JP2013161041A - Signal processor, camera and signal processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a signal processor that properly performs noise reduction.SOLUTION: A signal processor 1 includes transformation means 16 that transforms an acoustic signal to a spectrum signal for each prescribed frame, spectrum suppression means 16 that makes substantially zero a signal of frequency component to be a noise reduction object from the spectrum signal transformed by the transformation means 16, and inverse transformation means 16 that inversely transforms the spectrum signal after spectrum suppression to an acoustic signal.

Description

  The present invention relates to a signal processing device, a camera, and a signal processing program.

  When recording an acoustic signal acquired during photographing together with a photographed image, the mechanical sound at the time of the photographing operation that becomes noise is stored in advance as a spectrum signal, and the spectrum is determined based on the timing at which the mechanical sound is generated. There is known a technique for suppressing a mechanical sound mixed in an acoustic signal by subtracting a signal obtained by multiplying a signal by a predetermined coefficient from the acquired acoustic signal (see Patent Document 1).

JP 2006-279185 A

  In general, the mechanism sound at the time of AF (autofocus) operation is an unsteady sound that varies with time. For this reason, when a large mechanical sound is generated, there is a possibility that noise may remain due to undersubtraction. On the other hand, when a small mechanical sound is generated, excessive subtraction results in deterioration of sound quality.

  The signal processing apparatus according to the present invention includes a conversion unit that converts an acoustic signal into a spectrum signal for each predetermined frame, and a spectrum suppression that makes the frequency component signal to be reduced noise substantially zero from the spectrum signal converted by the conversion unit. Means, and inverse conversion means for inversely converting the spectrum signal after spectrum suppression into an acoustic signal.

  According to the present invention, it is possible to appropriately reduce noise.

It is a block diagram which illustrates the composition of the electronic camera carrying the signal processing device for noise reduction by one embodiment of the present invention. It is a flowchart explaining the flow of the noise reduction process which CPU performs. It is a figure which illustrates a computer apparatus.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating the configuration of an electronic camera 1 equipped with a signal processing device for noise reduction according to an embodiment of the present invention. In FIG. 1, an electronic camera 1 includes a photographing optical system 11, an image sensor 12, an image processing unit 13, a RAM 14, an LCD monitor 15, a CPU 16, a nonvolatile memory 17, and a card interface (I / F). 18, a communication interface (I / F) 19, an operation member 20, a microphone 21, and an acoustic processing circuit 22.

  The CPU 16, the nonvolatile memory 17, the card interface 18, the communication interface 19, the sound processing circuit 22, the image processing unit 13, the RAM 14, and the LCD monitor 15 are connected via a bus 25.

  The photographing optical system 11 includes a plurality of lens groups including a zoom lens and a focusing lens, and forms a subject image on the light receiving surface of the image sensor 12. In order to simplify FIG. 1, the photographing optical system 11 is shown as a single lens.

  The imaging element 12 is configured by a CMOS image sensor or the like in which light receiving elements are two-dimensionally arranged on the light receiving surface. The image sensor 12 photoelectrically converts an image of the light beam that has passed through the photographing optical system 11 to generate a digital image signal. The digital image signal is input to the image processing unit 13. The image processing unit 13 performs various types of image processing (color interpolation processing, gradation conversion processing, contour enhancement processing, white balance adjustment processing, etc.) on the digital image data.

  The LCD monitor 15 is composed of a liquid crystal panel or the like. The LCD monitor 15 displays an image, an operation icon, a menu screen, and the like according to an instruction from the CPU 16. The RAM 14 temporarily stores digital image data in a pre-process and post-process of image processing by the image processing unit 13 and temporarily stores digital acoustic data in a pre-process and post-process of noise reduction processing described later. Or used as a working memory when the CPU 16 executes the program. The nonvolatile memory 17 is configured by a flash memory or the like. Since the nonvolatile memory 17 retains the stored contents even when the power is turned off, the nonvolatile memory 17 stores a program executed by the CPU 16 and the like.

  The CPU 16 controls the operation performed by the electronic camera 1 by executing a program stored in the nonvolatile memory 17. The CPU 16 also performs AF (autofocus) operation control and automatic exposure (AE) calculation. In the AF operation, for example, an in-focus position of a focusing lens (not shown) is obtained based on the contrast information of the live view image. The live view image refers to a monitor image that is repeatedly acquired by the image sensor 12 at a predetermined time interval (for example, 30 frames / second) before the release operation.

  The card interface 18 has a connector (not shown), and a storage medium 30 such as a memory card is connected to the connector. The card interface 18 writes data to the connected storage medium 30 and reads data from the storage medium 30. The storage medium 30 is configured by a memory card incorporating a semiconductor memory, a hard disk drive, or the like.

  For example, the communication interface 19 performs communication using an TCP / IP protocol with an external device connected to a connector (not shown). Through this communication, commands and data from an external device are received, and image data, acoustic data, and the like stored in the storage medium 30 are transmitted to the external device. The operation member 20 includes a power switch, and other operation members such as a release button, a recording button, and a cross key switch. The operation member 20 sends an operation signal corresponding to each instruction such as a still image shooting instruction, a recording (and recording) instruction, a mode switching instruction, and various selection instructions to the CPU 16.

  The acoustic processing circuit 22 amplifies the acoustic signal collected by the microphone 22 and converts the amplified signal into digital acoustic data by an A / D conversion circuit (not shown). The lens driving mechanism 23 moves a focusing lens (not shown) forward and backward in the optical axis direction in response to an instruction from the CPU 16.

  The electronic camera 1 has a function of capturing a still image in response to a release button pressing operation and a function of recording and recording a moving image in response to a recording button pressing operation. When recording, the CPU 16 starts recording and recording a moving image when the recording button is pressed. For example, imaging is started at the same frame rate of 30 frames / second as that of the live view image, and acquisition of acoustic data is started. When the recording button is pressed again, image acquisition and sound acquisition are terminated, and a file for storing a moving image constituted by the acquired frame image group and a file for storing the acquired sound data are generated.

  Note that the acoustic data file generated by the CPU 16 of the present embodiment includes driving information of the lens driving mechanism 23 represented on the same time axis as the acoustic data. The drive information is data indicating whether or not the lens driving mechanism 23 is driving the focusing lens during recording. For example, time data indicating a driving start time and a driving end time may be used, or binary data indicating “1” during driving and “0” during non-driving may be used.

  Since the present embodiment has a feature in the noise reduction processing for the acoustic data acquired at the time of recording described above, the following description will be focused on the processing executed by the CPU 16 for noise reduction.

  FIG. 2 is a flowchart for explaining the flow of noise reduction processing executed by the CPU 16. For example, the CPU 16 temporarily stores the acoustic data in the RAM 14 and records the acoustic data after executing the noise reduction processing program in the storage medium 30. CPU16 repeats the process which performs noise reduction for every frame which comprises acoustic data.

  In step S10 of FIG. 2, the CPU 16 reads out acoustic data for one frame and driving information of the lens driving mechanism 23 from the RAM 14, and proceeds to step S20. In step S20, the CPU 16 performs a short-time Fourier transform on the sound data of the target frame and proceeds to step S30. Thereby, the amplitude (referred to as a frequency feature spectrum) for each frequency component of the acoustic signal is obtained. For example, when the sampling frequency is 44.1 kHz and divided into about 2000 frequency components, the spectral width per frequency component is about 20 Hz.

  In step S30, the CPU 16 determines whether sound is included in the acoustic data. Generally, a human voice has a plurality of “mountain” waveforms at a predetermined frequency interval at a frequency of 2 kHz or less. When the plurality of “mountain” waveforms are arranged at the frequency interval in the frequency band of 2 kHz or less in the frequency characteristic spectrum, and the peak value of the “mountain” waveform is equal to or greater than the first predetermined value, the CPU 16 affirms step S30. Determine and proceed to step S40. If the determination in step S30 is affirmative, there is a high possibility that the sound data of the target frame includes conversation, narration, and the like. Therefore, the sound data of the frame is not targeted for noise reduction, and the sound data of the frame is set to the environmental sound threshold value. Not used for decision. The environmental sound threshold is a determination threshold used for determining whether or not to perform noise reduction described later. On the other hand, if the “mountain” waveform in which the peak value is not less than the first predetermined value in the frequency band of 2 kHz or less is not arranged at the frequency interval, the CPU 16 makes a negative determination in step S30 and proceeds to step S60.

  In step S40, the CPU 16 performs short-time inverse Fourier transform on the frequency feature spectrum of the target frame, and proceeds to step S50. In step S50, the CPU 16 records the acoustic data after the above processing in the RAM 14, and ends the processing shown in FIG. In the present embodiment, the acoustic data of the target frame is not targeted for noise reduction unless the process proceeds to step S110 described later. That is, if the process does not proceed to step S110, the sound data of the target frame is directly stored, and if the process proceeds to step S110, the sound data processed by S110 is stored.

  In step S60, which proceeds after making a negative determination in step S30 described above, the CPU 16 determines whether or not the sum of the frequency feature spectra in the target frame (that is, the sum of the amplitude values for each of the approximately 2000 frequency components) is greater than or equal to the second predetermined value. Determine. If the sum is equal to or greater than the second predetermined value, the CPU 16 makes a positive determination in step S60 and proceeds to step S40. If the determination in step S60 is affirmative, there is a high possibility that sudden sound is included in the sound data of the target frame. Therefore, the sound data of the frame is not targeted for noise reduction, and the sound data of the frame is set to the environmental sound threshold value. Not used for decision. On the other hand, if the sum is less than the second predetermined value, the CPU 16 makes a negative determination in step S60 and proceeds to step S70.

  In step S <b> 70, the CPU 16 determines whether the acoustic data in the target frame is data acquired while the movable unit is being driven. For example, when the driving information of the lens driving mechanism 23 described above indicates that the frame corresponds to the driving of the focusing lens, the CPU 16 makes a positive determination in step S70 and proceeds to step S90. When the determination in step S70 is affirmative, there is a high possibility that the moving part drive sound is included in the sound data of the target frame, and therefore the sound data of the frame is not used for determining the environmental sound threshold. If the above-described driving information of the lens driving mechanism 23 indicates that the frame does not correspond to driving of the focusing lens, the CPU 16 makes a negative determination in step S70 and proceeds to step S80.

  In step S80, the CPU 16 determines and updates the environmental sound threshold using the frequency feature spectrum of the target frame, and proceeds to step S40. For example, the CPU 16 extracts a maximum value for each of the above approximately 2000 frequency components from the frequency feature spectrum of each frame from the first frame to the target frame constituting the acoustic data, and a frequency composed of the extracted approximately 2000 frequency components. The feature spectrum is an environmental sound threshold spectrum. When the value of any frequency component of the frequency characteristic spectrum of the frame corresponds to the maximum value, the CPU 16 replaces the value of the corresponding frequency component of the environmental sound threshold spectrum with the maximum value to replace the environmental sound threshold spectrum. Update. Thereby, the sound data of the most recent frame that does not include the movable part drive sound can be reflected in the environmental sound threshold spectrum.

  In step S90, the CPU 16 determines whether or not the frequency feature spectrum in the target frame has fluctuated by a third predetermined value or more. The CPU 16 makes an affirmative determination in step S90 when at least one frequency component of the frequency characteristic spectrum in the frame (that is, the frequency component of about 2000) increases or decreases more than a third predetermined value compared to the previous frame. Proceed to step S100. If the frequency feature spectrum in the frame does not include a frequency component that is larger or smaller than the third predetermined value compared to the previous frame, the CPU 16 makes a negative determination in step S90 and proceeds to step S40. If the determination in step S90 is affirmative, there is a high possibility that the acoustic data of the target frame has suddenly changed. Therefore, the acoustic data of the frame is not targeted for noise reduction, and the acoustic data of the frame is used to determine the environmental sound threshold. Also do not use.

  In step S100, the CPU 16 determines whether or not there is a frequency component having a value larger than the value of the corresponding frequency component of the environmental sound threshold spectrum among the frequency feature spectrum in the target frame (that is, the frequency component of about 2000). judge. The CPU 16 makes a positive determination in step S100 and proceeds to step S110 when at least one frequency component of the frequency characteristic spectrum in the frame has a value larger than the corresponding frequency component of the environmental sound threshold spectrum. If the frequency feature spectrum in the frame does not have a value greater than the value of the corresponding frequency component of the environmental sound threshold spectrum, the CPU 16 makes a negative determination in step S100 and proceeds to step S40. When a negative determination is made in step S100, it is determined that the moving part drive sound included in the sound data of the frame is within the allowable range, and the sound data of the frame is not targeted for noise reduction. However, since the sound data of the frame includes the moving part driving sound, the sound data of the frame is not used for determining the environmental sound threshold.

  In step S110, the CPU 16 replaces the value of the frequency component having a value larger than the value of the corresponding frequency component of the environmental sound threshold spectrum with 0 in the frequency feature spectrum (that is, the frequency component of about 2000) in the target frame. Then, the process proceeds to step S40. By suppressing to 0, even if the value before suppressing to 0 is an unsteady sound, there is no under-subtraction or over-subtraction, so noise remains (in the case of under-subtraction) and sound quality deteriorates. (In the case of excessive subtraction).

According to the embodiment described above, the following effects can be obtained.
(1) An electronic camera 1 equipped with a signal processing device that performs noise reduction is subject to noise reduction from a CPU 16 that Fourier transforms an acoustic signal into a frequency feature spectrum for each predetermined frame, and the frequency feature spectrum converted by the CPU 16. Since the CPU 16 that sets the frequency component signal to approximately 0 and the CPU 16 that performs inverse Fourier transform of the frequency feature spectrum after suppression to 0 to an acoustic signal are provided, noise reduction can be performed appropriately. That is, even if the noise to be suppressed is an unsteady sound, the frequency component of interest is surely suppressed to substantially zero, so that noise remains (in the case of under-subtraction) and the sound quality deteriorates unlike the spectral subtraction method. (In the case of excessive subtraction). Further, since no spectral subtraction is performed, it is not necessary to estimate a noise spectrum to be subtracted.

(2) The electronic camera 1 of (1) further includes a CPU 16 that determines whether or not an acoustic signal has been acquired during driving of the focusing lens, and the CPU 16 has a frequency characteristic corresponding to the affirmatively determined acoustic signal. Since the suppression process is performed on the spectrum, it is possible to appropriately reduce the noise of the acoustic signal due to the driving of the movable part.

(3) The electronic camera 1 of the above (2) further includes a CPU 16 that generates an environmental sound threshold spectrum corresponding to the environmental sound based on the frequency characteristic spectrum Fourier-transformed by the CPU 16. In this case, the CPU 16 sets a signal having a frequency component that is larger than the environmental sound threshold spectrum corresponding to the environmental sound, among the frequency characteristic spectrum obtained by Fourier transform, as a noise reduction target. Can be limited. For this reason, it is possible to suppress deterioration in sound quality as compared with the case where the signal of the frequency component targeted for noise reduction is always suppressed.

(4) In the electronic camera 1 of the above (3), the CPU 16 extracts the maximum value for each frequency component of the frequency feature spectrum of a plurality of frames each Fourier-transformed in different frames, and uses the extracted signal of each frequency component. Since the configured frequency feature spectrum is the environmental sound threshold spectrum corresponding to the environmental sound, it is possible to appropriately determine a criterion for determining whether or not the noise is within the allowable range.

(5) In the electronic camera 1, the CPU 16 determines the environment according to the environmental sound based on the frequency characteristic spectrum corresponding to the acoustic signal that is negatively determined in the determination as to whether or not the acoustic signal has been acquired during driving of the focusing lens. Since the sound threshold spectrum is generated, an environmental sound threshold spectrum corresponding to the environmental sound when the focusing lens is not driven is generated. For this reason, it is possible to appropriately determine a criterion for determining whether or not the noise is within an allowable range.

(6) The electronic camera 1 further includes a CPU 16 that determines whether the sum of the signal values of the frequency components in the frame of the frequency feature spectrum Fourier-transformed by the CPU 16 is greater than a second predetermined value. The CPU 16 generates an environmental sound threshold spectrum corresponding to the environmental sound based on the frequency feature spectrum determined that the sum of the signal values is smaller than the second predetermined value. For this reason, it is possible to appropriately determine a criterion for determining whether or not the noise is within an allowable range.

(7) The electronic camera 1 further includes a CPU 16 that determines whether sound is included in the acoustic signal. CPU16 produces | generates the environmental sound threshold spectrum according to environmental sound based on the frequency feature spectrum corresponding to the acoustic signal determined not to contain an audio | voice. For this reason, it is possible to appropriately determine a criterion for determining whether or not the noise is within an allowable range.

(8) In the electronic camera 1, since the CPU 16 performs the suppression process on the frequency feature spectrum corresponding to the acoustic signal determined not to include sound, the CPU 16 performs control so as not to suppress the sound information. be able to.

(9) The electronic camera 1 further includes a CPU 16 that determines whether or not a change in frequency feature spectrum between successive frames subjected to Fourier transform by the CPU 16 is greater than a third predetermined value. Since the CPU 16 performs the suppression process on the frequency feature spectrum for which it is determined that the change in the frequency feature spectrum between successive frames is smaller than the third predetermined value, the noise reduction process is appropriately performed. Can do.

(Modification 1)
Although the example in which the noise reduction processing program illustrated in FIG. 2 is started immediately before recording the acoustic data file in the storage medium 30 has been described, the acquisition of the acoustic data and the noise reduction processing may be performed in real time. . In this case, the CPU 16 can perform noise reduction processing for each frame and sequentially record the acoustic data after the noise reduction processing on the recording medium 30.

(Modification 2)
When the environmental sound threshold is determined and updated using the frequency feature spectrum of the target frame (step S80), it may be determined and updated as follows. For example, the CPU 16 of Modification 2 minimizes the sum of the frequency feature spectra of each frame from the first frame constituting the acoustic data to the target frame (that is, the sum of the amplitude values for each of the approximately 2000 frequency components). Select a frame. Then, the environmental sound threshold spectrum is updated by substituting the frequency characteristic spectrum obtained by multiplying the frequency characteristic spectrum composed of about 2000 frequency components for the frame by the fourth predetermined value with the previous environmental sound threshold spectrum.

  According to the modified example 2, since the environmental sound threshold spectrum is determined and updated based on the acoustic data of the frame that is highly likely not to include sudden sound, the environmental sound threshold spectrum has a large value at a specific frequency component. You can avoid having. For this reason, it is possible to reduce the possibility of noise reduction being insufficient for a specific frequency component.

(Modification 3)
When the environmental sound threshold is determined and updated using the frequency feature spectrum of the target frame (step S80), it may be determined and updated as follows. CPU16 of the modification 3 calculates the average value of an amplitude value for every said about 2000 frequency component, for example using the frequency characteristic spectrum of each flame | frame from the 1st frame which comprises acoustic data to an object frame. Then, the environmental sound threshold spectrum is updated by replacing the frequency characteristic spectrum obtained by multiplying the frequency characteristic spectrum formed by the average value of each frequency component by the fifth predetermined value with the previous environmental sound threshold spectrum.

  According to the third modification, the environmental sound threshold spectrum is determined and updated so that the sound component suddenly included in a certain frame is leveled. Therefore, the environmental sound threshold spectrum has a large value at a specific frequency component. You can avoid that. For this reason, it is possible to reduce the possibility of noise reduction being insufficient for a specific frequency component.

(Modification 4)
In the above-described determination of whether or not sound is included in the acoustic data (step S30), detection may be performed using the peak value of the autocorrelation function of the sound signal. The CPU 16 of the modification 4 uses the peak value of the autocorrelation function in a time delay region (about 880 points to about 110 points when the sampling frequency is 44.1 kHz) corresponding to one period of the fundamental frequency (for example, 50 Hz to 400 Hz) of the audio signal. Ask for. And CPU16 determines with the audio | voice being included when a peak value is larger than a predetermined determination threshold value. When the peak value is smaller than a predetermined determination threshold, it is determined that no sound is included.

(Modification 5)
Or you may detect using the envelope of a speech spectrum waveform (a horizontal axis is a signal frequency and a vertical axis | shaft is an amplitude of a signal). The CPU 16 of the modified example 5 obtains the envelope of the speech spectrum waveform by linear prediction analysis or the like. And CPU16 determines with the audio | voice being included, when the amount of time change of an envelope is larger than a predetermined determination threshold value. When the amount of time variation of the envelope is smaller than a predetermined determination threshold, it is determined that no sound is included.

(Modification 6)
In the embodiment described above, an example in which a signal processing device for noise reduction is mounted on the electronic camera 1 has been described. However, the signal processing device may be configured by a personal computer. A signal processing apparatus is configured by causing the computer apparatus 100 illustrated in FIG. 3 to execute a program that performs processing based on the flowchart illustrated in FIG. When the program is used by being loaded into the personal computer 100, the program is loaded into the data storage device of the personal computer 100 and then used as a signal processing device by executing the program.

  The loading of the program to the personal computer 100 may be performed by setting a storage medium 104 such as a CD-ROM storing the program in the personal computer 100, or to the personal computer 100 by a method via the communication line 101 such as a network. You may load. When passing through the communication line 101, the program is stored in the hard disk device 103 of the server (computer) 102 connected to the communication line 101. The program can be supplied as various forms of computer program products such as provision via the storage medium 104 or the communication line 101.

  According to the modified example 6 described above, the same function and effect as the above embodiment can be obtained. In addition to the configuration using a personal computer, a digital photo frame or a projector having a playback function may be configured to perform the above-described noise reduction processing.

  The above description is merely an example, and is not limited to the configuration of the above embodiment.

1 ... Electronic camera 14 ... RAM
15 ... LCD monitor 16 ... CPU
18 ... Card interface 20 ... Operation member 21 ... Microphone 22 ... Acoustic processing circuit 23 ... Lens drive mechanism 30 ... Storage medium 100 ... Personal computer

Claims (13)

Conversion means for converting an acoustic signal into a spectrum signal for each predetermined frame;
Spectrum suppression means for reducing the signal of the frequency component to be noise-reduced from the spectrum signal converted by the conversion means to substantially zero;
Inverse conversion means for inversely converting the spectrum signal after spectrum suppression into an acoustic signal;
A signal processing apparatus comprising: The signal processing device according to claim 1,
It further comprises an acquisition timing determination means for determining whether or not the acoustic signal is acquired at a noise generation timing,
The signal processing apparatus according to claim 1, wherein the spectrum suppression unit performs the spectrum suppression on a spectrum signal corresponding to the acoustic signal that has been affirmed by the acquisition timing determination unit. The signal processing device according to claim 2,
Based on the spectrum signal converted by the converting means, further comprising an environmental sound threshold value generating means for generating a spectral signal corresponding to the environmental sound,
The signal processing apparatus according to claim 1, wherein the spectrum suppression unit sets a signal having a frequency component that is larger than a spectrum signal corresponding to the environmental sound, among the spectrum signals converted by the conversion unit, as the noise reduction target. The signal processing device according to claim 3.
The environmental sound threshold generation unit extracts a maximum value for each frequency component of a plurality of frames of spectrum signals converted in different frames by the conversion unit, and extracts a spectrum signal composed of the extracted signals of each frequency component. A signal processing apparatus characterized in that it is a spectrum signal corresponding to the environmental sound. The signal processing device according to claim 3.
The environmental sound threshold generation unit multiplies a predetermined value by a spectral signal of a frame in which the sum of signal values of frequency components in the frame is minimum among the spectral signals of a plurality of frames converted in different frames by the conversion unit. A signal processing apparatus characterized in that the spectrum signal corresponds to the environmental sound. The signal processing device according to claim 3.
The environmental sound threshold generation unit is configured by a signal obtained by calculating an average value for each frequency component of a plurality of frames of spectrum signals converted in different frames by the conversion unit, and multiplying each average value by a predetermined value. A signal processing apparatus characterized in that a spectrum signal is a spectrum signal corresponding to the environmental sound. In the signal processing device according to any one of claims 3 to 6,
The environmental sound threshold generation unit generates a spectral signal corresponding to the environmental sound based on a spectral signal corresponding to the acoustic signal that is negatively determined by the acquisition timing determination unit. In the signal processing device according to any one of claims 3 to 7,
A sum total determining means for determining whether or not a sum of signal values of frequency components in a frame of the spectrum signal converted by the converting means is larger than a first determination threshold;
The signal processing apparatus according to claim 1, wherein the environmental sound threshold generation unit generates a spectral signal corresponding to the environmental sound based on the spectral signal that is negatively determined by the sum total determination unit. In the signal processing device according to any one of claims 3 to 8,
Voice determination means for determining whether or not sound is included in the acoustic signal;
The environmental sound threshold generation unit generates a spectral signal corresponding to the environmental sound based on a spectral signal corresponding to the acoustic signal that is negatively determined by the voice determination unit. The signal processing device according to claim 9,
The signal processing apparatus, wherein the spectrum suppression unit performs the spectrum suppression on a spectrum signal corresponding to an acoustic signal that is negatively determined by the voice determination unit. The signal processing device according to claim 10,
A change determination means for determining whether or not a change in the spectrum signal between successive frames converted by the conversion means is greater than a second determination threshold;
The signal processing apparatus, wherein the spectrum suppression unit performs the spectrum suppression on a spectrum signal that is negatively determined by the change determination unit.   A camera comprising the signal processing device according to claim 1. A first process for reading an acoustic signal;
A second process for converting the acoustic signal into a spectrum signal for each predetermined frame;
A third process for setting the frequency component signal to be reduced in noise from the converted spectrum signal to substantially zero;
A fourth process for inversely converting the spectral signal after the third process into an acoustic signal;
A signal processing program for causing a computer to execute.

Images