 53L or the frequency spectrum SSR that has been adjusted by the spectrum adjustment unit 54R, respectively. The time domain is converted (step S18).

According to the noise reduction process by the sound information processing unit 50 as described above, the left / right ratio of each frequency component of the frequency spectrum after the noise reduction process is maintained before and after the process. Thereby, it is possible to obtain a sound with reduced noise while maintaining the original orientation.
Further, it is possible to omit the calculation of the estimated noise spectrum in the right channel and the noise spectrum reduction (the noise spectrum estimation unit 52L and the noise spectrum reduction unit 53L in the left channel). Thereby, it is possible to reduce the amount of calculation and the necessary memory.

Here, in the above embodiment, the estimated noise spectrum is calculated and the noise spectrum is reduced in the left channel, and the right channel spectrum is adjusted based on the information.
However, the sharing of the left and right channels may be reversed. That is, the estimated noise spectrum may be calculated and the noise spectrum reduced in the right channel, and the left channel spectrum may be adjusted based on the information. Furthermore, it may be configured to switch according to the situation.

  For example, the control unit 12 determines whether the left microphone 41L or the right microphone 41R in the stereo sound collector 41 is closer to the noise generation source by switching the processing of the left and right channels depending on the situation. It is also possible to calculate the estimated noise spectrum and reduce the noise spectrum for the sound information of the channel of the closer microphone input, and adjust the spectrum of the other channel based on the information.

Further, as another configuration for switching the processing of the left and right channels depending on the situation, when the occurrence of AF driving noise in the sound signal is detected based on the output signal of the AF encoder 22, the channel on the side where the sound signal input is small The estimated noise spectrum may be calculated for the sound information and the noise spectrum may be reduced, and the spectrum of the other channel may be adjusted based on the information.
This is because if the conditions of noise input from the left microphone 41L and the right microphone 41R in the stereo sound collector 41 are substantially the same, the smaller the sound signal input, the greater the noise ratio. According to this, a more accurate noise reduction effect can be obtained.

Furthermore, as another configuration for switching the processing of the left and right channels depending on the situation, the fluctuation of the operating sound of the AF drive is detected, and the estimated noise spectrum of the sound information of the channel in which the sound signal fluctuates with the fluctuation is detected. You may comprise so that calculation and reduction of a noise spectrum may be performed and the spectrum adjustment of the other channel may be performed based on the information.
This is because there is a high possibility that the sound information of the channel in which the sound signal fluctuates with the fluctuation of the AF driving operation sound includes AF driving noise. Thereby, a highly accurate noise reduction effect can be obtained.
Further, as a further configuration for switching the processing of the left and right channels depending on the situation, a switch 43 for switching may be provided in the camera body 10 as indicated by a dotted line in FIG.

As described above, the first embodiment has the following effects.
(1) In the camera 1, the sound information processing unit 50 performs noise reduction processing on the sound information of one channel by the spectral subtraction method, and uses this information to convert the sound information of the other channel (right channel). A spectrum adjustment process is performed on the image.
The spectrum adjustment process is the ratio of the amplitude of the frequency spectrum of the other channel to the frequency spectrum of one channel before the noise reduction process and the frequency spectrum of the frequency spectrum after the noise reduction process of one channel. The frequency spectrum of the other channel is obtained by adjusting the amplitude.
Thereby, the left-right ratio of each frequency component of the frequency spectrum is maintained before and after the processing, and it is possible to obtain a sound with reduced noise while maintaining the original localization feeling and the realistic feeling.
(2) Moreover, calculation of the estimated noise spectrum and noise spectrum reduction in one channel can be omitted. Thereby, it is possible to reduce the amount of calculation and the necessary memory.

(Second Embodiment)
Next, a second embodiment of the present invention will be described with reference to FIG.
FIG. 7 is a block diagram of functional units of the sound information processing unit 500 in the second embodiment.
The sound information processing unit 500 in the second embodiment is replaced with the sound information processing unit 50 in the first embodiment described above, and other configurations are the same as those in the first embodiment. Is omitted. In addition, for each component in the sound information processing unit 500, components having the same functions as those of the sound information processing unit 50 in the first embodiment described above are denoted by the same reference numerals and description thereof is omitted.
Further, since the noise spectrum reduction unit 53L and the noise spectrum reduction unit 53R of the sound information processing unit 500 are the same as those in the first embodiment, the description and illustration thereof are omitted, but the impact sound noise reduction processing unit 53A and the drive sound, respectively, are omitted. A noise reduction processing unit 53B is included.

The sound information processing unit 500 illustrated in FIG. 7 includes a left channel priority processing unit 50L, a right channel priority processing unit 50R, and average value calculation units 56L and 56R.
The left channel priority processing unit 50L has the same configuration as the sound information processing unit 50 in the first embodiment described above, and includes spectrum conversion units 51L and 51R for both the left and right channels, and a noise spectrum estimation unit 52L for the left channel. A noise spectrum reduction unit 53L for the left channel, a spectrum adjustment unit 54R for the right channel, and inverse conversion units 55L and 55R for both the left and right channels. Then, the left channel priority processing unit 50L calculates the estimated noise spectrum and reduces the noise spectrum in the left channel, performs the right channel spectrum adjustment based on the information, and calculates the average value of the sound data of the left channel. The sound data of the right channel is output to the average value calculation unit 56R to the unit 56L.

  In the right channel priority processing unit 50R, the left and right channel processing is interchanged with the left channel priority processing unit 50L. The spectrum conversion units 51L and 51R for the left and right channels, the noise spectrum estimation unit 52R for the right channel, A noise spectrum reduction unit 53R for the channel, a spectrum adjustment unit 54L for the left channel, and inverse conversion units 55L and 55R for both the left and right channels are provided. Then, the right channel priority processing unit 50R calculates the estimated noise spectrum and reduces the noise spectrum in the right channel and adjusts the spectrum of the left channel based on the information, and converts the sound data of the left channel into an average value calculation unit. The sound data of the right channel is output to the average value calculation unit 56R at 56L.

  The average value calculation units 56L and 56R calculate average values from the left and right channel outputs of the left channel priority processing unit 50L and the right channel priority processing unit 50R, respectively. That is, the average value calculation unit 56L outputs an average value of the left channel output of the left channel priority processing unit 50L and the output of the left channel of the right channel priority processing unit 50R. The average value calculation unit 56R outputs an average value of the right channel output of the left channel priority processing unit 50L and the right channel output of the right channel priority processing unit 50R.

  According to the processing of the sound information processing unit 500 in the second embodiment, for example, even when the sound source is moving left and right, the processing is symmetric, and the same regardless of the moving direction of the sound source. Noise reduction effect can be obtained. In other words, if noise spectrum estimation and noise reduction processing is performed on only one of the left and right sides, the noise reduction effect can change depending on whether the sound source is moving from the left side to the right side or from the right side to the left side. This is not the case with this configuration.

  The sound information processing unit 500 includes a left channel priority processing unit 50L and a right channel priority processing unit 50R, and performs noise spectrum estimation and noise reduction processing in the left and right channels. As a result, there is a concern about an increase in the amount of calculation, but this may be dealt with by adjusting the number of samplings in the spectrum conversion.

As described above, the second embodiment has the following effects.
(1) The sound information processing unit 500 calculates the estimated noise spectrum and reduces the noise spectrum in the left channel, and adjusts the right channel spectrum based on the information, and estimates the right channel. The right channel priority processing unit 50R that calculates the noise spectrum and reduces the noise spectrum and adjusts the spectrum of the left channel based on the information, and the average value calculation unit 56L that averages the sound data of the left and right channels output by both , 56R. Thereby, it becomes a symmetrical process and the same noise reduction effect can be acquired irrespective of the movement of a sound source, etc.

(Deformation)
The present invention is not limited to the above-described embodiment, and various modifications and changes as described below are possible, and these are also within the scope of the present invention.
(1) In this embodiment, the present invention is applied to a camera 1 that includes a pair of (two) microphones (a left microphone 41L and a right microphone 41R) and performs stereo recording. However, the number of sound collection units (microphones) is not limited to this, and the present invention may be applied to, for example, a multichannel surround recording device including three or more microphones.
When a plurality of sound collection units are provided, noise reduction processing is performed on the sound information of a predetermined sound collection unit, and spectrum adjustment processing is performed on sound information of other sound collection units using the information. The processing does not necessarily have to be performed on all other sound information, and the sound information of some sound collecting units may not be processed.
Moreover, a plurality of sound collection units are divided into groups, and noise reduction processing is performed on sound information of a specific sound collection unit for each group, and the sound information of other sound collection units in the group is used by using the information. You may comprise so that a spectrum adjustment process may be performed.

(2) In the present embodiment, the noise to be reduced by the sound information processing unit 50 of the recording unit 40 is an AF drive sound generated by driving the AF drive motor 23 that moves and drives the focusing lens 21. However, the noise to be reduced is not limited to this. For example, the present invention may be applied to auto zoom drive sound, camera shake correction lens drive sound, and switch operation sound.

(3) In the above embodiment, the present invention is applied to a camera capable of recording simultaneously with moving image shooting, but may be applied to a recording apparatus that does not have a shooting function.
In addition, although embodiment and a deformation | transformation form can also be used in combination as appropriate, detailed description is abbreviate | omitted. Further, the present invention is not limited to the embodiment described above.

  1: Camera, 10: Camera body, 20: Lens barrel, 40: Recording unit, 41: Stereo sound collector, 41L: Left microphone, 41R: Right microphone, 50,500: Sound information processing unit, 50L: Left channel Priority processing unit, 50R: Right channel priority processing unit, 51L: Spectrum conversion unit, 51R: Spectrum conversion unit, 52L: Noise spectrum estimation unit, 52R: Noise spectrum estimation unit, 53L: Noise spectrum reduction unit, 53R: Noise spectrum reduction Part, 54L: spectrum adjustment part, 54R: spectrum adjustment part, 55L: inverse conversion part, 55R: inverse conversion part, 56L: average value calculation part, 56R: average value calculation part

Claims (6)

Multiple sound collectors;
A first noise estimation unit that performs noise estimation of the first sound collected by the first sound collection unit of the plurality of sound collection units;
A first noise reduction unit for reducing the noise estimated by the first noise estimation unit from the first sound;
A first relationship that is a relationship between the second sound collected by the second sound collecting unit of the plurality of sound collecting units and the first sound before noise reduction is obtained, and the first after noise reduction is obtained. A second adjustment unit that adjusts the second sound so that the relationship with the sound is the same as the first relationship;
A recording apparatus comprising: The recording device according to claim 1,
The first relationship is a ratio for each amplitude spectrum;
Recording device characterized by. The recording device according to claim 1 or 2,
A second noise estimation unit that performs noise estimation of the sound collected by the second sound collection unit;
A second noise reduction unit that reduces the noise estimated by the second noise estimation unit from the second sound;
A first adjustment unit that adjusts the first sound so that the relationship with the second sound after the noise reduction is the same as the first relationship;
A first averaging unit that averages the noise-reduced first sound from which the noise has been reduced by the first noise reduction unit and the adjusted first sound adjusted by the first adjustment unit;
A second averaging unit that averages the second sound after the noise has been reduced by the second noise reduction unit and the second sound after the adjustment adjusted by the second adjustment unit;
Providing
Recording device characterized by. The recording apparatus according to any one of claims 1 to 3,
A first converter that performs spectrum conversion on the first sound before the noise estimation or the adjustment;
A first inverse transform unit that inversely transforms the first sound after the noise reduction or the adjustment;
A second converter that performs spectral conversion of the second sound before the noise estimation or the adjustment;
A second inverse transform unit for performing spectrum inverse transform on the second sound after the noise reduction or the adjustment;
Providing
Recording device characterized by. The recording apparatus according to any one of claims 1 to 4,
For the right ear, a selection unit is provided that selects one of the first sound collection unit and the second sound collection unit and selects the other for the left ear,
Recording device characterized by. Performing noise estimation of the first sound collected by the first sound collection unit of the plurality of sound collection units;
Reducing the estimated noise from the first sound;
A first relationship that is a relationship between the second sound collected by the second sound collecting portion of the plurality of sound collecting portions and the first sound before noise reduction is obtained, and the first sound after noise reduction is obtained. A program that causes a computer device to execute a process of adjusting the second sound so that the relationship with one sound is the same as the first relationship.

Images