Active noise control system

Abstract

The present invention addresses the problem of providing an "active noise control system" capable of satisfactorily eliminating noise even in a high-frequency region in a blind configuration in which noise is not detected upstream. [Solution] An update control unit (16) causes an ICA processing unit (15) to perform independent component analysis processing when the level of sound output by a microphone increases. In the independent component analysis process, the ICA processing unit separates noise components from observation sound data output from the microphone by independent component analysis, and stores data representing the separated noise components as statistical noise data (14). An adaptive filter (13), which performs an adaptive operation using the output of the microphone as an error, outputs, from the speaker, a sound represented by noise cancelling sound data obtained by applying a filtering process to the stored statistical noise data as a noise cancelling sound (NC). In addition, the adaptive filter adapts the filter coefficient of the filter processing using the observation sound data output by the microphone as an error.

Description

Technical Field

[0001] The present invention relates to an active noise control (ANC) technology for reducing noise by using noise canceling sound for canceling noise. Background Art

[0002] As a technique for performing active noise control, feedforward type active noise control and feedback type active noise control are known (for example, Patent Document 1).

[0003] In active noise control, a sensor detects noise-related events such as vibration of a mechanism that is a noise source, rotation speed, or sound near the noise source at a position upstream of the noise cancellation target position, and generates a reference signal. Then, an adaptive filter is used to perform adaptation using the sound picked up by a microphone arranged near the noise cancellation target position as an error signal, and noise cancellation sound for canceling the noise is generated based on the reference signal and output from the speaker.

[0004] On the other hand, in the feedback type active noise control, the noise cancellation sound is also generated by the adaptive filter and output from the speaker. However, in the feedback type active noise control, the sound picked up by the microphone arranged near the position of the noise cancellation object is used as the error signal, and the signal obtained by adding the estimated noise cancellation sound propagated from the speaker to the microphone to the error signal is used as the noise signal representing the noise, and the noise cancellation sound for canceling the noise is generated according to the noise signal by the adaptive filter that uses the error signal for adaptation.

[0005] In addition, as technologies related to the present application, there are known technologies for separating a sound in which speech and noise are mixed into speech and noise through independent component analysis (ICA) (for example, Patent Document 2), or for separating individual audio sources from an audio signal in which multiple audio sources are mixed using a neural network (for example, Patent Document 3).

[0006] In addition, as a technology related to the present application, there is known a technology for obtaining the first principal component and the second principal component of a sound by principal component analysis using the intensity of each frequency band of the sound as data (variable) (for example, Patent Document 4).

[0007] Prior art literature:

[0008] Patent Literature:

[0009] Patent Document 1: Japanese Patent Application Publication No. 2017-090736

[0010] Patent Document 2: Japanese Patent Application No. 2006-510069

[0011] Patent Document 3: Japanese Patent Application No. 2009-511954

[0012] Patent Document 4: Japanese Patent Application Laid-Open No. 11-117875 Summary of the invention

[0013] Problems to be solved by the invention:

[0014] In the above-mentioned feedforward type active noise control, in addition to a microphone arranged near the position of the noise cancellation object for picking up errors used to adapt the adaptive filter, in order to generate a reference signal, a special sensor for detecting noise-related events is also required at a position upstream of the noise cancellation object position. If such a sensor cannot be arranged for various reasons, active noise control cannot be performed.

[0015] On the other hand, according to the feedback type active noise control, no sensor is required to detect noise-related events other than a microphone arranged near the noise cancellation target position, but there is a problem that it is difficult to cancel high-frequency noise due to the propagation delay of the feedback loop.

[0016] Furthermore, in feedforward active noise control, if a sensor that detects noise-related events is close to the noise cancellation target, there is insufficient computational processing time to generate high-frequency components of the required noise cancellation sound, making it difficult to effectively cancel noise in the high-frequency region.

[0017] Therefore, an object of the present invention is to realize active noise control capable of effectively canceling noise even in a high frequency range in a blind configuration in which no sensor for detecting an event related to noise is provided at a position upstream of a noise cancellation target position.

[0018] Means used to solve the problem:

[0019] In order to achieve the above-mentioned problem, the present invention relates to an active noise control system, which performs active noise control and comprises: a microphone, which outputs the picked-up sound as an observation signal; a speaker, which outputs a noise cancellation sound; an adaptive filter; a storage mechanism; a noise separation mechanism, which performs a noise separation process of separating a noise component from the observation signal at a specified time and storing a suspected noise signal representing the separated noise component in the storage mechanism; and an adaptive filter, which takes the suspected noise signal stored in the storage mechanism as input. Here, the adaptive filter performs an adaptive action with the observation signal as an error, and generates a noise cancellation sound output from the speaker according to the suspected noise signal.

[0020] In addition, in order to achieve the above-mentioned problem, the present invention relates to an active noise control system, which performs active noise control and comprises: a microphone; a frequency analysis mechanism, which converts the output of the microphone into a signal in the frequency domain and outputs it as an observation signal; a speaker, which outputs a noise cancellation sound; an adaptive filter; a storage mechanism; a time domain conversion mechanism; a noise separation mechanism, which performs a noise separation process of separating a noise component from the observation signal in the frequency domain and storing a pseudo noise signal representing the separated noise component in the frequency domain in the storage mechanism at a predetermined timing; and an adaptive filter, which takes the pseudo noise signal stored in the storage mechanism as input. Here, the adaptive filter performs adaptive action in the frequency domain with the observation signal as an error, and the time domain conversion mechanism converts the output of the adaptive filter into a signal in the time domain to generate a noise cancellation sound output from the speaker.

[0021] Here, the active noise control system can also be constructed as follows: in the frequency analysis mechanism, the output of the microphone is transformed into a frequency domain signal by fast Fourier transform and output as an observation signal; in the time domain transformation mechanism, the output of the adaptive filter is transformed into a time domain signal by inverse fast Fourier transform.

[0022] In addition, in order to achieve the above-mentioned problem, the present invention relates to an active noise control system, which performs active noise control and comprises: a microphone; an analysis mechanism, which transforms the output of the microphone into a weight coefficient of a prescribed transformation function and outputs it as an observation signal, wherein the prescribed transformation function is a function that generates a signal in the time domain by weighting an orthogonal basis; a speaker, which outputs a noise cancellation sound; an adaptive filter; a storage mechanism; a time domain transformation mechanism; a noise separation mechanism, which performs a noise separation process of separating a noise component from the observation signal at a prescribed timing and storing it in the storage mechanism as a suspected noise signal; and an adaptive filter, which takes the suspected noise signal stored in the storage mechanism as input. Here, the adaptive filter performs adaptive action with the observation signal as an error, and the time domain transformation mechanism transforms the output of the adaptive filter into a signal in the time domain through the transformation function, and generates a noise cancellation sound output from the speaker.

[0023] Here, in the active noise control system, the analysis mechanism may transform the output of the microphone into the weight coefficient by fast Fourier transform and output it as an observation signal, and the time domain transformation mechanism may transform the output of the adaptive filter into a time domain signal by inverse fast Fourier transform.

[0024] Alternatively, the analyzing means may transform the output of the microphone into the weight coefficient by wavelet transform and output it as an observation signal, and the time domain transform means may transform the output of the adaptive filter into a time domain signal by inverse wavelet transform.

[0025] Furthermore, the above active noise control system may be configured such that: in the noise separation mechanism, the noise separation process is performed at the predetermined timing when the output of the microphone becomes larger than a predetermined level.

[0026] Furthermore, the above active noise control system may be configured such that: in the noise separation means, the noise component is separated from the observation signal by independent component analysis (ICA).

[0027] Alternatively, the above active noise control system may be configured such that: in the noise separation mechanism, the noise component is inferred and separated from the observation signal by a neural network that has been previously learned in such a manner as to infer the noise component from the observation signal.

[0028] Alternatively, the above active noise control system may be configured such that: in the noise separation means, a principal component of a predetermined order of the observation signal is estimated by principal component analysis, and the estimated principal component is separated as the noise component.

[0029] According to the above active noise control system, the noise component in the output of the microphone that picks up the error used to make the adaptive filter adaptive can be separated and stored as a pseudo noise signal, and the noise cancellation sound is generated by the active noise control using the stored pseudo noise signal as the input of the adaptive filter. In addition, the problem of propagation delay of the feedback loop that causes the failure to cancel the noise well in the high frequency domain does not occur as in the feedback type active noise control. In addition, the pre-stored pseudo noise data is used as the input of the adaptive filter, so the problem of the failure to cancel the noise well in the high frequency domain in the feedforward type active noise control due to insufficient calculation processing time when the sensor that detects the event related to the noise at a position upstream of the noise cancellation target position approaches the noise cancellation target position does not occur.

[0030] Thus, according to the present invention, in a blind configuration in which no sensor for detecting events related to noise is provided upstream of a noise cancellation target position, active noise control capable of canceling noise well even in a high frequency range can be performed.

[0031] In addition, in order to achieve the above-mentioned problems, the present invention relates to an active noise control system, which performs active noise control and comprises: a microphone; an analysis mechanism, which transforms the output of the microphone into a weight coefficient of a prescribed transformation function and outputs it as an observation signal, wherein the prescribed transformation function is a function that generates a signal in the time domain by weighting an orthogonal basis; a speaker, which outputs a noise cancellation sound; an adaptive filter; a storage mechanism, which pre-stores the weight coefficient used to generate a pseudo noise through the transformation function as a pseudo noise signal; a time domain transformation mechanism; and an adaptive filter, which uses the pseudo noise signal stored in the storage mechanism as an input. The adaptive filter performs adaptive operation using the observation signal as an error, and the time domain transformation mechanism transforms the output of the adaptive filter into a signal in the time domain through the transformation function to generate a noise cancellation sound output from the speaker.

[0032] Here, in the active noise control system, the analysis mechanism may transform the output of the microphone into the weight coefficient by fast Fourier transform and output it as an observation signal, and the time domain transformation mechanism may transform the output of the adaptive filter into a time domain signal by inverse fast Fourier transform.

[0033] Alternatively, the analyzing means may transform the output of the microphone into the weight coefficient by wavelet transform and output it as an observation signal, and the time domain transform means may transform the output of the adaptive filter into a time domain signal by inverse wavelet transform.

[0034] According to such an active noise control system, active noise control capable of effectively canceling noise in a high frequency range can be performed even in a blind configuration in which no sensor for detecting events related to noise is provided upstream of a noise cancellation target position.

[0035] Effect of the invention:

[0036] As described above, according to the present invention, in a blind configuration in which no sensor for detecting events related to noise is provided upstream of a noise cancellation target position, active noise control capable of canceling noise well even in a high frequency range can be realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a block diagram showing the configuration of the noise control device according to the first embodiment of the present invention.

[0038] Figure 2 It is a block diagram showing the configuration of a noise control device according to a second embodiment of the present invention.

[0039] Figure 3It is a block diagram showing the configuration of a noise control device according to a third embodiment of the present invention.

[0040] Figure 4 It is a block diagram showing the configuration of a noise control device according to a fourth embodiment of the present invention.

[0041] Figure 5 It is a block diagram showing the configuration of a noise control device according to a fifth embodiment of the present invention.

[0042] Figure 6 It is a block diagram showing the structure of a noise control device according to a sixth embodiment of the present invention.

[0043] Figure 7 It is a block diagram showing the structure of a noise control device according to a seventh embodiment of the present invention.

[0044] Figure 8 It is a block diagram showing the structure of a noise control device according to an eighth embodiment of the present invention.

[0045] Fig. 9 This is a block diagram illustrating another configuration example of the noise control device according to the embodiment of the present invention.

[0046] Description of reference numerals:

[0047] 11…microphone, 12…speaker, 13…adaptive filter, 14…statistical noise data, 15…ICA processing unit, 16…update control unit, 21…microphone, 22…speaker, 23…adaptive filter, 24…frequency domain statistical noise data, 25…ICA processing unit, 26…update control unit, 27…frequency analysis unit, 28…time domain transformation unit, 31…microphone, 33…adaptive filter, 34…inference noise data, 35…learning neural network, 36…update control unit, 41…microphone, 42…speaker, 43…adaptive filter, 44…frequency domain inference noise data, 45…learning neural network, 46…update control unit, 47…frequency analysis unit, 48…time domain transformation unit, 51…microphone, 52…speaker, 53…adaptive filter, 54…principal component data, 55…principal component data Component analysis unit, 56…Update control unit, 61…Microphone, 62…Speaker, 63…Adaptive filter, 64…Frequency domain principal component data, 65…Principal component analysis unit, 66…Update control unit, 67…Frequency analysis unit, 68…Time domain transformation unit, 701…Microphone, 702…Speaker, 703…Adaptive filter, 704…Weight coefficient, 705…ICA processing unit, 706…Update control unit, 707…Frequency analysis unit, 708…Time domain transformation unit, 709…Base spectrum calculation unit, 710…Orthogonal base data, 801…Microphone, 802…Speaker, 803…Adaptive filter, 804…Weight coefficient, 805…ICA processing unit, 806…Update control unit, 807…Time-frequency analysis unit, 808…Time domain transformation unit, 809…Base spectrum calculation unit, 810…Orthogonal base data. DETAILED DESCRIPTION

[0048] Hereinafter, embodiments of the present invention will be described.

[0049] First, the first embodiment will be described.

[0050] Figure 1 The configuration of the noise control device according to the first embodiment is shown.

[0051] As shown in the figure, the noise control device includes: a microphone 11 arranged near the noise elimination target position, a speaker 12 that outputs a noise elimination sound NC for eliminating noise N at the noise elimination target position, an adaptive filter 13, statistical noise data 14, an ICA processing unit 15 that performs independent component analysis (ICA), and an update control unit 16.

[0052] The microphone 11 outputs a mixed sound of the collected noise N and the noise canceling sound NC as observation sound data. Here, the mixed sound data represents the remaining components of the noise N that cannot be completely canceled by the noise canceling sound NC when the noise canceling sound NC is output from the speaker 12.

[0053] When the noise control device is installed or the level of the sound represented by the observation sound data output by the microphone 11 increases to a predetermined level or more, the update control unit 16 stops the operation of the adaptive filter 13 and causes the ICA processing unit 15 to perform independent component analysis processing.

[0054] Here, while the operation of the adaptive filter 13 is stopped, the adaptive operation of the adaptive filter 13 and the output of the noise canceled sound NC are stopped. Therefore, while the operation of the adaptive filter 13 is stopped, the observed sound data output by the microphone 11 represents the noise N.

[0055] In the independent component analysis process, the ICA processing unit 15 separates noise components from the observation sound data output from the microphone 11 based on statistical independence by independent component analysis, and stores data representing the separated noise components as statistical noise data 14 .

[0056] Then, when the statistical noise data 14 is stored, the update control unit 16 restarts the operation of the adaptive filter 13 .

[0057] The stored statistical noise data 14 is output to the adaptive filter 13 as pseudo noise data.

[0058] The adaptive filter 13 outputs the sound represented by the noise-canceled sound data obtained by filtering the inputted pseudo-noise data as the noise-canceled sound NC from the speaker 12. In addition, the adaptive filter 13 uses the observed sound data outputted by the microphone 11 as an error, and adapts the filter coefficient of the filtering process applied to the pseudo-noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0059] Hereinafter, a second embodiment of the present invention will be described.

[0060] In the second embodiment, signal processing equivalent to the operation of the noise control device in the first embodiment is performed in the frequency domain.

[0061] Figure 2 The configuration of a noise control device according to the second embodiment is shown.

[0062] As shown in the figure, the noise control device includes: a microphone 21 arranged near the position of the noise elimination object, a speaker 22 that outputs a noise elimination sound NC for eliminating noise N at the position of the noise elimination object, an adaptive filter 23, frequency domain statistical noise data 24, an ICA processing unit 25 that performs independent component analysis (ICA), an update control unit 26, a frequency analysis unit 27, and a time domain transformation unit 28.

[0063] The frequency analysis unit 27 performs frequency analysis on the observation sound data output from the microphone 21 , and converts the observation sound data into frequency-domain observation sound data expressing the intensity of the sound in each frequency band.

[0064] As the frequency analysis performed by the frequency analysis unit 27 , for example, FFT (Fast Fourier Transform) can be used. When FFT is used, the frequency domain observation sound data becomes a frequency spectrum.

[0065] When the noise control device is set, or when it is detected based on the frequency domain observation sound data that the level of the sound represented by the observation sound data output by the microphone 21 increases to above a specified level, the update control unit 26 stops the operation of the adaptive filter 23 and causes the ICA processing unit 25 to perform independent component analysis processing.

[0066] The ICA processing unit 25 performs independent component analysis in the frequency domain during independent component analysis processing, separates data representing noise components in the frequency domain from the frequency domain observed sound data based on statistical independence, and stores the separated data as frequency domain statistical noise data 24 .

[0067] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 27 , the frequency domain statistical noise data 24 also becomes a frequency spectrum.

[0068] When the statistical noise data 14 is stored, the update control unit 26 restarts the operation of the adaptive filter 23 .

[0069] The stored frequency domain statistical noise data 24 is output to the adaptive filter 23 as frequency domain pseudo noise data.

[0070] The adaptive filter 23 applies filtering processing to the input frequency domain suspected noise data in the frequency domain, and outputs it as frequency domain noise-eliminated sound data to the time domain conversion unit 28. In addition, the adaptive filter 23 uses the frequency domain observed sound data output by the frequency analysis unit 27 as an error, and adapts the filter coefficient of the filtering processing applied to the frequency domain suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0071] The time domain conversion unit 28 converts the frequency domain noise cancellation sound data outputted from the adaptive filter 23 into a time domain sound, and outputs the sound from the speaker 22 as the noise cancellation sound NC.

[0072] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 27, the transformation performed by the time domain transformation unit 28 becomes IFFT (Inverse Fast Fourier Transform).

[0073] Hereinafter, a third embodiment of the present invention will be described.

[0074] The third embodiment uses a neural network instead of independent component analysis (ICA) to perform the process of separating noise from observed sound data by the noise control device of the first embodiment.

[0075] Figure 3 The configuration of a noise control device according to the third embodiment is shown.

[0076] As shown in the figure, the noise control device includes: a microphone 31 arranged near the noise elimination target position, a speaker 32 that outputs a noise elimination sound NC for eliminating noise N at the noise elimination target position, an adaptive filter 33, inferred noise data 34, a learned neural network 35, and an update control unit 36.

[0077] The microphone 31 outputs a mixed sound of the picked-up noise N and the noise-canceling sound NC as observation sound data.

[0078] When the noise control device is installed or the sound level represented by the observation sound data output by the microphone 31 increases to a predetermined level or above, the update control unit 36 ​​stops the operation of the adaptive filter 33 and causes the learned neural network 35 to perform noise inference processing.

[0079] The learned neural network 35 is a neural network that is pre-learned using appropriate training data to perform inference on separating noise components from observed sound data. In the noise inference process, noise is inferred based on the observed sound data output by the microphone 31, and data representing the inferred noise is stored as inferred noise data 34.

[0080] When the inferred noise data 34 is stored, the update control unit 36 ​​restarts the operation of the adaptive filter 33 .

[0081] The stored inferred noise data 34 is output to the adaptive filter 33 as pseudo noise data.

[0082] The adaptive filter 33 outputs the sound represented by the noise-canceling sound data obtained by filtering the inputted suspected noise data as the noise-canceling sound NC from the speaker. In addition, the adaptive filter 33 uses the observed sound data outputted by the microphone 31 as an error, and adapts the filter coefficient of the filtering process applied to the suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0083] Hereinafter, a fourth embodiment of the present invention will be described.

[0084] In the fourth embodiment, signal processing equivalent to the operation of the noise control device in the third embodiment is performed in the frequency domain.

[0085] Figure 4 The configuration of a noise control device according to the fourth embodiment is shown.

[0086] As shown in the figure, the noise control device includes: a microphone 41 arranged near the position of the noise elimination object, a speaker 42 that outputs a noise elimination sound NC for eliminating noise N at the position of the noise elimination object, an adaptive filter 43, frequency domain inference noise data 44, a learned neural network 45, an update control unit 46, a frequency analysis unit 47, and a time domain transformation unit 48.

[0087] The frequency analysis unit 47 performs frequency analysis on the observation sound data output from the microphone 41 , and converts the observation sound data into frequency-domain observation sound data expressing the intensity of the sound in each frequency band.

[0088] As the frequency analysis performed by the frequency analysis unit 47 , for example, FFT (Fast Fourier Transform) can be used. When FFT is used, the frequency domain observation sound data becomes a frequency spectrum.

[0089] When the noise control device is set up, or when it is detected based on the frequency domain observation sound data that the level of the sound represented by the observation sound data output by the microphone 41 increases to above a specified level, the update control unit 46 stops the operation of the adaptive filter 43 and causes the learned neural network 45 to perform noise inference processing.

[0090] The learned neural network 45 is a neural network that is pre-learned using appropriate training data to perform inference in the frequency domain to separate noise components from frequency domain observed sound data. In the noise inference processing, the noise components are inferred in the frequency domain based on the frequency domain observed sound data, and the inferred data representing the noise components in the frequency domain is stored as frequency domain inference noise data 44.

[0091] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 47 , the frequency domain inference noise data 44 also becomes a frequency spectrum.

[0092] When the frequency domain inference noise data 44 is stored, the update control unit 46 restarts the operation of the adaptive filter 43 .

[0093] The stored frequency domain inference noise data 44 is output to the adaptive filter 43 as frequency domain pseudo noise data.

[0094] The adaptive filter 43 applies filtering processing to the input frequency domain suspected noise data in the frequency domain, and outputs it as frequency domain noise-eliminated sound data to the time domain conversion unit 48. In addition, the adaptive filter 43 uses the frequency domain observed sound data output by the frequency analysis unit 47 as an error, and adapts the filter coefficient of the filtering processing applied to the frequency domain suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0095] The time domain conversion unit 48 converts the frequency domain noise cancellation sound data outputted from the adaptive filter 43 into a time domain sound, and outputs the sound from the speaker 42 as the noise cancellation sound NC.

[0096] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 47, the transformation performed by the time domain transformation unit 48 becomes IFFT (Inverse Fast Fourier Transform).

[0097] Hereinafter, a fifth embodiment of the present invention will be described.

[0098] The fifth embodiment uses principal component analysis instead of independent component analysis (ICA) to perform the process of separating noise from observed sound data by the noise control device of the first embodiment.

[0099] Figure 5 The configuration of a noise control device according to the fifth embodiment is shown.

[0100] As shown in the figure, the noise control device includes: a microphone 51 arranged near the noise cancellation target position, a speaker 52 that outputs a noise cancellation sound NC for canceling noise N at the noise cancellation target position, an adaptive filter 53, principal component data 54, a principal component analysis unit 55, and an update control unit 56.

[0101] The microphone 51 outputs a mixed sound of the picked-up noise N and the noise-canceling sound NC as observation sound data.

[0102] When the noise control device is installed or the level of the sound represented by the observation sound data output by the microphone 51 increases to a predetermined level or more, the update control unit 56 stops the operation of the adaptive filter 53 and causes the principal component analysis unit 55 to perform principal component analysis.

[0103] The principal component analysis unit 55 analyzes the principal components of the observed sound data a specified number of times (for example, only the first principal component, or only the second principal component, or the first principal component and the second principal component) through principal component analysis, and stores the data of the sound when the cumulative contribution rate of each analyzed principal component becomes the maximum as principal component data 54.

[0104] When the main component data 54 is stored, the update control unit 56 restarts the operation of the adaptive filter 53 .

[0105] The stored main component data 54 is output to the adaptive filter 53 as pseudo noise data.

[0106] The adaptive filter 53 outputs the sound represented by the noise-canceled sound data obtained by filtering the inputted suspected noise data as the noise-canceled sound NC from the speaker 52. In addition, the adaptive filter 53 uses the observed sound data outputted by the microphone 51 as an error, and adapts the filter coefficient of the filtering process applied to the suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0107] Next, a sixth embodiment of the present invention will be described.

[0108] In the sixth embodiment, signal processing equivalent to the operation of the noise control device in the fifth embodiment is performed in the frequency domain.

[0109] Figure 6 The structure of the noise control device according to the sixth embodiment is shown.

[0110] As shown in the figure, the noise control device includes: a microphone 61 arranged near the position of the noise elimination object, a speaker 62 that outputs a noise elimination sound NC for eliminating noise N at the position of the noise elimination object, an adaptive filter 63, frequency domain principal component data 64, a principal component analysis unit 65, an update control unit 66, a frequency analysis unit 67, and a time domain transformation unit 68.

[0111] The frequency analysis unit 67 performs frequency analysis on the observation sound data output from the microphone 61 , and converts the observation sound into frequency-domain observation sound data expressing the intensity of the sound in each frequency band.

[0112] As the frequency analysis performed by the frequency analysis unit 67 , for example, FFT (Fast Fourier Transform) can be used. When FFT is used, the frequency domain observation sound data becomes a frequency spectrum.

[0113] When the noise control device is set up, or when it is detected based on the frequency domain observation sound data that the level of the sound represented by the observation sound data output by the microphone 61 increases to above a specified level, the update control unit 66 stops the operation of the adaptive filter 63 and causes the principal component analysis unit 65 to perform principal component analysis.

[0114] The principal component analysis unit 65 analyzes the principal components of a specified number of frequency domain observation sound data (for example, only the first principal component, or only the second principal component, or the first principal component and the second principal component) through principal component analysis in the frequency domain, and stores the data of the sound when the cumulative contribution rate of each principal component analyzed in the frequency domain becomes the maximum as frequency domain principal component data 64.

[0115] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 67, the frequency domain main component data 64 also becomes a frequency spectrum.

[0116] When the frequency domain main component data 64 is stored, the update control unit 66 restarts the operation of the adaptive filter 63 .

[0117] The stored frequency domain main component data 64 is output to the adaptive filter 63 as frequency domain pseudo noise data.

[0118] The adaptive filter 63 applies filtering processing to the input frequency domain suspected noise data in the frequency domain, and outputs it as frequency domain noise-eliminated sound data to the time domain conversion unit 68. In addition, the adaptive filter 63 uses the frequency domain observed sound data output by the frequency analysis unit 67 as an error, and adapts the filter coefficient of the filtering processing applied to the frequency domain suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0119] The time domain conversion unit 68 converts the frequency domain noise cancellation sound data outputted from the adaptive filter 63 into a time domain sound, and outputs the sound from the speaker 62 as the noise cancellation sound NC.

[0120] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 67, the transformation performed by the time domain transformation unit 68 becomes IFFT (Inverse Fast Fourier Transform).

[0121] Next, a seventh embodiment of the present invention will be described.

[0122] Figure 7 The structure of the noise control device according to the seventh embodiment is shown.

[0123] As shown in the figure, the noise control device includes: a microphone 701 arranged near the position of the noise elimination object, a speaker 702 that outputs a noise elimination sound NC for eliminating noise N at the position of the noise elimination object, an adaptive filter 703, a weight coefficient 704, an ICA processing unit 705 that performs independent component analysis (ICA), an update control unit 706, a frequency analysis unit 707, a time domain transformation unit 708, a basis spectrum calculation unit 709, and orthogonal basis data 710.

[0124] The orthogonal basis data 710 is the data of the orthogonal basis used in the transformation function, which is a function that generates a signal in the time domain by weighting a predetermined orthogonal basis with a weight. As such a transformation function, an inverse fast Fourier transform or the like can be used. When the inverse fast Fourier transform is used, the orthogonal basis data 710 becomes Fourier basis data.

[0125] The frequency analysis unit 707 performs frequency analysis on the observation sound data output from the microphone 701 , and transforms the observation sound data into frequency-domain observation sound data expressing weight coefficients of an orthogonal basis of the above-mentioned transformation function.

[0126] Here, when the orthogonal basis data 710 is set as Fourier basis data, FFT (Fast Fourier Transform) is used as the frequency analysis performed by the frequency analysis unit 707, and the frequency domain observation sound data becomes a frequency spectrum.

[0127] When the noise control device is set up, or when it is detected based on the frequency domain observation sound data that the level of the sound represented by the observation sound data output by the microphone 701 increases to above a specified level, the update control unit 706 stops the operation of the adaptive filter 703 and causes the ICA processing unit 705 to perform independent component analysis processing.

[0128] The ICA processing unit 705 performs independent component analysis in the frequency domain during independent component analysis processing, separates data representing noise components in the frequency domain from the frequency domain observed sound data based on statistical independence, and sends the separated data as frequency domain statistical noise data to the base spectrum calculation unit 709.

[0129] The basis spectrum calculation unit 709 estimates the complex spectrum of each element of the orthogonal basis stored in the orthogonal basis data 710 of the noise represented in the frequency domain by the frequency domain statistical noise data sent from the ICA processing unit 705, and sets the weight coefficient of the orthogonal basis obtained based on the estimated complex spectrum as the weight coefficient 704.

[0130] The complex spectrum estimated by the basis spectrum calculation unit 709 expresses gain and phase, and the weight coefficients stored in the weight coefficients 704 express the gain and phase applied to each element of the orthogonal basis in a complex form.

[0131] In addition, when FFT is used as the frequency analysis performed by the frequency analysis unit 707, the weight coefficients stored in the weight coefficient 704 become the frequency spectrum.

[0132] When the weight coefficient 704 is stored, the update control unit 706 restarts the operation of the adaptive filter 703 .

[0133] The stored weight coefficient 704 is output to the adaptive filter 703 as frequency-domain pseudo-noise data representing the weight coefficient of the orthogonal basis stored in the orthogonal basis data 710 .

[0134] The adaptive filter 703 applies filtering processing to the input frequency domain suspected noise data in the frequency domain, and outputs it as frequency domain noise-eliminated sound data to the time domain transform unit 708. In addition, the adaptive filter 703 uses the frequency domain observed sound data output by the frequency analysis unit 707 as an error, and adapts the filter coefficient of the filtering processing applied to the frequency domain suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0135] The time domain conversion unit 708 converts the frequency domain noise cancellation sound data outputted from the adaptive filter 703 into a time domain sound, and outputs the sound from the speaker 702 as the noise cancellation sound NC.

[0136] Here, when FFT is used as the frequency analysis performed by the frequency analysis unit 707, the transformation performed by the time domain transformation unit 708 becomes IFFT (Inverse Fast Fourier Transform).

[0137] In addition, in the above-mentioned seventh embodiment, a processing unit can be set instead of the ICA processing unit 15, which completes the neural network or principal component analysis by learning, separates the data representing noise in the frequency domain from the frequency domain observation sound data, and uses the separated data as frequency domain statistical noise data.

[0138] Next, an eighth embodiment of the present invention will be described.

[0139] Figure 8 The following shows the structure of the noise control device according to the eighth embodiment.

[0140] As shown in the figure, the noise control device includes: a microphone 801 arranged near the position of the noise elimination object, a speaker 802 that outputs a noise elimination sound NC for eliminating noise N at the position of the noise elimination object, an adaptive filter 803, a weight coefficient 804, an ICA processing unit 805 that performs independent component analysis (ICA), an update control unit 806, a time-frequency analysis unit 807, a time domain transformation unit 808, a basis spectrum calculation unit 809, and orthogonal basis data 810.

[0141] The orthogonal basis data 810 is the data of the orthogonal basis used in the transformation function, which is a function that generates a time domain signal whose spectrum changes in time by weighting a predetermined orthogonal basis with a weight. As such a transformation function, an inverse wavelet transform or the like can be used. When the inverse wavelet transform is used, the orthogonal basis data 810 becomes the data of the wavelet basis. Here, the wavelet basis is generally a basis obtained by making various changes to the scale or time shift of the mother wavelet.

[0142] The time-frequency analysis unit 807 performs time-frequency analysis on the observation sound data outputted from the microphone 801 , and transforms the observation sound data into time-frequency domain observation sound data expressing weight coefficients of an orthogonal basis of the above-mentioned transformation function.

[0143] Here, when the orthogonal basis data 810 is set as wavelet basis data, wavelet transform is used as the time-frequency analysis performed by the frequency analysis unit, and the time-frequency domain observation sound data becomes wavelet coefficients.

[0144] When the noise control device is set up, or when it is detected based on the time-frequency domain observation sound data that the level of the sound represented by the observation sound data output by the microphone 801 increases to above a specified level, the update control unit 806 stops the operation of the adaptive filter 803 and causes the ICA processing unit 805 to perform independent component analysis processing.

[0145] The ICA processing unit 805 performs independent component analysis in the time-frequency domain during the independent component analysis processing, separates the data representing the noise component in the time-frequency domain from the time-frequency domain observed sound data based on statistical independence, and sends the separated data as time-frequency domain statistical noise data to the base spectrum calculation unit 809.

[0146] The basis spectrum calculation unit 809 estimates the spectrum of each element of the orthogonal basis stored in the orthogonal basis data 810 of the noise represented in the time-frequency domain by the time-frequency domain statistical noise data sent from the ICA processing unit 805, and determines the weight coefficient of the orthogonal basis obtained based on the estimated spectrum as the weight coefficient 804.

[0147] In addition, when wavelet transform is used as the frequency analysis performed by the frequency analysis unit, the weight coefficient stored in the weight coefficient 804 is a wavelet coefficient.

[0148] When the weight coefficient 804 is stored, the update control unit 806 restarts the operation of the adaptive filter 803 .

[0149] The stored weight coefficient 804 is output to the adaptive filter 803 as time-frequency domain pseudo-noise data representing the weight coefficient of the orthogonal basis stored in the orthogonal basis data 810 .

[0150] The adaptive filter 803 applies filtering processing to the input time-frequency domain suspected noise data in the frequency domain, and outputs it as time-frequency domain noise-eliminated sound data to the time domain conversion unit 808. In addition, the adaptive filter 803 uses the time-frequency domain observed sound data output by the time-frequency analysis unit 807 as an error, and adapts the filter coefficient of the filtering processing applied to the time-frequency domain suspected noise data by an adaptive algorithm such as LMS or NLMS so as to minimize the power of the error.

[0151] The time domain conversion unit 808 converts the time-frequency domain noise-cancelled sound data outputted from the adaptive filter 803 into a sound in the time domain, and outputs the sound from the speaker 802 as noise-cancelled sound NC.

[0152] Here, when wavelet transform is used as the frequency analysis performed by the time-frequency analysis unit 807, the transform performed by the time domain transform unit 808 becomes an inverse wavelet transform.

[0153] In addition, in the above-mentioned eighth embodiment, a processing unit can be set instead of the ICA processing unit 15, and the processing unit completes the neural network or principal component analysis by learning, separates the data representing noise in the time-frequency domain from the time-frequency domain observation sound data, and uses the separated data as time-frequency domain statistical noise data.

[0154] The eighth embodiment of the present invention has been described above.

[0155] According to the above embodiments, the noise component can be analyzed and stored based on the sound picked up by the microphone that picks up the error for making the adaptive filter adapt, and the pseudo noise data representing the stored noise component is used as the input of the adaptive filter to generate noise cancellation sound through active noise control. In addition, the problem of propagation delay of the feedback loop that causes the failure to cancel the noise well in the high frequency domain will not occur as in the feedback type active noise control. In addition, the pre-stored pseudo noise data is used as the input of the adaptive filter, so the problem of insufficient computing processing time that causes the failure to cancel the noise well in the high frequency domain in the case of the feedforward type active noise control when the sensor that detects noise-related events at a position upstream of the noise cancellation target position is close to the noise cancellation target position will not occur.

[0156] Thus, according to each embodiment, in a blind configuration in which no sensor for detecting events related to noise is provided upstream of a noise cancellation target position, active noise control that can cancel noise well even in a high frequency range can be performed.

[0157] Here, as the weight coefficient 704 in the seventh embodiment shown above, a weight coefficient that is fixedly set in advance may be used.

[0158] That is, in this case, if Fig. 9 As shown in a, the ICA processing unit 705, the update control unit 706, and the basis spectrum calculation unit 709 are deleted, and the weight coefficient of the orthogonal basis determined by the analysis performed in advance is pre-set as the weight coefficient 704.

[0159] Similarly, as the weight coefficient 804 in the eighth embodiment, a weight coefficient that is fixedly set in advance can be used.

[0160] That is, in this case, if Fig. 9 As shown in FIG. 8 b , the ICA processing unit 805 , the update control unit 806 , and the basis spectrum calculation unit 809 are deleted, and the weight coefficient of the orthogonal basis determined by analysis performed in advance is pre-set as the weight coefficient 804 .

[0161] With these configurations, even in a blind configuration in which no sensor for detecting an event related to noise is provided at a position upstream of a noise cancellation target position, noise can be cancelled well in a high frequency region.

Claims

1. An active noise control system for active noise control, characterized in that: have: A microphone outputs the picked-up sound as an observation signal; A speaker, outputting noise cancellation sound; Adaptive filter; Depository institutions; a noise separation unit that performs a noise separation process of separating a noise component from the observation signal at a predetermined timing and storing a pseudo noise signal representing the separated noise component in the storage unit; as well as an adaptive filter, taking the suspected noise signal stored in the storage mechanism as input, The adaptive filter performs an adaptive operation using the observation signal as an error, and generates a noise-canceled sound to be output from the speaker based on the pseudo-noise signal.

2. An active noise control system for active noise control, characterized in that: have: microphone; A frequency analysis mechanism, which converts the output of the microphone into a signal in the frequency domain and outputs it as an observation signal; A speaker, outputting noise cancellation sound; Adaptive filter; Depository institutions; Time domain transformation mechanism; A noise separation mechanism that performs a noise separation process of separating a noise component from the observation signal in the frequency domain and storing a pseudo-noise signal representing the separated noise component in the frequency domain in the storage mechanism at a predetermined timing; as well as an adaptive filter, taking the suspected noise signal stored in the storage mechanism as input, The adaptive filter performs adaptive action in the frequency domain using the observed signal as an error. The time domain conversion unit converts the output of the adaptive filter into a signal in the time domain to generate a noise cancellation sound to be output from the speaker.

3. The active noise control system according to claim 2, characterized in that: The frequency analysis mechanism transforms the microphone output into a frequency domain signal through fast Fourier transform and outputs it as an observation signal. The time domain transformation mechanism transforms the output of the adaptive filter into a time domain signal by inverse fast Fourier transformation.

4. An active noise control system for active noise control, characterized in that: have: microphone; An analysis mechanism that transforms the output of the microphone into a weight coefficient of a prescribed transformation function, and outputs the result as an observation signal, wherein the prescribed transformation function is a function that generates a signal in the time domain by weighting an orthogonal basis; A speaker, outputting noise cancellation sound; Adaptive filter; Depository institutions; Time domain transformation mechanism; a noise separation unit that performs a noise separation process of separating a noise component from the observation signal and storing the noise component as a suspected noise signal in the storage unit at a predetermined timing; as well as an adaptive filter, taking the suspected noise signal stored in the storage mechanism as input, The adaptive filter uses the observed signal as an error to perform adaptive action. The time domain conversion unit converts the output of the adaptive filter into a signal in the time domain by using the conversion function, and generates a noise cancellation sound to be output from the speaker.

5. The active noise control system according to claim 4, characterized in that: The analysis mechanism transforms the microphone output into the weight coefficient by fast Fourier transform and outputs it as an observation signal. The time domain transformation mechanism transforms the output of the adaptive filter into a time domain signal by inverse fast Fourier transformation.

6. The active noise control system according to claim 4, characterized in that: The analysis mechanism transforms the microphone output into the weight coefficient by wavelet transform and outputs it as an observation signal. The time domain transformation mechanism transforms the output of the adaptive filter into a time domain signal by inverse wavelet transformation.

7. The active noise control system according to any one of claims 1 to 6, characterized in that: The noise separation mechanism performs the noise separation process at a time when the output of the microphone becomes larger than a predetermined level as the predetermined timing.

8. The active noise control system according to any one of claims 1 to 6, characterized in that: The noise separation mechanism separates the noise component from the observation signal by independent component analysis (ICA).

9. The active noise control system according to any one of claims 1 to 6, characterized in that: The noise separation means infers and separates the noise component from the observation signal using a neural network that has been previously learned in such a manner as to infer the noise component from the observation signal.

10. The active noise control system according to any one of claims 1 to 6, characterized in that: The noise separation means estimates a principal component of a predetermined order of the observation signal by principal component analysis, and separates the estimated principal component as the noise component.

11. An active noise control system for performing active noise control, characterized in that: have: microphone; An analysis mechanism that transforms the output of the microphone into a weight coefficient of a prescribed transformation function, and outputs the result as an observation signal, wherein the prescribed transformation function is a function that generates a signal in the time domain by weighting an orthogonal basis; A speaker, outputting noise cancellation sound; Adaptive filter; A storage device for pre-storing the weight coefficient used to generate the suspected noise through the transformation function as a suspected noise signal; A time domain transformation mechanism; and an adaptive filter, taking the suspected noise signal stored in the storage mechanism as input, The adaptive filter uses the observed signal as an error to perform adaptive action. The time domain conversion unit converts the output of the adaptive filter into a signal in the time domain by using the conversion function, and generates a noise cancellation sound to be output from the speaker.

12. The active noise control system according to claim 11, characterized in that: The analysis mechanism transforms the microphone output into the weight coefficient by fast Fourier transform and outputs it as an observation signal. The time domain transformation mechanism transforms the output of the adaptive filter into a time domain signal by inverse fast Fourier transformation.

13. The active noise control system according to claim 11, characterized in that: The analysis mechanism transforms the microphone output into the weight coefficient by wavelet transform and outputs it as an observation signal. The time domain transformation mechanism transforms the output of the adaptive filter into a time domain signal by inverse wavelet transformation.

Images