JP2012173371A - Imaging apparatus and noise reduction method for imaging apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an imaging apparatus and a noise reduction method for the imaging apparatus capable of reducing noise appropriately without degrading a target sound such as voice.SOLUTION: An imaging apparatus according to the present invention comprises: a sound collection device 131; a voice section detecting part 134 for detecting a voice section from sound information collected by the sound collection device 131; and a noise reduction processing part 133 for performing a different noise reduction process based on the detection result of the voice section detecting part 134.

Description

  The present invention relates to an image pickup apparatus that performs noise reduction processing from sound information input during shooting, and a noise reduction method for the image pickup apparatus.

At the time of moving image shooting in an image pickup apparatus capable of shooting a moving image, noise such as an operation sound (hereinafter referred to as AF noise) generated with the operation of the driving unit of the autofocus lens is collected by a sound collecting device such as a microphone, and the subject The sound may be mixed with the target sound such as voice and the quality of the target sound may be impaired.
As a method of reducing such AF noise, the power value of the audio signal input before the operation of the AF drive unit is acquired, and the flooring coefficient is controlled (changed) based on the power value of the audio signal. A method for removing noise has been proposed (see, for example, Patent Document 1).

JP 2008-252389 A

  However, in the case of the noise reduction processing according to Patent Document 1, although AF noise can be reduced, there is a problem that there is a high possibility that the target sound such as voice is deteriorated.

  An object of the present invention is to provide an image pickup apparatus and a noise reduction method for the image pickup apparatus that can appropriately reduce noise without causing deterioration of a target sound such as voice.

  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.

The invention according to claim 1 is a sound collecting device (131), a sound section detecting unit (134) for detecting a sound section from sound information collected by the sound collecting device (131), and the sound section detection. An image pickup apparatus (100) comprising: a noise reduction processing unit (133) that performs different noise reduction processing based on a detection result of the unit (134).
A second aspect of the present invention is the imaging apparatus (100) according to the first aspect, wherein the noise timing detection unit detects the generation timing of the operational noise from the operational information of the drive unit in the imaging apparatus (100). (135), and the noise reduction processing unit (133) performs different noise reduction processing based on the detection result of the noise timing detection unit (135). .
The invention described in claim 3 is the imaging apparatus (100) according to claim 1 or 2, wherein the noise reduction processing unit (133) is detected as a voice section by the voice section detection unit (134). In this case, the imaging apparatus (100) is characterized in that it performs a lower first noise reduction process that is weaker than when it is detected as a non-speech section by the speech section detection unit (134).
Invention of Claim 4 is an imaging device (100) of any one of Claim 1 to 3, Comprising: The said noise reduction process part (133) is in the said audio | voice area detection part (134). Image pickup characterized in that noise is estimated from sound information when it is determined to be a non-speech interval, and a second noise reduction process is performed to subtract the estimated noise from the estimated noise subtraction sound information. Device (100).
Invention of Claim 5 is an imaging device (100) of any one of Claim 1 to 4, Comprising: The said noise reduction process part (133) is in the said audio | voice area detection part (134). An imaging apparatus (100) characterized in that a flooring spectrum is obtained from sound information when it is determined to be a non-speech section, and flooring processing is performed on the sound information before flooring processing using the flooring spectrum. ).
A sixth aspect of the present invention is the imaging apparatus (100) according to any one of the first to fifth aspects, wherein the detection of the voice section by the voice section detection unit (134) is performed using one of the voice waveforms. An imaging apparatus (100) is characterized in that an autocorrelation function is obtained by cutting out a part and detecting using a peak value of the obtained autocorrelation function.
According to a seventh aspect of the present invention, the noise of the image pickup apparatus (100) is characterized in that a voice section is detected from collected sound information and different noise reduction processing is performed based on the detection result of the voice section. This is a reduction method.
The invention according to claim 8 is the noise reduction method according to claim 7, wherein the generation timing of the operation noise is detected from the operation information of the drive unit in the imaging device (100), and the generation timing of the operation noise is detected. The noise reduction method of the imaging apparatus (100) is characterized in that different noise reduction processing is performed based on the detection result of.
The invention according to claim 9 is the noise reduction method according to claim 7 or 8, wherein the first noise reduction is weaker when detected as a non-speech interval when detected as a speech interval. It is the noise reduction method of an imaging device (100) characterized by performing a process.
A tenth aspect of the present invention is the noise reduction method according to any one of the seventh to ninth aspects, wherein noise is estimated from sound information when it is determined to be a non-voice section, and the estimation is performed. A noise reduction method for the imaging apparatus (100), characterized in that a second noise reduction process is performed to subtract the generated noise from the pre-estimation noise subtraction sound information.
The invention according to claim 11 is the noise reduction method according to any one of claims 7 to 10, wherein a flooring spectrum is obtained from sound information when it is determined to be a non-voice section, A noise reduction method for an imaging apparatus (100), wherein flooring processing is performed on sound information before flooring processing using a flooring spectrum.
The invention according to claim 12 is the noise reduction method according to any one of claims 7 to 11, wherein the speech section is detected by extracting a part of the speech waveform and obtaining an autocorrelation function. It is a noise reduction method of the imaging device (100) characterized by detecting using the peak value of the calculated autocorrelation function.
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 an imaging apparatus and a noise reduction method for the imaging apparatus that can appropriately reduce noise without causing deterioration of a target sound such as voice.

It is a block diagram which shows the structure of the imaging device of embodiment of this invention. It is an audio | voice waveform diagram. It is a figure explaining the autocorrelation function of an audio | voice waveform. FIG. 4A shows an example of detecting a speech section using an autocorrelation function. FIG. 4A shows an output waveform from a microphone, and FIG. 4B shows a threshold value at a peak of the autocorrelation function. It is the waveform which showed the above part as High. It is a figure explaining the detail of the generation timing detection of the operation noise by a noise timing detection part. It is a flowchart which shows the flow of a noise reduction process operation. It is the schematic explaining the form of the 1st process target sound used as the object of a noise reduction process. It is a figure which shows the spectrum of the area A. FIG. It is a figure which shows the spectrum of the area B. FIG. FIG. 6 is a diagram showing a spectrum of a section C. It is a figure which shows an estimated noise spectrum. It is a figure which shows the spectrum which subtracted the noise from the spectrum of the area C. FIG. It is a figure which shows the spectrum after the flooring which uses the flooring spectrum A. FIG. It is a figure which shows the flooring spectrum A. It is a figure which shows the flooring spectrum B. FIG. It is a figure which shows the spectrum after the flooring which uses the flooring spectrum B. FIG. It is the schematic explaining the form of the 2nd process target sound used as the object of a noise reduction process. It is a figure which shows the spectrum of the background sound and noise of the area E. FIG. It is a figure which shows the estimation noise which uses the spectrum of the area E. FIG. It is a figure which shows the spectrum of the area F. FIG. It is a figure which shows the spectrum after performing a flooring process using the estimated noise of the area E. FIG. It is a figure which shows the estimation noise which uses the spectrum of the area F. FIG. It is a figure which shows the spectrum after performing a flooring process using the estimated noise of the area F. FIG.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of an imaging apparatus according to an embodiment of the present invention.

  As shown in FIG. 1, the imaging apparatus 100 includes a lens barrel 110 and an image processing unit that captures and subjects an object image that has passed through the lens barrel 110 to A / D conversion and generates image data by performing image processing. 120, the sound information processing unit 130 that performs A / D conversion on the collected sound information and performs noise reduction processing, the image data obtained by the image processing unit 120, and the audio signal obtained by the sound information processing unit 130 Recording section 140 and CPU 150 are recorded.

  The lens barrel 110 includes a focus adjustment lens (hereinafter referred to as AF (Auto Focus) lens, a camera shake correction lens (hereinafter referred to as VR (Vibration Reduction) lens), a zoom lens, a zoom lens driving unit, a zoom encoder, an image blur correction unit, and the like. A VR unit 111, an AF encoder 112, and an AF drive motor 113 are provided.

  The AF encoder 112 detects the position of the AF lens of the optical system and outputs it to the CPU 150. A driving control signal for controlling the position of the AF lens is input from the CPU 150 to the AF driving motor 113, and the position of the AF lens is controlled in accordance with the driving control signal.

  The CPU 150 controls the lens barrel 110 according to the set imaging conditions (for example, aperture value, exposure value, etc.). The CPU 150 generates a drive control signal for driving the zoom lens driving unit and the AF driving motor 113 and outputs the drive control signal to the zoom lens driving unit and the AF driving motor 113.

  The sound information processing unit 130 includes a microphone 131 that is a sound collection device, a sound signal processing unit 132 that processes sound information that has been collected and A / D converted, and a noise reduction processing unit 133.

  The sound signal processing unit 132 includes a voice section detection unit 134 that detects a voice section from sound information collected by the microphone 131, and a noise timing detection unit that detects timing at which operation noise occurs from the operation information of the AF driving motor 113. 135.

  Based on the peak value of the autocorrelation function, the speech section detection unit 134 determines a section (speech section) including a speech signal and other sections (non-speech section) from the sound information collected by the microphone 131. Determine. The outline of the voice zone detection by the voice zone detector 134 will be described as follows.

  FIG. 2 shows a speech waveform. When an arbitrary part of this speech waveform is cut out to obtain an autocorrelation function, the waveform shown in FIG. 3 is obtained. This voice waveform has the property (harmonic nature) that peaks concentrate on the fundamental frequency corresponding to the frequency of the voice, that is, the vocal cords, and the frequency band corresponding to its harmonics. Thus, it is possible to distinguish between speech and non-speech based on the magnitude of the peak value of the autocorrelation function.

  FIG. 4 shows an example in which a speech section is detected using an autocorrelation function. FIG. 4A shows an output waveform from the microphone 131, in which AF noise is generated in the first half and voice and AF noise are generated in the second half. When an autocorrelation function is obtained for the output waveform as shown in FIG. 4A, a threshold is set at the peak of the autocorrelation function, and a portion equal to or higher than the threshold is shown as High, as shown in FIG. Waveform is obtained. As a result, it can be detected that there is a voice section that matches the voice position in the latter half of the output waveform.

  The noise timing detection unit 135 detects the timing at which operation noise occurs from the operation information of the AF driving motor 113. The generation timing of the operation noise by the noise timing detector 135 is detected (estimated) using an AF drive command that instructs the CPU 150 to output a drive control signal for the AF drive motor 113 and an output from the AF encoder 112. .

The details of operation noise generation timing detection by the noise timing detection unit 135 will be described as follows.
As shown in FIG. 5, when the AF drive motor 113 is operated by the output of the AF drive command, the operation continues from the operation start time t1 of the AF drive motor 113, which is the output time of the AF drive command, to the operation end time t3. Operation noise is generated. The microphone 131 collects sound information in which operation noise is superimposed on the recording target sound such as the sound of the subject, and the collected sound information is output from the microphone 131.

At this time, the AF encoder 112 may output from a time t2 that is later than the operation start time t1 of the AF driving motor 113 due to the influence of backlash or the like that occurs in the gear train of the AF driving system. Therefore, the noise timing detection unit 135 detects from the output time t1 of the AF drive command to the output stop t3 of the AF encoder 112 as operation noise generation timing, and detects the other as non-noise timing.
Note that the signal actually output from the microphone 131 during the AF operation is a signal in which the operation noise is superimposed on the target sound, but only the operation noise is shown in FIG. 5 to simplify the description.

  The noise reduction processing unit 133 reduces the impact noise generated at the start of the AF operation and at the end of the AF operation among the operation noises shown in FIG. The noise reduction processing unit 133 obtains the first frequency spectrum of the window X before the generation of the operational noise and the second frequency spectrum of the window Y after the generation of the operational noise shown in FIG. The obtained first frequency spectrum is compared with the second frequency spectrum, and if the second frequency spectrum is larger than the first frequency spectrum as a result of the comparison, the first frequency spectrum is replaced with the first frequency spectrum, thereby replacing the first frequency spectrum with the first frequency spectrum. Perform noise reduction processing.

  Here, when it is detected that the speech section is a speech section by the speech section detection unit 134, the spectrum up to a predetermined frequency (for example, 4000 Hz) is stored without replacement, and when it is detected that it is a non-speech section, The spectrum up to a predetermined frequency (for example, 500 Hz) smaller than that is stored without being replaced. That is, the upper limit of the frequency to be saved when it is detected as a voice section is, for example, 4000 Hz, and the upper limit of the frequency to be saved when it is detected as a non-voice section is, for example, 500 Hz. Is detected, a first impact noise reduction process that is weaker than that detected when the non-speech section is detected is performed.

  In addition, the noise reduction processing unit 133 estimates noise from the frequency spectrum when the voice segment detection unit 134 detects that it is a non-speech segment and performs strong impact noise reduction processing, and updates the estimated noise. Then, using the estimated noise, a spectrum subtraction process (second noise reduction process) for generating a frequency spectrum by subtracting from the frequency spectrum on which the first impact sound noise reduction process has been performed is performed.

  In addition to the above-described configuration, the sound information processing unit 130 divides the sound information output from the microphone 131 into predetermined intervals and weights them with a window function, and Fourier-transforms the sound data for each interval. (FFT: Fast Fourier Transform) and a processing unit that converts the frequency domain. Further, an inverse Fourier transform (IFFT: IFFT) is performed on a spectrum that is divided into frequency domain amplitude information and phase information by FFT processing and subjected to noise reduction processing (spectral subtraction processing) using the frequency domain amplitude information. A processing unit that converts the spectrum (sound information) after the noise reduction processing into a time domain by performing an Inverse Fast Fourier Transform). The illustration of these processing units is omitted.

Further, the noise reduction processing unit 133 has a flooring function for correcting the spectrum when the spectrum is significantly reduced or the spectrum disappears by the second noise reduction processing (spectrum subtraction processing). This flooring is a flooring spectrum generated based on sound information when the noise timing detection unit 135 detects non-noise timing and the voice interval detection unit 134 detects non-sound timing. And the spectrum after subtraction in the second noise reduction processing, and if the spectrum after subtraction is lower than the flooring spectrum (spectrum intensity is small), the sound information (spectrum) adopting the flooring spectrum And IFFT process this.
However, if the spectrum after subtraction exceeds the flooring spectrum (the spectrum intensity is high), the flooring process may or may not be performed.

  Further, the flooring spectrum used for the flooring function uses sound information when the noise timing detection unit 135 detects non-noise timing and the voice segment detection unit 134 detects non-speech segment. Update. As a result, the flooring spectrum does not include either the operating noise spectrum or the voice spectrum, but only the background sound is included. The voice spectrum is added during the flooring process, and is not originally present in the sound information after the noise reduction process. The sound is not generated.

  Next, the noise reduction processing operation (noise reduction method) in the imaging apparatus 100 of the present embodiment will be described based on the drawings. FIG. 6 is a flowchart showing the flow of the noise reduction processing operation. FIG. 7 is a schematic diagram illustrating the form of the first processing target sound that is the target of the noise reduction processing.

(First processing target sound)
As shown in FIG. 7, the first processing target sound is a form in which section A generates only background sound, section B generates background sound and sound (target sound), and section C generates background sound and AF noise. . An operation for reducing AF noise from the sound information collected and output by the microphone 131 in the section C of the first processing target sound shown in FIG. 7 and flooring update will be described.

(Step ST1)
First, the noise timing detection unit 135 starts detection of noise timing based on sound information output from the microphone 131.
The sound information (spectrum) collected by the microphone 31 at this time is shown in FIG.

(Step ST2)
Subsequently, the voice segment detection unit 134 starts detection of the voice segment based on the sound information output from the microphone 131.

(Step ST3)
Sound information output from the microphone 131 is subjected to FFT processing, and is divided into frequency domain amplitude information and phase information.

(Step ST4)
Next, the noise timing detection unit 135 detects (determines) whether it is the generation timing of the operation noise or the non-noise timing (that is, whether or not it is the AF section).

(Step ST4, YES)
In step ST4, it is determined that section C is the timing of occurrence of operation noise (AF section, YES), and the process proceeds to step ST5.
(Step ST4, NO)
The sections A and B are determined to be non-noise timing, and the process proceeds to step ST11.

(Step ST5)
In step ST5, the speech segment detection unit 134 detects (determines) whether it is a speech segment or a non-speech segment. Since section C is a non-voice section (NO), the process proceeds to step ST7.
(Step ST7)
Here, when the AF operation start time and AF operation end time are included, a strong impact noise reduction process is performed in which the upper limit is stored without replacing the spectrum up to a predetermined frequency (for example, 500 Hz). The spectrum of FIG. 10 is obtained.
When the AF operation start time and AF operation end time are not included, it is determined that the impact noise is not included, and the impact noise reduction process is not performed.

(Step ST8)
Next, the noise in the spectrum (FIG. 10) obtained by the noise reduction process in step ST7 is estimated, and an estimated noise spectrum as shown in FIG. 11 is output to step ST9.

(Step ST9)
Subsequently, a spectrum subtraction process (second noise reduction process) for subtracting the estimated noise spectrum (FIG. 11) obtained by the estimation of step ST8 from the spectrum (FIG. 10) obtained by the shock noise reduction process of step ST7. And a spectrum as shown in FIG. 12 is obtained.

(Step ST10)
The second noise reduction process (spectrum subtraction process) may cause the spectrum of FIG. 12 to be significantly reduced or lost, and therefore, flooring for correcting the spectrum of FIG. 12 is performed to cope with this. .
In this flooring, the magnitudes of the spectrum of FIG. 12 and the reference flooring spectrum are compared. Then, as a result of comparison, a spectrum having a high intensity is adopted to generate a spectrum shown in FIG. Although the flooring spectrum used here is mentioned later, it is a flooring spectrum calculated | required from the area A. FIG.

(Step ST11)
Returning to step ST11, here, the voice section detection unit 134 detects (determines) whether it is a voice section or a non-speech section (section of only background sound). As a result, it is determined that the section B is a voice section (YES), and noise reduction processing, spectrum subtraction, and flooring are not performed, and the process proceeds to step ST13. The section A is determined to be a non-voice section (NO), and the process proceeds to step ST12.

(Step ST12)
In step ST12, the amplitude at each frequency of the spectrum of the section A in which only the background sound shown in FIG. 8 is generated is halved to obtain a flooring spectrum as shown in FIG. This flooring spectrum (FIG. 14) is used for flooring in step ST10 as described above, and is updated to this flooring spectrum.
If flooring is performed using the flooring spectrum of FIG. 15 obtained by halving the amplitude at each frequency of the spectrum shown in FIG. 9 in section B, the spectrum shown in FIG. 16 is obtained. If the spectrum of FIG. 16 is the spectrum of the section C, the components (particularly f2 and f4) of the speech spectrum included in the section B (FIG. 9) are also included, and an accurate target sound cannot be obtained.
However, according to this embodiment, the flooring spectrum (FIG. 14) used for flooring does not include the spectrum of voice and operation noise. For this reason, in flooring of step ST10, it is possible to prevent AF noise and voice spectrums from being added, and undesired operation noise and voice from being generated in the sound information after the noise reduction processing.

(Step ST13)
In the final step ST13, IFFT processing is performed using the phase divided in step ST3, thereby converting the spectrum after the noise reduction processing into the time domain and outputting it to the recording unit 140.

(Second processing target sound)
Next, a noise reduction processing operation (noise reduction method) when using a second processing target sound having a form different from the first processing target sound described above will be described. In addition, since each step of the noise reduction processing operation flow is substantially the same in the case of the first processing target sound, the description will mainly focus on differences in processing contents in each step.

  FIG. 17 is a schematic diagram illustrating the form of the second processing target sound that is the target of the noise reduction processing. As shown in FIG. 17, the processing target sound has a form in which section D generates only background sound, section E generates background sound and AF noise, and section F generates background sound, voice, and AF noise. The operation for reducing AF noise from the sound information collected and output by the microphone 131 in the section E and the section F of the processing target sound shown in FIG. 17 and the flooring update will be described.

Steps ST1 to ST4 are the same as the above-described section C of the first processing target sound, and are therefore omitted.
(Step ST5)
In step ST5, it is determined that the section F is a voice section (YES), and the process proceeds to step ST6.
(Step ST6)
In step ST6, when the AF operation start time and AF operation end time are included, a weak first impact noise reduction process is performed in which the upper limit is stored without replacing the spectrum up to a predetermined frequency (for example, 4000 Hz). It is.
When the AF operation start time and AF operation end time are not included, it is determined that the impact noise is not included, and the impact noise reduction process is not performed.

The spectrum subjected to the first impact sound noise reduction processing includes speech vector components f2 and f4. This spectrum is not used for updating the estimated noise, and the process proceeds to step ST9 for performing the spectrum subtraction process which is the second noise reduction process.
In the case of the second processing target sound, a spectrum as shown in FIG. 18 is obtained in the section E which is the generation timing of the operation noise and is a non-speech section, and in the section F, a spectrum as shown in FIG. can get.

  Therefore, in step ST8, noise is estimated from the spectrum obtained in section E and updated. The estimated noise after the update has a spectrum as shown in FIG.

In step ST9, the estimated noise spectrum (FIG. 19) is subtracted from the spectrum in the section F (FIG. 20), and further, flooring is performed in step ST10, thereby generating a spectrum as shown in FIG.
The flooring spectrum in the case of the second processing target sound is obtained from the section D in which only the background sound is generated. As the flooring spectrum, the spectrum of FIG. 14 obtained by halving FIG. 8 as in the case of the first processing target sound is used.

  Here, if spectrum subtraction is performed based on the estimated noise diagram 22 obtained by multiplying the spectrum of the section F (FIG. 20) by 0.9, the spectrum shown in FIG. 23 is obtained. In this case, the sound spectrums indicated by f2 and f4 are also subtracted, and correct sound information cannot be obtained. However, according to the present embodiment, the voice spectrum can exist as shown in FIG.

As described above, this embodiment has the following effects.
(1) When a voice section is detected from the sound information collected by the microphone 131 and is detected as a voice section, the first noise reduction processing that is weaker than when it is detected as a non-voice section is performed. Therefore, noise is appropriately reduced without causing deterioration of the target sound, particularly the voice portion of the target sound consisting of the voice and the background sound, as compared with the case of performing strong noise reduction processing without distinguishing between the voice section and the non-voice section. Can do.
(2) Second noise reduction processing (spectrum subtraction processing) in which noise is estimated from sound information when it is determined to be a non-speech section after the first noise reduction processing, and the estimated noise is subtracted. To do. Therefore, since the noise is obtained from the sound information in the non-speech section, it is possible to obtain a processed sound very close to the target sound without deleting the sound itself.
(3) The operation noise generation timing is detected from the operation information of the drive unit in the imaging apparatus 100, and the process proceeds to the noise reduction process when the noise generation timing is detected. Therefore, it is possible to appropriately and rationally perform noise reduction processing only when necessary without performing useless noise reduction processing.
(4) Since flooring is performed on the sound information after the second noise reduction process (spectrum subtraction process), it is possible to correct a spectrum that may be reduced or disappear due to spectrum subtraction. As a result, it is possible to prevent the noise from being excessively reduced and to secure (record) a sound close to the target sound among the collected sound information.

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.
For example, in the present embodiment, the configuration has been described in which noise information collected by the microphone 131 is subjected to noise reduction processing in real time. However, the present invention is not limited to this, and the sound information collected by the microphone 131 is temporarily stored in a buffer memory or the like, and the sound information is read from the buffer memory or the like as necessary to perform noise reduction processing. May be. In this case, it is possible to reduce the load on the apparatus when processing in real time.
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.

  100: imaging device, 131: microphone (sound collecting device), 133: noise reduction processing unit, 134: voice section detection unit, 135: noise timing detection unit, 136: first noise reduction processing unit, 137: second Noise reduction processing section

Claims (12)

A sound collector;
A voice section detector for detecting a voice section from sound information collected by the sound collecting device;
A noise reduction processing unit that performs different noise reduction processing based on the detection result of the voice section detection unit,
An imaging apparatus characterized by the above. The imaging apparatus according to claim 1,
A noise timing detection unit for detecting operation noise generation timing from the operation information of the drive unit in the imaging device;
The noise reduction processing unit performs different noise reduction processing based on a detection result of the noise timing detection unit;
An imaging apparatus characterized by the above. The imaging apparatus according to claim 1, wherein:
The noise reduction processing unit
Performing a first noise reduction process that is weaker than when detected as a non-speech segment in the speech segment detection unit when the speech segment detection unit detects a speech segment;
An imaging apparatus characterized by the above. The imaging apparatus according to any one of claims 1 to 3,
The noise reduction processing unit
Performing a second noise reduction process for estimating noise from sound information when it is determined as a non-speech section in the speech section detection unit, and subtracting the estimated noise from estimated sound subtraction sound information ,
An imaging apparatus characterized by the above. The imaging apparatus according to any one of claims 1 to 4, wherein:
The noise reduction processing unit
Obtaining a flooring spectrum from sound information when it is determined as a non-voice section in the voice section detection unit, and flooring processing the sound information before flooring processing using the flooring spectrum;
An imaging apparatus characterized by the above. The imaging apparatus according to any one of claims 1 to 5,
The detection of the voice section by the voice section detection unit is to cut out a part of the voice waveform to obtain an autocorrelation function, and to detect using the peak value of the obtained autocorrelation function,
An imaging apparatus characterized by the above. Detects the voice section from the collected sound information,
Performing different noise reduction processing based on the detection result of the voice section,
A noise reduction method for an image pickup apparatus. The noise reduction method according to claim 7,
Detecting the generation timing of the operation noise from the operation information of the drive unit in the imaging device;
Perform different noise reduction processing based on the detection result of the operation noise occurrence timing,
A noise reduction method for an image pickup apparatus. The noise reduction method according to claim 7 or 8,
Performing a first noise reduction process that is weaker than when it is detected as a non-speech section when it is detected as a speech section;
A noise reduction method for an image pickup apparatus. It is the noise reduction method of the imaging device according to any one of claims 7 to 9,
Performing a second noise reduction process for estimating noise from sound information when it is determined to be a non-speech section and subtracting the estimated noise from the pre-estimated noise subtraction sound information;
A noise reduction method for an image pickup apparatus. The noise reduction method according to any one of claims 7 to 10,
Obtaining a flooring spectrum from sound information when it is determined to be a non-speech section, and flooring the sound information before flooring using the flooring spectrum;
A noise reduction method for an image pickup apparatus. The noise reduction method according to any one of claims 7 to 11,
The detection of the speech section is to cut out a part of the speech waveform to obtain an autocorrelation function, and to detect using the peak value of the obtained autocorrelation function,
A noise reduction method for an image pickup apparatus.

Images