Signal processing device, signal processing method, and non-transitory recording medium

Abstract

The signal processing device of one embodiment of the present application includes a filter circuit that generates a second signal by performing filter processing on a first signal using a filter coefficient; a first Fourier transform circuit that generates a first frequency spectrum signal by performing a discrete Fourier transform on the first signal; a subtraction circuit that generates a fourth signal by performing subtraction processing on a third signal supplied from a sensor and corresponding to the first signal and the second signal; a second Fourier transform circuit that generates a second frequency spectrum signal by performing a discrete Fourier transform on the fourth signal; a frequency spectrum subtraction processing circuit that generates a third frequency spectrum signal by performing frequency spectrum subtraction processing on the second frequency spectrum signal using a noise frequency spectrum signal indicated by a noise model prepared in advance; and a control unit that calculates the filter coefficient based on the first frequency spectrum signal and the third frequency spectrum signal.

Description

Technical Field

[0001] This invention relates to a signal processing apparatus and method for reducing steady-state noise, as well as a non-transient recording medium containing software for reducing steady-state noise. Background Technology

[0002] In signal processing devices, there are apparatuses that process the output signal of a sensor. For example, Patent Document 1 discloses a technique for calculating an error signal by subtracting noise contained in the output signal of a sensor. [Existing Technical Documents] [Patent Literature]

[0003] Patent Document 1: Japanese Patent Application Publication No. 2020-10803 Summary of the Invention

[0004] In signal processing devices, it is desirable to reduce steady-state noise, and more effective reduction of steady-state noise is expected.

[0005] It is desirable to provide a signal processing device, signal processing method, and non-transient recording medium that can effectively reduce steady-state noise.

[0006] One embodiment of the signal processing apparatus of the present invention includes a filtering circuit, a first Fourier transform circuit, a subtraction circuit, a second Fourier transform circuit, a spectrum subtraction processing circuit, and a control unit. The filtering circuit generates a second signal by filtering a first signal using filtering coefficients. The first Fourier transform circuit generates a first spectral signal by performing a discrete Fourier transform on the first signal. The subtraction circuit generates a fourth signal by subtracting the second signal from a third signal supplied from a sensor and corresponding to the first signal. The second Fourier transform circuit generates a second spectral signal by performing a discrete Fourier transform on the fourth signal. The spectrum subtraction processing circuit generates a third spectral signal by performing spectrum subtraction on the second spectral signal using a noise spectrum signal shown in a pre-prepared noise model. The control unit calculates filtering coefficients based on the first and third spectral signals to reduce the signal components of the third spectral signal.

[0007] One embodiment of the signal processing method of the present invention includes: generating a second signal by filtering a first signal using filtering coefficients; generating a first spectral signal by performing a discrete Fourier transform on the first signal; generating a fourth signal by subtracting the second signal from a third signal supplied from a sensor and corresponding to the first signal; generating a second spectral signal by performing a discrete Fourier transform on the fourth signal; generating a third spectral signal by performing spectral subtraction on the second spectral signal using a noise spectral signal shown by a pre-prepared noise model; and calculating filtering coefficients based on the first spectral signal and the third spectral signal to reduce the signal components of the third spectral signal.

[0008] One embodiment of the present invention includes a non-transient recording medium recording software that enables a processor to perform processing. The processing includes: generating a second signal by filtering a first signal using filtering coefficients; generating a first spectral signal by performing a discrete Fourier transform on the first signal; generating a fourth signal by subtracting the second signal from a third signal supplied from a sensor and corresponding to the first signal; generating a second spectral signal by performing a discrete Fourier transform on the fourth signal; generating a third spectral signal by performing spectral subtraction on the second spectral signal using a noise spectral signal shown in a pre-prepared noise model; and calculating filtering coefficients based on the first and third spectral signals to reduce the signal components of the third spectral signal.

[0009] The signal processing apparatus, signal processing method, and non-transient recording medium according to one embodiment of the present invention can effectively reduce steady-state noise. Attached Figure Description

[0010] [ Figure 1 [A block diagram showing a structural example of a signal processing apparatus according to an embodiment of the present invention.] [ Figure 2 ]express Figure 1 The diagram illustrates an example of an operation of the Fourier transform unit. [ Figure 3 ]express Figure 1 The diagram illustrates an example of an operation of the spectrum subtraction processing unit. [ Figure 4 ]express Figure 1 A diagram illustrating one operating state of the signal processing device. [ Figure 5A ]express Figure 1 The diagram illustrates an example of an action of the noise model generation unit. [ Figure 5B ]express Figure 1 The diagram illustrates another example of the operation of the noise model generation unit. [ Figure 6 ]express Figure 1 A diagram illustrating another operating state of the signal processing device shown. [ Figure 7 [This is an explanatory diagram showing an example of spectral subtraction processing in the signal processing apparatus involved in the reference example.] [ Figure 8 ]express Figure 1 An explanatory diagram illustrating an example of spectral subtraction processing in the signal processing apparatus shown. Detailed Implementation

[0011] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0012] <Implementation Method> [Structure Example] Figure 1 This invention illustrates a structural example of a signal processing apparatus (signal processing apparatus 1) according to an embodiment of the present invention. The signal processing apparatus 1 is configured, for example, to detect the transmission characteristics of an audio signal from a system B containing a certain space A. The signal processing apparatus 1 includes: a signal generation unit 11, a digital-to-analog (DA) converter 12, a loudspeaker 13, a microphone 14, an analog-to-digital (AD) converter 15, a Fourier transform unit 21, an adaptive filter 22, a subtraction unit 23, a Fourier transform unit 24, a spectral subtraction processing unit 25, an adaptive algorithm calculation unit 26, an inverse Fourier transform unit 27, a noise model generation unit 31, a noise model selection unit 32, and a user interface 33. The signal processing apparatus 1 is configured, for example, to include a processor and a memory, and to perform processing by executing a program.

[0013] The signal generation unit 11 is configured to generate a signal S11 as voice data. The signal generation unit 11 sequentially generates data x at a predetermined processing frequency f (e.g., 16 kHz). This processing frequency f is a frequency sufficiently high than the frequency range of the audio signal processed by the signal processing device 1. Figure 1 The data x(n) shown represents the nth data x. In addition, the signal generation unit 11 outputs this series of data x as signal S11.

[0014] The DA conversion unit 12 is configured to generate an analog signal by performing DA conversion based on the signal S11 supplied from the signal generation unit 11.

[0015] The speaker 13 is configured to convert the analog signal supplied from the DA converter 12 into sound. The sound emitted from the speaker 13 propagates through space A and reaches the microphone 14.

[0016] The microphone 14 is configured to convert sound emitted from the speaker 13 and propagating through space A into an analog signal as an electrical signal. In this example, the microphone 14 is constructed using MEMS (Micro Electro Mechanical Systems) technology. In this example, the microphone 14 is integrally formed with the semiconductor chip of the signal processing device 1.

[0017] Because microphone 14 may generate steady-state noise, the output signal of microphone 14 contains not only the sound components propagating through space A, but also the steady-state noise component. In signal processing device 1, by reducing this steady-state noise, for example, the transmission characteristics of the acoustic signal in system B containing space A can be detected with higher accuracy, and the convergence speed of the adaptive algorithm operation described later can be accelerated.

[0018] The AD converter 15 is configured to generate a signal S15 by performing AD conversion based on the analog signal supplied from the microphone 14. The AD converter 15 generates data sequentially by performing AD conversion at a processing frequency f, and outputs this series of data as the signal S15.

[0019] The Fourier transform unit 21 is configured to generate a spectral signal SS21 by performing a discrete Fourier transform using a fast Fourier transform algorithm based on the signal S11 supplied from the signal generation unit 11.

[0020] Figure 2 This describes an example of the operation of the Fourier transform unit 21. For example, whenever a predetermined number (e.g., 512) of data x are supplied via signal S11, the Fourier transform unit 21 performs a discrete Fourier transform on these data x to generate spectral data X. The spectral data X includes amplitude spectral data and phase spectral data. Furthermore, in this… Figure 2 The amplitude spectrum data is shown, but the phase spectrum data is omitted. Figure 2 In the amplitude spectrum data shown, the horizontal axis represents frequency, and the vertical axis represents amplitude. During processing period T, the Fourier transform unit 21 receives a predetermined number (e.g., 512) of data x from the signal generation unit 11. During each processing period T, the Fourier transform unit 21 generates spectrum data X by performing discrete Fourier transforms on these data x. Figure 1 The spectrum data X(k) shown represents the k-th spectrum data X. In addition, the Fourier transform unit 21 outputs this series of spectrum data X as the spectrum signal SS21.

[0021] Adaptive filter 22 ( Figure 1The system is configured to filter the signal S11 supplied from the signal generation unit 11 using filter coefficients supplied from the inverse Fourier transform unit 27, thereby generating the signal S22. The adaptive filter 22 is an FIR (Finite Impulse Response) filter, which performs a convolution operation on the signal S11 using filter coefficients supplied from the inverse Fourier transform unit 27. Thus, the adaptive filter 22 adjusts the amplitude and phase of the signal S11 and outputs the signal with adjusted amplitude and phase as the signal S22. As will be described later, the filter coefficients of the adaptive filter 22 at the convergence of the negative feedback operation correspond to the transmission characteristics of the audio signal in the system B containing space A. System B includes a DA converter 12, a loudspeaker 13, space A, a microphone 14, and an AD converter 15.

[0022] The subtraction unit 23 is configured to perform subtraction processing on the signal S15 supplied from the AD converter 15 and the signal S22 supplied from the adaptive filter 22 to generate the signal S23. The subtraction unit 23 sequentially receives data contained in the signal S15 from the AD converter 15 and sequentially receives data contained in the signal S22 from the adaptive filter 22. The subtraction unit 23 generates data m sequentially by subtracting the data contained in the signal S22 from the data contained in the signal S15, and outputs this series of data m as the signal S23. Figure 1 The data m(n) shown represents the nth data m.

[0023] The signal S23 is the so-called error signal when the signal processing device 1 performs negative feedback. The data m(n) can be represented by the following formula EQ1. [Number 1] m(n) represents the error signal data. e(n) represents the error signal data assuming no noise is generated in microphone 14. no(n) represents the data corresponding to the noise generated in microphone 14. Thus, the data m(n) contains the data no(n) corresponding to the noise generated in microphone 14.

[0024] The Fourier transform unit 24 is configured to generate a spectral signal SS24 by performing a discrete Fourier transform using a fast Fourier transform algorithm based on the signal S23 supplied from the subtraction unit 23. The operation of the Fourier transform unit 24 is the same as that of the Fourier transform unit 21. That is, for example, whenever a predetermined number (e.g., 512) of data m are supplied via the signal S23, the Fourier transform unit 24 performs a discrete Fourier transform based on these data m, thereby generating spectral data M. The spectral data M includes amplitude spectral data and phase spectral data. Figure 1The spectrum data M(k) shown represents the k-th spectrum data M. Furthermore, the Fourier transform unit 21 outputs this series of spectrum data M as the spectrum signal SS24.

[0025] The spectrum data M(k) can be represented by the following formula EQ2. [Number 2] E1(k) is the spectral data of data e(n). That is, the spectral data E1(k) is the spectral data of the error signal assuming no noise is generated in microphone 14. NO(k) is the spectral data of data no(n). That is, the spectral data NO(k) is the spectral data of the noise generated in microphone 14.

[0026] The spectrum subtraction processing unit 25 is configured to perform spectrum subtraction processing on the spectrum signal SS24 supplied from the Fourier transform unit 24 and the noise spectrum signal SN32 related to the noise model supplied from the noise model selection unit 32, thereby generating the spectrum signal SS25. The noise spectrum signal SN32 includes amplitude spectrum data |NM|. Figure 1 The amplitude spectrum data |NM(k)| shown represents the k-th amplitude spectrum data |NM|. This amplitude spectrum data |NM| represents the amplitude spectrum data of the noise generated in the microphone 14. This amplitude spectrum data |NM| is generated by pre-measuring the characteristics of the microphone 14 and stored in the noise model selection unit 32.

[0027] Figure 3 This represents an example of a spectrum subtraction process, where (A) represents the amplitude spectrum data |M(k)| contained in the spectrum data M(k) supplied from the Fourier transform unit 24, (B) represents the amplitude spectrum data |NM(k)| supplied from the noise model selection unit 32, and (C) represents the amplitude spectrum data |E2(k)| contained in the spectrum data E2(k) as the result of the spectrum subtraction process. Figure 3 In the diagram, the horizontal axis represents frequency, and the vertical axis represents amplitude. Figure 3 In the diagram, the parts represented by dotted shading indicate the components of the error signal, and the parts represented by diagonal shading indicate the noise components of the microphone 14.

[0028] The spectrum subtraction processing unit 25 calculates the spectrum data M(k) supplied from the Fourier transform unit 24 based on the spectrum data M(k). Figure 3 (A)) and amplitude spectrum data |NM(k)| supplied from noise model selection unit 32 Figure 3 (B)), and performs spectrum subtraction processing. As a result, the spectrum subtraction processing unit 25 generates spectrum data E2(k) with reduced noise components. Figure 3 (C)). Furthermore, because the amplitude spectrum data |NM(k)| ( Figure 3 (B) is generated by pre-measuring the characteristics of microphone 14, so the noise component shown by the amplitude spectrum data |NM(k)| is the same as the spectrum data M(k) Figure 3 The noise components contained in (A) are similar but not identical. Therefore, as Figure 3 As shown in (C), the generated spectrum data E2 may contain components corresponding to the differences between these noise components, in addition to components of the error signal.

[0029] The spectral subtraction processing can utilize various calculation methods, such as spectral subtraction, MAP (maximum a posteriori) estimation, and MMSE-ETSA (minimum mean-square-error short-time spectral amplitude) estimation. The spectral subtraction processing unit 25 can, for example, select the calculation method to use based on instructions from the user interface 33. For example, when using spectral subtraction, the spectral data E2(k) can be represented by the following equations EQ3 and EQ4. [Number 3] [Number 4]

[0030] The spectral data M and the amplitude spectral data |NM| related to the noise model are sequentially supplied to the spectral subtraction processing unit 25. The spectral subtraction processing unit 25 generates spectral data E2 sequentially by performing spectral subtraction processing. In addition, the spectral subtraction processing unit 25 outputs this series of spectral data E2 as a spectral signal SS25.

[0031] The adaptive algorithm calculation unit 26 is configured to perform adaptive algorithm calculations based on the spectrum signal SS21 supplied from the Fourier transform unit 21 and the spectrum signal SS25 supplied from the spectrum subtraction processing unit 25. Specifically, the adaptive algorithm calculation unit 26 sets the filter coefficients of the adaptive filter 22 through the inverse Fourier transform unit 27 based on the spectrum signals SS21 and SS25, so as to reduce the spectrum data E2 contained in the spectrum signal SS25. That is, the signal processing device 1 performs negative feedback operation to reduce the spectrum data E2. The adaptive algorithm calculation unit 26 generates spectrum data of the filter coefficients sequentially by performing adaptive algorithm calculations. This spectrum data includes amplitude spectrum data and phase spectrum data. The adaptive algorithm calculation unit 26 outputs this series of spectrum data as the spectrum signal SS26.

[0032] The inverse Fourier transform unit 27 is configured to generate signal S27 by performing an inverse discrete Fourier transform on the spectral signal SS26 supplied from the adaptive algorithm operation unit 26. The spectral data of the filter coefficients are sequentially supplied to the inverse Fourier transform unit 27. The inverse Fourier transform unit 27 generates filter coefficients sequentially by performing an inverse discrete Fourier transform on the spectral data of the filter coefficients. Furthermore, the inverse Fourier transform unit 27 supplies these filter coefficients to the adaptive filter 22.

[0033] The noise model generation unit 31 is configured to generate a noise model relating to the noise generated in the microphone 14. The noise model generation unit 31 generates the noise model before the signal processing device 1 performs normal processing operations. This noise model is, for example, the amplitude spectrum data |NM| of the noise.

[0034] When the signal processing device 1 generates a noise model, it is, for example, placed in a quiet space where there is almost no surrounding sound. Furthermore, the signal processing device 1 is configured to operate the following six modules: microphone 14, AD conversion unit 15, subtraction unit 23, Fourier transform unit 24, noise model generation unit 31, and noise model selection unit 32. Under these conditions, the data e(n) related to the error signal and the spectral data E1(k) related to the error signal are zero. Therefore, the noise model generation unit 31 can generate amplitude spectral data |NM| of the noise generated in the microphone 14 based on the spectral signal SS24 containing spectral data NO.

[0035] The noise model selection unit 32 is configured to select a noise model according to instructions from the user interface 33. The noise model selection unit 32 stores amplitude spectrum data |NM| generated by the noise model generation unit 31. The noise model selection unit 32 can store multiple amplitude spectrum data |NM|. According to instructions from the user interface 33, the noise model selection unit 32 selects one of these multiple amplitude spectrum data |NM| and outputs the selected amplitude spectrum data |NM| as a noise spectrum signal SN32.

[0036] The user interface 33 is configured to receive operations from the user using the signal processing device 1. The user interface 33 may be configured using, for example, a liquid crystal display, various indicators, a touch panel, buttons, etc. In this example, the user interface 33 can receive user operations for selecting a noise model and for selecting a calculation method for spectral subtraction.

[0037] Here, the adaptive filter 22 corresponds to a specific example of a "filtering circuit" in one embodiment of this disclosure. The Fourier transform unit 21 corresponds to a specific example of a "first Fourier transform circuit" in one embodiment of this disclosure. The subtraction unit 23 corresponds to a specific example of a "subtraction circuit" in one embodiment of this disclosure. The Fourier transform unit 24 corresponds to a specific example of a "second Fourier transform circuit" in one embodiment of this disclosure. The spectrum subtraction processing unit 25 corresponds to a specific example of a "spectrum subtraction processing circuit" in one embodiment of this disclosure. The adaptive algorithm operation unit 26 and the inverse Fourier transform unit 27 correspond to a specific example of a "control unit" in one embodiment of this disclosure. The noise model generation unit 31 corresponds to a specific example of a "noise model generation circuit" in one embodiment of this disclosure. The noise model selection unit 32 corresponds to a specific example of a "noise model selection circuit" in one embodiment of this disclosure. The user interface 33 corresponds to a specific example of a "user interface" in one embodiment of this disclosure. The signal S11 corresponds to a specific example of a "first signal" in one embodiment of this disclosure. Signal S22 corresponds to a specific example of a "second signal" in one embodiment of this disclosure. Spectrum signal SS21 corresponds to a specific example of a "first spectrum signal" in one embodiment of this disclosure. Signal S15 corresponds to a specific example of a "third signal" in one embodiment of this disclosure. Signal S23 corresponds to a specific example of a "fourth signal" in one embodiment of this disclosure. Spectrum signal SS24 corresponds to a specific example of a "second spectrum signal" in one embodiment of this disclosure. Spectrum signal SS25 corresponds to a specific example of a "third spectrum signal" in one embodiment of this disclosure. Noise spectrum signal SN32 corresponds to a specific example of a "noise spectrum signal" in one embodiment of this disclosure.

[0038] [Actions and Functions] Next, the operation and function of the signal processing device 1 in this embodiment will be explained.

[0039] (Overall Action Summary) First, refer to Figure 1 The following is a summary of the overall operation of the signal processing device 1. First, the signal generation unit 11 generates a signal S11 as voice data. The DA converter 12 generates an analog signal by performing DA conversion based on the signal S11. The speaker 13 converts the analog signal supplied from the DA converter 12 into sound. The sound emitted from the speaker 13 propagates through space A and reaches the microphone 14. The microphone 14 converts the sound emitted from the speaker 13 and propagated through space A into an analog signal as an electrical signal. The AD converter 15 performs AD conversion based on the analog signal supplied from the microphone 14 to generate signal S15.

[0040] Fourier transform unit 21 generates a spectral signal SS21 by performing a discrete Fourier transform on signal S11. Adaptive filter 22 filters signal S11 using filter coefficients supplied from inverse Fourier transform unit 27 to generate signal S22. Subtraction unit 23 generates signal S23 by subtracting signal S15 supplied from AD converter 15 and signal S22 supplied from adaptive filter 22. Fourier transform unit 24 generates a spectral signal SS24 by performing a discrete Fourier transform on signal S23. Spectral subtraction processing unit 25 generates spectral signal SS25 by performing spectral subtraction on spectral signal SS24 supplied from Fourier transform unit 24 and noise spectrum signal SN32 related to the noise model supplied from noise model selection unit 32. Adaptive algorithm calculation unit 26 performs adaptive algorithm calculation on spectral signal SS21 supplied from Fourier transform unit 21 and spectral signal SS25 supplied from spectral subtraction processing unit 25. The adaptive algorithm calculation unit 26 generates a spectral signal SS26 by performing the adaptive algorithm calculation. The inverse Fourier transform unit 27 generates a signal S27 by performing an inverse discrete Fourier transform on the spectral signal SS26 supplied from the adaptive algorithm calculation unit 26. In addition, the inverse Fourier transform unit 27 supplies the filter coefficients contained in the signal S27 to the adaptive filter 22.

[0041] The noise model generation unit 31 generates a noise model (amplitude spectrum data |NM|) for the noise generated in the microphone 14. The noise model selection unit 32 stores multiple noise models and selects one of them according to instructions from the user interface 33. Furthermore, the noise model selection unit 32 supplies the selected noise model as a noise spectrum signal SN32 to the spectrum subtraction processing unit 25. The user interface 33 receives operations from the user using the signal processing device 1.

[0042] (Detailed actions) In its normal processing operation, signal processing device 1 detects the transmission characteristics of the acoustic signal in system B, which includes space A. Before performing this normal processing operation, signal processing device 1 first generates a noise model (amplitude spectrum data |NM|) for the noise generated in microphone 14.

[0043] Figure 4 This illustrates an example of the operational state of signal processing device 1 when generating a noise model. Figure 4 In the diagram, the modules depicted by solid lines are those that operate when the signal processing device 1 generates the noise model, while the modules depicted by dashed lines are those that do not operate when the signal processing device 1 generates the noise model.

[0044] When the signal processing device 1 generates a noise model, the signal processing device 1 is, for example, placed in a quiet space where there is almost no surrounding sound. In this example, the signal processing device 1 is configured to operate six modules: microphone 14, AD conversion unit 15, subtraction unit 23, Fourier transform unit 24, noise model generation unit 31, and noise model selection unit 32.

[0045] Microphone 14 converts sound into an electrical signal, i.e., an analog signal. Because the signal processing device 1 is located in a quiet space, most of the output signal of microphone 14 consists of noise components generated within microphone 14. The AD conversion unit 15 generates signal S15 by performing AD conversion based on the analog signal supplied from microphone 14. Because the adaptive filter 22 is not activated, the subtraction unit 23 outputs signal S15 as is, as signal S23. Under these conditions, since the data e(n) related to the error signal is zero, the data m(n) contained in signal S23 can be represented by the following equation EQ5. [Number 5] The data no(k) is the noise data generated in the microphone 14. The Fourier transform unit 24 generates a spectrum signal SS24 by performing a discrete Fourier transform on the signal S23 containing this series of data m. Under this condition, since the spectrum data E1(k) related to the error signal is zero, the spectrum data M(k) contained in the spectrum signal SS24 can be represented by the following formula EQ6. [Number 6] The spectrum data NO(k) is the spectrum data of the noise generated in the microphone 14. Thus, when the signal processing device 1 generates a noise model, signals S15, S23, and spectral signal SS24 are all signals related to the noise generated in the microphone 14, and are substantially equivalent to each other. The noise model generation unit 31 generates a noise model (amplitude spectrum data |NM|) about the noise generated in the microphone 14 based on a series of spectrum data NO.

[0046] Figure 5A , 5B This represents an example of an action of the noise model generation unit 31.

[0047] like Figure 5AAs shown, the noise model generation unit 31 can generate a noise model (amplitude spectrum data |NM|) by multiplying an amplitude spectrum data |NO| contained in one of a series of spectrum data NO by a subtraction coefficient α. Specifically, the noise model generation unit 31 can generate a noise model, for example, by multiplying the amplitude magnitude at each frequency in the amplitude spectrum data |NO| by, for example, "0.7". It is not limited to this; the noise model generation unit 31 can also generate a noise model, for example, by multiplying the amplitude magnitude at each frequency in the amplitude spectrum data |NO| by, for example, "1.2".

[0048] In addition, such as Figure 5B As shown, the noise model generation unit 31 can generate a noise model (amplitude spectrum data |NM|) by calculating the average value of the two amplitude spectrum data |NO| contained in two spectrum data in a series of spectrum data NO. Not limited to this, the noise model generation unit 31 can also generate a noise model, for example, by calculating the average value of the three amplitude spectrum data |NO| contained in three spectrum data in a series of spectrum data NO.

[0049] Alternatively, for example, it can also be Figure 5A Examples and Figure 5B Combining examples, the average of the two amplitude spectrum data is multiplied by a subtraction factor.

[0050] Thus, the noise model generation unit 31 generates multiple noise models (amplitude spectrum data |NM|) based on a series of spectrum data NO. Furthermore, the noise model generation unit 31 supplies these noise models to the noise model selection unit 32. The noise model selection unit 32 stores these noise models.

[0051] Therefore, the signal processing device 1 can use these noise models to detect the transmission characteristics of acoustic signals in system B containing space A.

[0052] Figure 6 This illustrates an example of the operating state of the signal processing device 1 when it detects transmission characteristics. Figure 6 In the diagram, the modules depicted by solid lines are those that operate when the signal processing device 1 detects transmission characteristics, while the modules depicted by dashed lines are those that do not operate when detecting transmission characteristics.

[0053] In this operation, the signal processing device 1 is positioned in space A where the transmission characteristics of the audio signal are to be detected. Furthermore, in this example, the signal processing device 1 is configured to operate modules other than the noise model generation unit 31.

[0054] The user selects a noise model and a spectral subtraction processing method by operating the user interface 33. The user interface 33 receives these user operations. The noise model selection unit 32 selects one noise model (amplitude spectrum data |NM|) from multiple noise models according to the instructions from the user interface 33, and outputs the selected amplitude spectrum data |NM| as a noise spectrum signal SN32. The spectral subtraction processing unit 25 selects the appropriate processing method according to the instructions from the user interface 33. Thus, the spectral subtraction processing unit 25 performs spectral subtraction processing using the selected processing method.

[0055] First, the signal generation unit 11 generates a signal S11 as voice data. The DA converter 12 generates an analog signal by performing DA conversion based on the signal S11. The speaker 13 converts the analog signal supplied from the DA converter 12 into sound. The sound emitted from the speaker 13 propagates through space A and reaches the microphone 14. The microphone 14 converts the sound emitted from the speaker 13 and propagated through space A into an electrical signal, i.e., an analog signal. The AD converter 15 generates signal S15 by performing AD conversion based on the analog signal supplied from the microphone 14.

[0056] Because microphone 14 may generate steady-state noise, the output signal of microphone 14 contains not only the sound component propagating through space A, but also the steady-state noise component. Therefore, signal S15 also contains not only the sound component propagating through space A, but also the steady-state noise component.

[0057] Fourier transform unit 21 generates a spectral signal SS21 by performing a discrete Fourier transform on signal S11. Adaptive filter 22 filters signal S11 using filter coefficients supplied from inverse Fourier transform unit 27, thereby generating signal S22. Subtraction unit 23 generates signal S23 by performing subtraction on signal S15 supplied from AD converter 15 and signal S22 supplied from adaptive filter 22.

[0058] The signal S23 is the so-called error signal when the signal processing device 1 performs negative feedback. The signal S23 contains a series of data m. The data m(n) is as shown in Equation EQ1, and includes: error signal data e(n) assuming no noise is generated in the microphone 14, and data no(n) corresponding to the noise generated in the microphone 14.

[0059] The Fourier transform unit 24 generates a spectral signal SS24 by performing a discrete Fourier transform on the signal S23 supplied from the subtraction unit 23. This spectral signal SS24 contains a series of spectral data M. The spectral data M(k), as shown in Equation EQ2, includes: spectral data E1(k) of the error signal assuming no noise is generated in the microphone 14, and spectral data NO(k) corresponding to the noise generated in the microphone 14.

[0060] Spectrum subtraction processing unit 25, such as Figure 3 As shown, based on the spectrum signal SS24 supplied from the Fourier transform unit 24 and the noise spectrum signal SN32 related to the noise model supplied from the noise model selection unit 32, spectrum subtraction is performed to generate the spectrum signal SS25. Figure 3 As shown, the spectrum subtraction processing unit 25 processes the spectrum data M(k) supplied from the Fourier transform unit 24. Figure 3 (A)) and amplitude spectrum data |NM(k)| supplied from noise model selection unit 32 Figure 3 (B)), generating spectral data E2(k) with reduced noise components ( Figure 3 (C)). In addition, the spectrum subtraction processing unit 25 generates a spectrum signal SS25 containing a series of spectrum data E2(k).

[0061] The adaptive algorithm calculation unit 26 performs adaptive algorithm calculations based on the spectrum signal SS21 supplied from the Fourier transform unit 21 and the spectrum signal SS25 supplied from the spectrum subtraction processing unit 25. By performing this adaptive algorithm calculation, the adaptive algorithm calculation unit 26 sets the filter coefficients of the adaptive filter 22 via the inverse Fourier transform unit 27, thereby reducing the spectrum data E2 contained in the spectrum signal SS25. In other words, the signal processing device 1 performs negative feedback to reduce the spectrum data E2. The adaptive algorithm calculation unit 26 sequentially generates spectrum data of the filter coefficients by performing this adaptive algorithm calculation. This spectrum data includes amplitude spectrum data and phase spectrum data. The amplitude spectrum data and phase spectrum data are updated sequentially. The changes in the amplitude spectrum data and phase spectrum data are limited to a predetermined step size. The adaptive algorithm calculation unit 26 outputs this series of spectrum data as the spectrum signal SS26.

[0062] Since the adaptive algorithm calculation unit 26 processes data in the frequency domain, it generates spectral data of filter coefficients by processing multiple frequencies individually. Therefore, the convergence speed may vary depending on the frequency. The adaptive algorithm calculation unit 26 can, for example, change the step size for each of the multiple frequencies based on their respective convergence speeds. Thus, the adaptive algorithm calculation unit 26 can, for example, improve the convergence speed at frequencies with slower convergence by changing the step size at those frequencies.

[0063] The inverse Fourier transform unit 27 performs an inverse discrete Fourier transform based on the spectral data of the filter coefficients contained in the spectral signal SS26 supplied from the adaptive algorithm operation unit 26, thereby generating filter coefficients. The adaptive filter 22 uses these filter coefficients for filtering.

[0064] Signal processing device 1 performs negative feedback to reduce the spectral data E2. The filter coefficients of adaptive filter 22 at the convergence of the negative feedback operation correspond to the transmission characteristics of the acoustic signal in system B containing space A. The user can understand the transmission characteristics of the acoustic signal in space A through these filter coefficients.

[0065] Thus, the signal processing device 1 includes: a filtering circuit (adaptive filter 22) that can generate a second signal (signal S22) by filtering the first signal (signal S11) using filtering coefficients; a first Fourier transform circuit (Fourier transform unit 21) that can generate a first spectral signal (spectral signal SS21) by performing a Fourier transform on the first signal (signal S11); a subtraction circuit (subtraction unit 23) that can generate a fourth signal (signal S23) by subtracting the second signal (signal S22) from a third signal (signal S15) supplied from the sensor (microphone 14) and corresponding to the first signal; and a second Fourier transform circuit (Fourier transform unit 23). 24) A second spectral signal (spectral signal SS24) can be generated by performing a Fourier transform on the fourth signal (signal S23); the spectral subtraction processing circuit (spectral subtraction processing unit 25) can generate a third spectral signal (spectral signal SS25) by performing spectral subtraction on the second spectral signal (spectral signal SS24) using a noise spectral signal (noise spectral signal SN32) shown in a pre-prepared noise model; and the control unit (adaptive algorithm operation unit 26 and inverse Fourier transform unit 27) can calculate the filtering coefficients based on the first spectral signal (spectral signal SS21) and the third spectral signal (spectral signal SS25) to reduce the signal components of the third spectral signal. Thus, in the signal processing device 1, because the adaptive algorithm operation is performed in the frequency domain and processed separately at each frequency, the convergence speed of the adaptive algorithm operation can be improved, for example. In addition, in the signal processing device 1, the computational load of the adaptive algorithm operation can be suppressed by using a fast Fourier transform. As a result, steady-state noise can be effectively reduced. In the signal processing device 1, by reducing steady-state noise in this way, the external disturbance components contained in the error signal can be suppressed. The result is that the convergence speed of the adaptive algorithm operation can be improved, and the operation accuracy of the adaptive algorithm operation can also be improved in the signal processing device 1.

[0066] Furthermore, in signal processing device 1, since noise components are reduced by performing spectral subtraction in the frequency domain, the computational load can be suppressed. That is, for example, when using an adaptive algorithm that operates in the time domain, performing spectral subtraction in the frequency domain results in... Figure 7 As shown, Fourier transform, spectral subtraction, and inverse Fourier transform are required, thus increasing the computational load. On the other hand, in signal processing device 1, because adaptive algorithm calculations are performed in the frequency domain, and spectral subtraction is performed in that frequency domain, therefore... Figure 8 As shown, no additional Fourier transform and its inverse transform are required. That is, in signal processing device 1, Fourier transform and its inverse transform are performed for adaptive algorithm operations in the frequency domain. Therefore, no additional Fourier transform and its inverse transform are needed for spectral subtraction in the frequency domain. Thus, signal processing device 1 can suppress computational load and effectively reduce steady-state noise.

[0067] Furthermore, in signal processing device 1, because noise components are reduced by performing spectral subtraction in the frequency domain, computational complexity can be suppressed, and noise over a wide frequency range can be reduced. In other words, for example, if a high-pass filter is provided in the circuit section operating in the time domain, this signal processing device can reduce noise in the low-frequency region, but not in the high-frequency region. Additionally, this signal processing device reduces the low-frequency signal components of the acoustic signal propagating through space A. On the other hand, in signal processing device 1, because noise components are reduced by performing spectral subtraction using a noise model in the frequency domain, noise over a wide frequency range can be reduced, and steady-state noise can be effectively reduced.

[0068] Furthermore, the signal processing apparatus 1 further includes a noise model generation circuit (noise model generation unit 31) that can generate a noise model based on the third signal (signal S15). As a result, since the noise model generation and normal processing operations can be performed using a single device in the signal processing apparatus 1, the structure can be simplified and the user's convenience can be improved.

[0069] Furthermore, in the signal processing device 1, the noise model generation circuit (noise model generation unit 31) can change the generation parameters ( Figure 5A The subtraction coefficients α and α shown are... Figure 5B The number of spectral data NO shown is used to generate multiple noise models based on the third signal (signal S15). Furthermore, although the noise model generation unit 31... Figure 4As shown, multiple noise models are generated based on the spectral signal SS24, but when generating the noise models, signals S15, S23, and the spectral signal SS24 are equivalent to each other. Therefore, in the signal processing device 1, for example, the optimal noise model for various situations corresponding to the application can be used, thus effectively reducing steady-state noise.

[0070] Furthermore, the signal processing apparatus 1 includes a user interface 33 that allows user operation, and a noise model selection circuit (noise model selection unit 32) that can select one noise model from multiple noise models based on user operation. Thus, the user can, for example, select the optimal noise model for the corresponding application and various situations. Therefore, steady-state noise can be effectively reduced in the signal processing apparatus 1.

[0071] Furthermore, in the signal processing apparatus 1, the spectrum subtraction processing circuit (spectrum subtraction processing unit 25) can perform spectrum subtraction processing using multiple spectrum subtraction methods, and can select one of the multiple spectrum subtraction methods as the spectrum subtraction method to be used. Therefore, in the signal processing apparatus 1, for example, since the optimal spectrum subtraction method corresponding to the application and various situations can be used, steady-state noise can be effectively reduced.

[0072] Furthermore, the signal processing apparatus 1 includes a user interface 33 that allows user operation, and the spectrum subtraction processing circuit (spectrum subtraction processing unit 25) can select one of multiple spectrum subtraction methods as the appropriate method based on user operation. Thus, the user can, for example, select the optimal spectrum subtraction method for the corresponding application and various situations. Therefore, steady-state noise can be effectively reduced in the signal processing apparatus 1.

[0073] [Effect] As described above, this embodiment includes: a filtering circuit that generates a second signal by filtering a first signal using filtering coefficients; a first Fourier transform circuit that generates a first spectral signal by performing a Fourier transform on the first signal; a subtraction circuit that generates a fourth signal by subtracting the second signal from a third signal supplied from a sensor and corresponding to the first signal; a second Fourier transform circuit that generates a second spectral signal by performing a Fourier transform on the fourth signal; a spectral subtraction processing circuit that generates a third spectral signal by performing spectral subtraction on the second spectral signal using a noise spectral signal shown in a pre-prepared noise model; and a control unit that calculates filtering coefficients based on the first and third spectral signals to reduce the signal components of the third spectral signal. This effectively reduces steady-state noise.

[0074] In addition, in this embodiment, the noise model generation circuit can generate multiple noise models based on the third signal by changing the generation parameters, thus effectively reducing steady-state noise.

[0075] Furthermore, this embodiment further includes a user interface that allows user operation, and the noise model selection circuit can select one from multiple noise models based on user operation, thus effectively reducing steady-state noise.

[0076] Furthermore, in this embodiment, the spectrum subtraction processing circuit can use multiple spectrum subtraction methods to perform spectrum subtraction processing, and can select one of the multiple spectrum subtraction methods as the spectrum subtraction method to be used, thus effectively reducing steady-state noise.

[0077] Furthermore, in this embodiment, a user interface that can accept user operation is provided, and the spectrum subtraction processing circuit can select one of multiple spectrum subtraction methods as the spectrum subtraction method to be used according to the user operation, thus effectively reducing steady-state noise.

[0078] [Variation Example 1] In the above embodiment, although the noise model selection unit 32 selects one of multiple noise models (amplitude spectrum data |NM|) based on user operation, it is not limited to this. Alternatively, the noise model selection unit 32 can also select one of multiple noise models by performing a calibration operation. Specifically, for example, in a calibration operation, the signal processing device 1 sequentially uses one of the multiple noise models (amplitude spectrum data |NM|) to detect the transmission characteristics of the acoustic signal in the system B containing space A. In this calibration operation, the noise model selection unit 32 selects one of the multiple noise models as the noise model to be used based on the convergence characteristics of the negative feedback operation. Specifically, the noise model selection unit 32 can, for example, select the noise model that maximizes the reduction value as the noise model to be used. Here, the reduction value is a value representing the amount of reduction in the error signal, based on the ratio of the power of the microphone's output signal to the power of the error signal. For example, the smaller the spectrum data E2(k), the larger the reduction value. In the signal processing device 1, by using a noise model that maximizes the reduction value in this way, the transmission characteristics can be obtained more accurately.

[0079] Furthermore, in this example, although the signal processing device 1 sequentially uses one of multiple noise models (amplitude spectrum data |NM|) to detect the transmission characteristics of the acoustic signal in system B containing space A during the calibration operation, it is not limited to this. For example, the signal processing device 1 may also select a subset of the multiple noise models (amplitude spectrum data |NM|), i.e., two or more noise models, corresponding to the application specified by the user operation, and sequentially use one of these two or more noise models to detect the transmission characteristics. In this case, the signal processing device 1 may, for example, select the noise model that maximizes the reduction value as the noise model used.

[0080] [Modification Example 2] In the above embodiment, although the spectrum subtraction processing unit 25 selects one of multiple calculation methods based on user operation, it is not limited to this. Alternatively, the spectrum subtraction processing unit 25 may select one of multiple calculation methods by performing a calibration operation. Specifically, for example, in the calibration operation, the signal processing device 1 sequentially uses one of multiple calculation methods to detect the transmission characteristics of the acoustic signal in the system B containing space A. In this calibration operation, the spectrum subtraction processing unit 25 selects one of the multiple calculation methods as the calculation method to be used based on the convergence characteristics of the negative feedback operation. Specifically, in this calibration operation, the spectrum subtraction processing unit 25 can select the calculation method that maximizes the reduction value as the calculation method to be used. In the signal processing device 1, by using the calculation method that maximizes the reduction value, the transmission characteristics can be obtained more accurately.

[0081] Furthermore, in this example, although the signal processing device 1 sequentially uses one of multiple calculation methods to detect the transmission characteristics of the audio signal in system B containing space A during the calibration operation, it is not limited to this. For example, the signal processing device 1 may also select a subset of multiple calculation methods, i.e., two or more calculation methods, corresponding to the application specified by the user operation, and sequentially use one of these two or more calculation methods to detect the transmission characteristics. In this case, the signal processing device 1 may, for example, select the calculation method that maximizes the reduction value as the calculation method used.

[0082] [Modification Example 3] In the above embodiment, although the noise model is used as amplitude spectrum data, it is not limited to this. As an alternative, it could also be data representing the power spectrum. In this case, the spectrum subtraction processing unit 25 could, for example, perform spectrum subtraction processing based on the data representing the power spectrum.

[0083] [Other variations] Alternatively, two or more of these variations can be combined.

[0084] The present invention has been described above with examples of embodiments and variations, but the present invention is not limited to these embodiments and various changes can be made.

[0085] For example, in the above embodiment, although a microphone constructed using MEMS technology is used as microphone 14, it is not limited to this. As an alternative, a condenser microphone may also be used as microphone 14.

[0086] For example, in the above embodiments, although this technology is used to detect the transmission characteristics of audio signals in a system B containing space A, it is not limited to this and can be used for various other purposes.

[0087] For example, in the embodiments described above, although a microphone 14 is used to reduce the steady-state noise generated in the microphone 14, the invention is not limited to this. Alternatively, a magnetic sensor can be used, for example, to reduce the steady-state noise generated in the magnetic sensor. In this case, the signal processing device 1 can be used, for example, in applications such as detecting biological information using a magnetic sensor. Furthermore, this technology can also be applied to devices that use various sensors.

[0088] For example, in the above embodiments, although this technology is used to reduce noise generated in the microphone 14, it is not limited to this. For example, it can also be used to reduce various steady noises such as dark noise and ambient noise.

[0089] The effects described in this specification are merely illustrative, and the effects of this disclosure are not limited to those described herein. Therefore, other effects can also be obtained with this disclosure.

[0090] Furthermore, this disclosure can be implemented in the following ways. (1) A signal processing device comprising: The filtering circuit can generate the second signal by filtering the first signal using filtering coefficients; The first Fourier transform circuit can generate a first spectral signal by performing a discrete Fourier transform on the first signal; The subtraction circuit can generate a fourth signal by subtracting the second signal from a third signal supplied from the sensor and corresponding to the first signal. The second Fourier transform circuit can generate a second spectral signal by performing a discrete Fourier transform on the fourth signal. The spectrum subtraction processing circuit can generate a third spectrum signal by performing spectrum subtraction on the second spectrum signal using a noise spectrum signal shown in a pre-prepared noise model; and The control unit can calculate the filtering coefficients based on the first spectrum signal and the third spectrum signal to reduce the signal components of the third spectrum signal. (2) The signal processing device described in (1), wherein, Furthermore, it has a noise model generation circuit. The noise model generation circuit can generate the noise model based on the third signal. (3) The signal processing device described in (2), wherein, Furthermore, it features a noise model selection circuit. The noise model generation circuit can generate multiple noise models based on the third signal by changing the generation parameters. The noise model selection circuit can select one of the multiple noise models as the noise model to be used. (4) The signal processing device described in (3), wherein, Further enhance the user interface, The user interface can accept user operations. The noise model selection circuit can select one of the multiple noise models based on the user's operation. (5) The signal processing device described in (3), wherein, The signal processing device has a calibration mode. In the calibration mode, the noise model selection circuit can select the noise model to be used based on the convergence characteristics when using the multiple noise models respectively. (6) The signal processing apparatus according to any one of (1) to (5), wherein, The spectrum subtraction processing circuit can perform the spectrum subtraction processing using multiple spectrum subtraction methods, and can select one of the multiple spectrum subtraction methods as the spectrum subtraction method to be used. (7) The signal processing apparatus described in (6), wherein, Further enhance the user interface, The user interface can accept user operations. The spectrum subtraction processing circuit can select one of the multiple spectrum subtraction methods as the spectrum subtraction method to be used based on the user operation. (8) The signal processing apparatus described in (6), wherein, The signal processing device has a calibration mode. In the calibration mode, the spectrum subtraction processing circuit can select the appropriate spectrum subtraction method based on the convergence characteristics of the multiple spectrum subtraction methods used respectively. (9) The signal processing apparatus according to any one of (1) to (8), wherein, Furthermore, it possesses the aforementioned sensor, The sensor is a microphone constructed using MEMS technology. (10) The signal processing apparatus according to any one of (1) to (9), wherein, The control unit can generate multiple coefficients corresponding to multiple frequencies, calculate the filter coefficients by performing an inverse discrete Fourier transform based on the multiple coefficients, and set the change amount of the multiple coefficients according to the convergence speed of the multiple coefficients. (11) A signal processing method, comprising: The second signal is generated by filtering the first signal using filter coefficients. A first spectral signal is generated by performing a discrete Fourier transform on the first signal. A fourth signal is generated by subtracting the second signal from a third signal supplied from the sensor and corresponding to the first signal. The second spectral signal is generated by performing a discrete Fourier transform on the fourth signal. A third spectral signal is generated by performing spectral subtraction on a noise spectral signal shown in a pre-prepared noise model based on the second spectral signal; and The filtering coefficients are calculated based on the first spectral signal and the third spectral signal to reduce the signal component of the third spectral signal. (12) A non-transient recording medium that records software that enables a processor to perform operations. The process includes: The second signal is generated by filtering the first signal using filter coefficients. A first spectral signal is generated by performing a discrete Fourier transform on the first signal. A fourth signal is generated by subtracting the second signal from a third signal supplied from the sensor and corresponding to the first signal. The second spectral signal is generated by performing a discrete Fourier transform on the fourth signal. A third spectral signal is generated by performing spectral subtraction on a noise spectral signal shown in a pre-prepared noise model based on the second spectral signal; and The filtering coefficients are calculated based on the first spectral signal and the third spectral signal to reduce the signal component of the third spectral signal.

Claims

1. A signal processing apparatus, comprising: The filtering circuit can generate the second signal by filtering the first signal using filtering coefficients; The first Fourier transform circuit can generate a first spectral signal by performing a discrete Fourier transform on the first signal; The subtraction circuit can generate a fourth signal by subtracting the second signal from a third signal supplied from the sensor and corresponding to the first signal. The second Fourier transform circuit can generate a second spectral signal by performing a discrete Fourier transform on the fourth signal. The spectrum subtraction processing circuit can generate a third spectrum signal by performing spectrum subtraction on the second spectrum signal using a noise spectrum signal shown in a pre-prepared noise model; and The control unit can calculate the filtering coefficients based on the first spectrum signal and the third spectrum signal to reduce the signal components of the third spectrum signal.

2. The signal processing apparatus according to claim 1, wherein, Furthermore, it includes a noise model generation circuit. The noise model generation circuit can generate the noise model based on the third signal.

3. The signal processing apparatus according to claim 2, wherein, Furthermore, it features a noise model selection circuit. The noise model generation circuit can generate multiple noise models based on the third signal by changing the generation parameters. The noise model selection circuit can select one of the multiple noise models as the noise model to be used.

4. The signal processing apparatus according to claim 3, wherein, Further enhance the user interface, The user interface can accept user operations. The noise model selection circuit can select one of the multiple noise models based on the user's operation.

5. The signal processing apparatus according to claim 3, wherein, The signal processing device has a calibration mode. In the calibration mode, the noise model selection circuit can select the noise model to be used based on the convergence characteristics when using the multiple noise models respectively.

6. The signal processing apparatus according to claim 1, wherein, The spectrum subtraction processing circuit can perform the spectrum subtraction processing using multiple spectrum subtraction methods, and can select one of the multiple spectrum subtraction methods as the spectrum subtraction method to be used.

7. The signal processing apparatus according to claim 6, wherein, Further enhance the user interface, The user interface can accept user operations. The spectrum subtraction processing circuit can select one of the multiple spectrum subtraction methods as the spectrum subtraction method to be used based on the user operation.

8. The signal processing apparatus according to claim 6, wherein, The signal processing device has a calibration mode. In the calibration mode, the spectrum subtraction processing circuit can select the appropriate spectrum subtraction method based on the convergence characteristics of the multiple spectrum subtraction methods used respectively.

9. The signal processing apparatus according to claim 1, wherein, Furthermore, it includes the aforementioned sensor. The sensor is a microphone constructed using MEMS technology.

10. The signal processing apparatus according to claim 1, wherein, The control unit can generate multiple coefficients corresponding to multiple frequencies, calculate the filter coefficients by performing an inverse discrete Fourier transform based on the multiple coefficients, and set the change amount of the multiple coefficients according to the convergence speed of the multiple coefficients.

11. A signal processing method, comprising: The second signal is generated by filtering the first signal using filter coefficients. A first spectral signal is generated by performing a discrete Fourier transform on the first signal. A fourth signal is generated by subtracting the second signal from a third signal supplied from the sensor and corresponding to the first signal. The second spectral signal is generated by performing a discrete Fourier transform on the fourth signal. A third spectral signal is generated by performing spectral subtraction on a noise spectral signal shown in a pre-prepared noise model based on the second spectral signal; and The filtering coefficients are calculated based on the first spectral signal and the third spectral signal to reduce the signal component of the third spectral signal.

12. A non-transient recording medium containing software that enables a processor to perform processing. The process includes: The second signal is generated by filtering the first signal using filter coefficients. A first spectral signal is generated by performing a discrete Fourier transform on the first signal. A fourth signal is generated by subtracting the second signal from a third signal supplied from the sensor and corresponding to the first signal. The second spectral signal is generated by performing a discrete Fourier transform on the fourth signal. A third spectral signal is generated by performing spectral subtraction on a noise spectral signal shown in a pre-prepared noise model based on the second spectral signal; and The filtering coefficients are calculated based on the first spectral signal and the third spectral signal to reduce the signal component of the third spectral signal.

Images