Noise suppression device, noise suppression system, noise suppression method, and program

Abstract

Provided is, inter alia, a noise suppression device with which it is possible, in a case in which a noise suppression device is installed with respect to each area such as a seat in a single moving body such as a train or an aircraft, to increase the suppression performance so as to be higher than that of conventional noise suppression devices without increasing the size of, or complicating, each of the devices. The noise suppression device performs active noise control. The noise suppression device has suppression information, which relates to noise suppression, that is shared with a noise suppression device that is disposed within the same moving body and that corresponds to a reference microphone different from a reference microphone of this noise suppression device, the noise being suppressed using the common suppression information.

Description

Technical Field

[0001] The present invention relates to an active noise control (ANC) technology for suppressing external noise at a specific position.

Background Art

[0002] Non-Patent Document 1 is known as a conventional active noise control technology. In active noise control, a reference microphone, an error microphone, and a cancellation speaker are generally used. FIG. 1 shows a configuration example of a conventional noise suppression device. The reference microphone 91 picks up the noise emitted by the noise source. The cancellation speaker 92 plays back the cancellation signal generated by the suppression signal generation device 90 to emit a cancellation sound that cancels the noise. Further, the error microphone 93 picks up and feeds back the remaining noise. The suppression signal generation device 90 actively controls and generates a cancellation signal using the sound pickup signal of the reference microphone 91 and the sound pickup signal of the error microphone 93 so that the remaining noise is reduced. In order to emit a sound that can appropriately suppress the noise from the cancellation speaker 92, the suppression signal generation device 90 applies an adaptive filter to the sound picked up by the reference microphone 91 to generate a cancellation signal.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The closer the installation position of the reference microphone 91 is to the noise source compared to the position where noise suppression is desired, i.e., the installation position of the error microphone 93, the more time lag occurs from when the noise reaches the reference microphone 91 until it reaches the position where noise suppression is desired. Therefore, the suppression performance is more likely to improve. For this reason, arranging a plurality of reference microphones over a wide range can achieve higher suppression performance. However, complication of wiring and enlargement of the device become problems.

[0005] An object of the present invention is to provide a noise suppression device, a noise suppression system, a noise suppression method, and a program that can enhance the suppression performance compared to a conventional noise suppression device without making each device large or complicated in a case where noise suppression devices are installed for each area such as seats in the same moving body such as a train or an aircraft.

Means for Solving the Problem

[0006] In order to solve the above problems, according to one aspect of the present invention, a noise suppression device performs active noise control. The noise suppression device has suppression information, which is information related to noise suppression and is common to a noise suppression device corresponding to a reference microphone different from the reference microphone of this noise suppression device arranged in the same moving body, and suppresses noise using the common suppression information.

[0007] In order to solve the above problems, according to another aspect of the present invention, a noise suppression system includes N noise suppression devices that perform active noise control. The N noise suppression devices respectively correspond to N reference microphones arranged in the same moving body. The noise suppression device generates a cancellation signal for suppressing noise using the sound collection signal of the corresponding reference microphone and a model, and the noise suppression system includes a suppression information sharing unit that receives suppression information from a certain noise suppression device and outputs, to another noise suppression device, the suppression information to be used in the other noise suppression device among the suppression information. The suppression information of a certain noise suppression device corresponding to a front area is used by another noise suppression device corresponding to a rear area with respect to the traveling direction of the moving body.

Effects of the Invention

[0008] According to the present invention, there is an effect that the noise suppression performance can be improved as compared with the conventional noise suppression device.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Mode for Carrying Out the Invention

[0010] Hereinafter, embodiments of the present invention will be described. In the drawings used in the following description, components having the same function and steps performing the same process are denoted by the same reference numerals, and redundant descriptions are omitted.

[0011] <Highlights of the First Embodiment> · In a case where noise suppression devices are installed in each area such as seats in the same moving body such as a train or an aircraft, a cancellation sound, which is a signal for suppressing the noise in the target area, is generated using the information of each noise suppression device. With such a configuration, especially when having a similar structure in any location such as a train or an aircraft, (i) the calculation results at each noise suppression device are transmitted from the noise suppression device at the front of the traveling direction to the noise suppression device at the rear of the traveling direction, or (ii) transmitted to a system for aggregating information, and the suppression performance is enhanced by sharing the calculation results.

[0012] · Using the sound collection signal of the reference microphone of another noise suppression device, a cancellation sound is generated as a pseudo multi-channel reference microphone. With such a configuration, by using the sound collection signal of the reference microphone of the noise suppression device close to the noise source for generating the cancellation sound, the installation position of the reference microphone is closer to the noise source compared to the position where the noise is to be suppressed, and a time delay occurs from when the noise arrives at the reference microphone until it arrives at the position where the noise is to be suppressed, making it easier to improve the suppression performance.

[0013] <First Embodiment> FIG. 2 shows a functional block diagram of the noise suppression system according to the first embodiment, FIG. 3 shows its processing flow, and FIG. 4 shows a functional block diagram of the noise suppression device according to the first embodiment.

[0014] The noise suppression system includes N reference microphones 91-n, N cancellation speakers 92-n, N error microphones 93-n, N noise suppression devices 100-n, and a suppression information sharing unit 130. Here, let n = 1, 2, …, N, and N be any integer of 2 or more. The N reference microphones 91-n, the N cancellation speakers 92-n, and the N error microphones 93-n are arranged in the same moving body. The n-th noise suppression device 100-n includes a suppression signal generation unit 110-n and a determination unit 120-n.

[0015] The n-th noise suppression device 100-n takes as inputs the sound pickup signal x(r, n) of the n-th reference microphone 91-n and the sound pickup signal x(e, n) of the n-th error microphone 93-n, has common suppression information with the noise suppression device 100-n' corresponding to different reference microphones 91-n', error microphones 93-n', and cancellation speakers 92-n' arranged in the same moving body, and uses this suppression information to generate a cancellation signal (hereinafter, also referred to as a "suppression signal") y(n) and output it to the n-th cancellation speaker 92-n. Here, n' = 1, 2, …, N, and n ≠ n'. The n-th reference microphone 91-n, the error microphone 93-n, and the cancellation speaker 92-n are arranged at appropriate positions for realizing noise suppression in the n-th area. The appropriate position is, for example, when the same moving body is a train, a position where the microphones are linearly arranged at equal intervals at the same position of each seat in the same direction as the traveling direction.

[0016] The noise suppression device is a special device configured by loading a special program into a known or dedicated computer having, for example, a central processing unit (CPU) and a main memory device (RAM: Random Access Memory). The noise suppression device executes each process under the control of, for example, the central processing unit. The data input to the noise suppression device and the data obtained by each process are stored, for example, in the main memory device, and the data stored in the main memory device is read out to the central processing unit as needed and used for other processes. Each processing unit of the noise suppression device may be configured at least in part by hardware such as an integrated circuit. Each storage unit included in the noise suppression device can be configured by, for example, a main memory device such as a RAM (Random Access Memory), or middleware such as a relational database or a key-value store. However, each storage unit does not necessarily have to be provided inside the noise suppression device, and may be configured by an auxiliary storage device configured by a semiconductor memory element such as a hard disk, an optical disk, or a flash memory (Flash Memory), and may be provided outside the noise suppression device.

[0017] Hereinafter, each part will be described.

[0018] <Reference microphone 91-n> The reference microphone 91-n picks up the sound to be suppressed (S91-n) and outputs a picked-up signal x(r,n). The sound to be suppressed picked up by the reference microphone 91-n is hereinafter referred to as "noise".

[0019] <Error microphone 93-n> The error microphone 93-n picks up the sound not suppressed by the reproduced sound reproduced from the cancellation speaker 92, including the remaining noise (S93-n), and outputs a picked-up signal x(e,n).

[0020] In at least one of the following determination unit 120-n and suppression signal generation unit 110-n, the suppression information p(n) output by the suppression information sharing unit 130 is shared. Note that "sharing" means that a certain noise suppression device outputs suppression information to the suppression information sharing unit 130, the suppression information sharing unit 130 receives and stores the suppression information, another noise suppression device receives the suppression information from the suppression information sharing unit 130, and a certain noise suppression device and another noise suppression device store the same suppression information. Also, a state in which a certain suppression information is shared means a state in which the suppression information shared by a certain noise suppression device and another noise suppression device is stored in each of them. First, the process of not sharing the suppression information p(n) will be described, and then the process of sharing the suppression information p(n) in combination with the process of the suppression information sharing unit 130 will be described.

[0021] <Determination unit 120-n> The determination unit 120-n takes as input the cancellation signal y(n) of the suppression signal generation unit 110-n described later, and uses the cancellation signal y(n) to determine whether ANC operates stably (S120-n), and outputs a determination result j(n). For example, it is determined whether ANC operates stably based on the magnitude relationship between the amount of change per unit time of the cancellation signal y(n) and a predetermined threshold value. For example, when the amount of change in the power of the cancellation signal y(n) is greater than a predetermined threshold value, it is determined that ANC does not operate stably, and when it is less than or equal to the predetermined threshold value, it is determined that ANC operates stably. The predetermined threshold value may be calculated in advance by simulation or the like.

[0022] When it is determined that the ANC does not operate stably (NO in S120-n), the determination unit 120-n outputs a control signal indicating that the ANC operation is not to be performed. Since it suffices to be able to control so as not to perform the ANC operation, as the output destination of the control signal, the suppression signal generation unit 110-n, the cancellation speaker 92-n, etc. described later can be considered. (i) The determination unit 120-n may control so as not to perform the ANC operation by stopping the reproduction of the cancellation signal y in the cancellation speaker 92-n. In this case, a control signal may be output to the cancellation speaker 92-n to stop the reproduction of the cancellation signal y(n), or a control signal may be output to the suppression signal generation unit 110-n to stop the generation or output of the cancellation signal y(n) in the suppression signal generation unit 110-n. Also, (ii) the determination unit 120-n may control so as not to perform the ANC operation by outputting a control signal to the suppression signal generation unit 110-n and stopping the update of the adaptive filter. Depending on the situation, (i) to (ii) may be switched for control. However, by adopting (i) and maintaining the update of the adaptive filter, it is possible to prevent the operation at the time of restarting the ANC operation from becoming unstable. The time during which the ANC operation is not performed may be set to the time required until the operation of the ANC becomes stable.

[0023] When it is determined that the ANC operates stably (YES in S120-n), the determination unit 120-n outputs a control signal indicating that the ANC operation is to be performed. However, when a control signal indicating that the ANC operation is not to be performed is not received in each unit, it may be configured to basically perform the ANC operation, and the determination unit 120-n may be configured not to output a control signal indicating that the ANC operation is to be performed. In this case, when it is determined that the ANC operates stably (YES in S120-n), the determination unit 120-n does not perform any processing and transfers the processing to the suppression signal generation unit 110-n.

[0024] <Suppression signal generation unit 110-n> The suppression signal generation unit 110-n takes the picked-up signal x(r,n) and the picked-up signal x(e,n) as inputs, and uses the picked-up signal x(r,n) and the model to generate (S110-n) and output a cancellation signal y(n) for suppressing noise. Further, the suppression signal generation unit 110 updates the model using the picked-up signals x(r,n) and x(e,n). For example, the model is an adaptive filter, and its filter coefficients are updated. Note that the model is not limited to filter coefficients, and may be other models, or may be a model generated using technologies such as deep learning.

[0025] As a method for generating the cancellation signal, a conventional technique can be used. For example, the method of Non-Patent Document 1 can be used. In the present embodiment, a feedforward type ANC is realized by the picked-up signal x(r,n), the picked-up signal x(e,n), and the cancellation signal y(n). The interference sound between the noise from the noise source and the reproduced sound of the cancellation signal y is detected by the error microphone 93-n, and the noise from the noise source is detected by the reference microphone 91-n. The cancellation signal y(n) is generated by inputting the picked-up signal x(r,n) of the reference microphone 91-n to the adaptive filter realized by a digital filter, and is reproduced by the cancellation speaker 92-n. The reproduced sound of the cancellation signal y(n) propagates through the secondary path, which is a series of transmission systems from the cancellation speaker 92-n to the error microphone 93-n. Then, using the picked-up signal x(r,n) of the reference microphone 91-n and the picked-up signal x(e,n) of the error microphone 93-n, the filter coefficients of the adaptive filter are updated by an adaptive algorithm so that the input of the error microphone 93-n is minimized. Since a conventional update method can be used as the method for updating the filter coefficients of the adaptive filter, the description thereof is omitted. In the feedforward type ANC, a secondary path model obtained by estimating the secondary path is used to compensate for the influence of the secondary path in the adaptive algorithm.

[0026] <cancellation speaker 92-n> The cancellation speaker 92-n takes the cancellation signal y(n) as input and reproduces the cancellation signal y(n) (S92-n). When the reproduced sound from the cancellation speaker 92 and the noise to be suppressed are in a completely opposite phase, the reproduced sound and the noise to be suppressed overlap, that is, the sound waves overlap, and the waves cancel each other out, so the noise is suppressed. As described above, the sound not suppressed by the reproduced sound from the cancellation speaker 92-n is picked up by the error microphone 93-n.

[0027] <Suppression information sharing unit 130> The suppression information sharing unit 130 receives and stores the suppression information from N noise suppression devices 100-n. Further, the suppression information sharing unit 130 outputs, to each noise suppression device 100-n, the suppression information to be used in each noise suppression device 100-n among the suppression information. Hereinafter, the suppression information output to the nth noise suppression device 100-n is denoted as p(n).

[0028] The suppression information is information related to noise suppression. For example, it is the sound pickup signals x(r,n), x(e,n), the cancellation signal y(n), the determination result j(n), and the model used in the suppression signal generation unit 110-n. These suppression information are shared among N noise suppression devices 100-n.

[0029] (1) When used for generating cancellation sound by pseudo multi-channel reference microphones The suppression information sharing unit 130 receives N sound pickup signals x(r,n) and outputs N - 1 sound pickup signals x(r,n') to the nth suppression signal generation unit 110-n. The nth suppression signal generation unit 110-n generates and outputs a cancellation signal y(n) for suppressing noise using the N - 1 sound pickup signals x(r,n'), the sound pickup signal x(r,n), and the model. In this case, the model takes N sound pickup signals x(r,n) as input and outputs the cancellation signal y(n).

[0030] (2) When sharing the suppression information of the noise suppression device corresponding to the front area and using it in the noise suppression device corresponding to the rear area The suppression information sharing unit 130 may receive suppression information of a noise suppression device corresponding to a front area (for example, a sound collection signal of a reference microphone located in front with respect to the traveling direction of the moving body, a sound collection signal of an error microphone, a cancellation signal input to a cancellation speaker, a determination result of a determination unit of a noise suppression device corresponding to a front area, a model of a suppression signal generation unit of a noise suppression device corresponding to a front area), and output it to a noise suppression device corresponding to a rear area. As shown in FIG. 5, the suppression information output by the noise suppression device corresponding to the foremost front area may be output to noise suppression devices corresponding to all other rear areas at once. Alternatively, as shown in FIG. 6, the suppression information output by the noise suppression device corresponding to a front area may be output to a noise suppression device corresponding to an area one position rearward of that noise suppression device, and the suppression information may be transmitted in order. In particular, by using the suppression information output by the noise suppression device corresponding to the foremost front area in a noise suppression device corresponding to a rear area, it becomes easier to create a time margin in the rear area. In a moving body such as a train, since it travels on a fixed route with a fixed vehicle, the characteristics of the noise are similar in each vehicle, it is easy to predict the noise suppression environment, and the processing result of the noise suppression device corresponding to the front area can be shared and used in a noise suppression device corresponding to a rear area. When it consists of one or more moving bodies (such as vehicles) such as a train, since the front vehicle and the rear vehicle have a similar structure, the area inside the front vehicle may be regarded as the front area, and the area inside the rear vehicle may be regarded as the rear area. Also, since the characteristics of the noise are similar even within one vehicle, the front seats within one vehicle may be regarded as the front area, and the rear seats in the same vehicle may be regarded as the rear area.

[0031] For example, it can be used as follows.

[0032] (i) For example, the sound collection signal of a reference microphone located in front may be used by a noise suppression device corresponding to a reference microphone located in the rear to generate a cancellation signal y(n), or may be used for updating the model.

[0033] (ii) The cancellation signal input to the determination unit of the noise suppression device corresponding to the front area may be used as the input to the determination unit of the noise suppression device corresponding to the rear area, and whether the ANC operates stably may be determined using the cancellation signal.

[0034] (iii) The determination result of the determination unit of the noise suppression device corresponding to the front area may be used as the determination result of the determination unit of the noise suppression device corresponding to the rear area. In this case, the determination unit of the noise suppression device corresponding to the rear area can omit the determination process.

[0035] (iv) When having a similar structure in any location such as a train or an aircraft, the model updated by the suppression signal generation unit of the noise suppression device corresponding to the front area may be used as the model of the suppression signal generation unit of the noise suppression device corresponding to the rear area. When performing active noise suppression at a plurality of positions with similar noise characteristics such as train seats, stable operation can be realized by using the calculation result of ANC at one seat for other seats as well. In particular, since the noise characteristics in the rear area follow the noise characteristics in the front area, sharing the model obtained by the noise suppression device corresponding to the front area with the noise suppression device corresponding to the rear area creates a time margin and can improve the suppression performance. In this case, the update process of the model of the suppression signal generation unit of the noise suppression device corresponding to the rear area can be omitted.

[0036] <Effect> With the above configuration, the suppression performance can be improved compared to the conventional noise suppression device. By sharing the sound collection signals obtained from a large number of microphones (reference microphones, error microphones) distributed in the same moving body, the cancellation signals, determination results, and models obtained using these sound collection signals with other noise suppression devices, it is possible to increase the accuracy and processing speed of the ANC. The ANC system can operate safely against noise situations that vary depending on the location, and the convergence speed of signal calculation in the suppression signal generation unit can be improved.

[0037] <Modification Example> In this embodiment, it has been described that the suppression information is shared by N noise suppression devices 100-n, but the suppression information may be shared by a part of the N noise suppression devices 100-n.

[0038] <Second Embodiment> The description will focus on the parts different from the first embodiment. In this embodiment, in order to generate a cancellation signal for canceling noise, the received signal x(r,n) (the signal to be suppressed) that will be received by the reference microphone 91-n in the future is predicted from the N-1 received signals x(r,n') respectively received by N-1 reference microphones 91-n'. By using the predicted value of the received signal x(r,n) that will be received in the future, it is possible to cope with the deterioration of the suppression performance due to the delay generated in the process of processing.

[0039] In time series analysis for predicting general time series signals, a state space model that performs linear prediction on stationary signals has been studied for a long time (see Reference 1). (Reference 1) Kalman, R.E., "A new approach to linearfiltering and prediction problems", Trans. ASME-J, Basic Eng.(ser.D), 82, pp.35-45., 1960.

[0040] Also, in Reference 2, in order to improve the noise suppression performance in earphones, microphones are installed on the outside and inside of the earphones, and a function of predicting and suppressing the noise in the surrounding environment is realized by using a predictor configured with a deep neural network (DNN). (Reference 2) Jang, Young-Jae, Jaehyun Park, Won-Cheol Lee, and Hong-June Park., "A Convolution-Neural-Network Feedforward Active-Noise-Cancellation System on FPGA for In-Ear Headphone", Applied Sciences 12, no. 11: 5300., 2022.

[0041] First, noise is often non - stationary for the following reasons. · The position of the noise source may change from moment to moment. · The noise itself may consist of a variety of sources and is likely to be non - stationary.

[0042] Therefore, in Reference 1, since a state - space model that performs linear prediction on a stationary signal is used, it is difficult to simply apply it to noise with non - stationary mean, variance, and covariance for prediction. Also, when applying the DNN of Reference 2 to multiple channels, the amount of calculation becomes large, resulting in a heavy computational load, and it is difficult to calculate simultaneously with multiple microphones.

[0043] In this embodiment, such problems are solved. FIG. 7 shows a functional block diagram of the noise suppression system according to the second embodiment, FIG. 8 shows a processing flow of the determination stage of the noise suppression system, and FIG. 9 shows a processing flow of the suppression stage of the noise suppression system.

[0044] The noise suppression system includes N reference microphones 91 - n, N canceling speakers 92 - n, N error microphones 93 - n, N noise suppression devices 100 - n, a suppression information sharing unit 130, a weight set determination unit 240, and a sound collection signal prediction unit 250. The reference microphone 91 - n outputs the sound collection signal x(r,n,t) not to the noise suppression device 100 - n but to the suppression information sharing unit 130. However, t is an index indicating the sample time. The n - th noise suppression device 100 - n takes as input the predicted value x’(r,n,t + τ1) of the sound collection signal x(r,n,t+τ1) after τ1 samples of the reference microphone 91 - n instead of the sound collection signal x(r,n,t). The processing of the noise suppression system consists of a determination stage and a suppression stage.

[0045] When the sound field is regarded as a linear system, the sound field around the area A to be suppressed in FIG. 10 can be expressed as a simple superposition of the sounds existing around the area A. Among the sounds that are first generated and picked up in area A, there must be sounds that are first picked up in area B except for the sounds picked up in area A. Assume that the sound that is first generated and picked up in area A is likely to be emitted by the person using area A, and it is an important sound (outside the suppression target) for the person using area A. It is sufficient to estimate the sound to be suppressed from the sounds picked up in the area B group. Therefore, in the present embodiment, the future sound pickup signal of the area to be suppressed is predicted by superimposing the current or past sound pickup signals picked up in the area outside the suppression target.

[0046] In the present embodiment, in the suppression stage, the future observed sound pickup signal of the suppression target is predicted from the sound pickup signals other than the sound pickup signal x of the suppression target. In the determination stage, the weights used for predicting the future observed sound pickup signal of the suppression target are determined. Since the optimal combination of weights at each time and further the optimal combination of weights over the entire time period can be considered, for example, L1 regularization (LASSO) is used for optimization at each time, and for example, the greedy method is used for optimization over the entire time period.

[0047] First, the determination stage will be described. <Determination Stage> In the determination stage, first, non-stationary noise is picked up by N reference microphones 91-n, and the picked-up signal x(r, n, t) is stored in the suppression information sharing unit 130.

[0048] <Weight Set Determination Unit 240> The weight set determination unit 240 extracts the N picked-up signals x(r, n, t) from the suppression information sharing unit 130 from the picked-up signal x(r, n, t+S) at time t+S to the picked-up signal x(r, n, t-τ1-τ2) at time t-τ1-τ2. Here, S is the time width used for prediction. Let τ1 be a parameter indicating the time difference between the current time and the time to be predicted, and τ2 be a parameter indicating the time width to go back for prediction. FIG. 11 shows a functional block diagram of the weight set determination unit 240. The weight set determination unit 240 includes a selection unit 241, a weight set estimation unit 243, and a weight set selection unit 245.

[0049] <Selection unit 241> The selection unit 241 takes as input the N sound collection signals x(r, n, t) from the sound collection signal x(r, n, t+S) at time t+S to the sound collection signal x(r, n, t−τ1−τ2) at time t−τ1−τ2, selects the sound collection signal x(r, m, t) to be suppressed (S241), and among the selected sound collection signals, the S+1 sound collection signals x(r, m, t), x(r, m, t+1),..., x(r, m, t+S) from time t to time t+S are used as the correct data X(r, m, t)=[x(r, m, t), x(r, m, t+1),..., x(r, m, t+S)] T ∈R (S+1)×1 And among the unselected sound collection signals, the (τ2+S+1)×(N−1) sound collection signals x(r, m', t−τ1−τ2), x(r, m', τ1−τ2+1),..., x(r, m', t−τ1+S) from time t−τ1−τ2 to time t−τ1+S are used as the explanatory data. Note that the explanatory data means data used for estimating the correct data X(r, m, t). m is any one of 1, 2,..., N and is an index indicating the selected sound collection signal, m' is an index indicating a sound collection signal other than the selected sound collection signal, m'=1, 2,..., N, and m≠m'.

[0050] To consider the influence of the signal observation delay due to the physical distance between the reference microphone 91−m corresponding to the sound collection signal x(r, m, t) to be suppressed and the other N−1 reference microphones 91−m', the sound collection signals obtained by shifting the sound collection signals at one reference microphone by one time step are combined in the channel direction. FIG. 12 is a diagram for explaining the correct data and the explanatory data. For example, taking the sound collection signals x(r, m', t−τ1−τ2), x(r, m', τ1−τ2+1),..., x(r, m', t−τ1+S) of a certain reference microphone 91−m' as one set for each time width S, τ2+1 sets are created by shifting by one time step as follows. X(r,m',t - τ1)=[x(r,m',t - τ1),x(r,m',t - τ1 + 1),…,x(r,m',t - τ1 + S)]∈R (S+1)×1 , X(r,m',t - τ1 - 1)=[x(r,m',t - τ1 - 1),x(r,m',t - τ1 + 0),…,x(r,m',t - τ1 - 1 + S)]∈R (S+1)×1 , … X(r,m',t - τ1 - τ2)=[x(r,m',t - τ1 - τ2),x(r,m',t - τ1 - τ2 + 1),…,x(r,m',t - τ1 - τ2 + S)]∈R (S+1)×1 Furthermore, integrate in the channel direction to generate the following matrix X(m,t)∈R (S+1)×(N-1)τ_2 as explanatory data. Here, the superscript τ_2 means τ2.

Number

[0051] The selection unit 241 outputs the correct data X(r,m,t)∈R (S+1)×1 and the explanatory data X(m,t)∈R (S+1)×(N-1)τ_2 . The selection unit 241 may select, as the sound collection signal to be suppressed, the sound collection signal specified by some input means (such as a mouse or a keyboard) manually, or may automatically select the sound collection signal to be suppressed randomly or according to a predetermined rule. Also, the selection unit 241 may sequentially select N sound collection signals as the sound collection signals to be suppressed, or may sequentially select only a part of the N sound collection signals as the sound collection signals to be suppressed.

[0052] <Weight set estimation unit 243> The weight set estimation unit 243 takes the correct data X(r,m,t)∈R (S+1)×1 and the explanatory data X(m,t)∈R (S+1)×(N-1)τ_2 as inputs, and the optimal weight W(m,t)=[W(m,t,1),W(m,t,2),…,W(m,t,(N - 1)τ2)] T ∈R ((N-1)τ_2)×1Estimate it (S243) and output the estimated weight W(m,t).

[0053] In this embodiment, under the assumption that the correct data X(r,m,t) is obtained by the regression of the explanatory data X(m,t), the optimal weight W(m,t) is estimated by the least squares method. The loss function is L = |X(r,m,t) - X(m,t) T W(m,t)| 2 +Q, where Q is a regularization term, and the regularization term Q is with the update constant as α, Q = α|W(m,t)| or Q = α|W(m,t)| 2 is.

[0054] The weight set estimation unit 243 optimizes the weights at each time by, for example, L1 regularization (LASSO). Fix the time (from t - T to t) to be predicted during suppression, and set t(P) < … < t(3) < t(2) < t(1) < t. Using the explanatory data X(m,t) ∈ R (S+1)×(N-1)τ_2 to calculate the loss function L = |X(r,m,t) - X(m,t) T W(m,t)| 2 +Q. For the minimization problem of the loss function L, as an index of the model fit, for example, the Akaike information criterion (AIC) or the Bayesian information criterion (BIC) can be used.

[0055] For example, when predicting the correct data X(r,m,t(p)) at time t(p), as shown in FIG. 13, multiply the explanatory data

Equation

[0056] The weight set estimation unit 243 changes the time t P times, estimates and outputs the respective optimal weights W(m, t(1)), …, W(m, t(P)) at P times t(1), …, t(P). P is any integer greater than or equal to 2. FIG. 14 shows the results of multiplying the optimal weights W(m, t(1)), …, W(m, t(P)) at each time by the explanatory data. The black dot squares indicate that the value remains after multiplying by the weight, and the blank squares indicate that the value does not remain after multiplying by the weight.

[0057] <Weight set selection unit 245> The weight set selection unit 245 takes as input the P optimal weights W(m, t(p)), applies the weights W(m, t(1)), W(m, t(2)), …, W(m, t(P)) obtained at each time t(1), t(2), …, t(P)) to other times as well, selects the weight that minimizes the loss function L the most as the final optimal weight W(m) (S245), and outputs it to the voice signal prediction unit 250. For example, the greedy method is used to obtain the optimal weight set over the entire time period. For example, the weight W(m, t(u)) obtained at a certain time t(u) is applied to the explanatory data X(m, t(p)) at each time t(p) to obtain P loss functions L(t(p), W(m, t(u)), and the sum thereof is obtained. The same process is performed for u = 1, 2, …, P, and the weight with the smallest sum is taken as the final optimal weight. The above process is performed before moving to the suppression stage.

[0058] <Suppression stage> <Voice signal prediction unit 250> Before the suppression process, the voice signal prediction unit 250 receives the optimal weight W(m). The voice signal prediction unit 250 extracts N - 1 voice signals x(r, m', t) other than the voice signal to be suppressed from the voice signal x(r, m', t) at time t to the voice signal x(r, m', t - τ2) at time t - τ2 from the suppression information sharing unit 130.

[0059] The sound collection signal prediction unit 250 uses X(m,t) and W(m) to obtain a predicted value X'(r,m,t + τ1) of the sound collection signal to be suppressed after τ1 samples, where X'(r,m,t + τ1)=[x'(r,m,t - S + τ1), x'(r,m,t - S + 1 + τ1),…,x'(r,m,t + τ1)] T =X(m,t) T W(m)∈R (S+1)×1 and outputs it to the noise suppression device 100 - m corresponding to the sound collection signal to be suppressed (S250). Note that

Equation

[0060] The processing of the noise suppression device 100 - m in the suppression stage is the same as that of the first embodiment, except that the processing is performed using the predicted value x'(r,m,t + τ1) of the sound collection signal instead of the sound collection signal x(r,m).

[0061] <Suppression signal generation unit 110 - n> The suppression signal generation unit 110 - n takes the predicted value x'(r,n,t + τ1) of the sound collection signal and the sound collection signal x(e,n,t) as inputs, and uses the predicted value x'(r,n,t + τ1) and the model to generate (S110 - n) and output a cancellation signal y(n,t) for suppressing noise. Also, the suppression signal generation unit 110 updates the model using the predicted value x'(r,n,t + τ1) and the sound collection signal x(e,n,t).

[0062] <Effect> With the above configuration, since the difference between the time when the sound will be collected in the future and the time until the noise actually arrives at the position where the noise is to be suppressed becomes larger, the time margin becomes larger, and the suppression performance can be improved compared to the conventional noise suppression device.

[0063] <Other modification examples> The present invention is not limited to the above-described embodiments and modifications. For example, the various processes described above may be executed not only in time series according to the description, but also in parallel or individually according to the processing capabilities of the device executing the processes or as necessary. In addition, various modifications can be made as appropriate without departing from the spirit of the present invention.

[0064] <Program and Recording Medium> The various processes described above can be implemented by causing a program for executing each step of the above method to be read into the storage unit 2020 of the computer shown in FIG. 15 and operating the control unit 2010, the input unit 2030, the output unit 2040, etc.

[0065] The program describing this processing content can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any of a magnetic recording device, an optical disk, a magneto-optical recording medium, a semiconductor memory, etc. may be used.

[0066] In addition, the distribution of this program can be performed, for example, by selling, transferring, lending, etc. a portable recording medium such as a DVD or a CD-ROM on which the program is recorded. Furthermore, the program may be stored in the storage device of a server computer, and configured to distribute the program by transferring the program from the server computer to other computers via a network.

[0067] A computer that executes such a program first stores, for example, a program recorded on a portable recording medium or a program transferred from a server computer in its own storage device. Then, at the time of executing processing, the computer reads the program stored in its own recording medium and executes processing according to the read program. As another execution form of this program, the computer may directly read the program from the portable recording medium and execute processing according to the program. Further, each time a program is transferred from the server computer to this computer, the computer may sequentially execute processing according to the received program. Further, the transfer of the program from the server computer to this computer may not be performed, and the above-described processing may be executed by a so-called ASP (Application Service Provider) type service that realizes a processing function only by the execution instruction and result acquisition. Note that the program in this embodiment includes information for use in processing by an electronic computer that conforms to the program (data having a property of defining the processing of the computer but not being a direct instruction to the computer).

[0068] Further, in this embodiment, the present apparatus is configured by causing a computer to execute a predetermined program, but at least a part of these processing contents may be realized hardware-wise.

Claims

1. A noise suppression device that performs active noise control, comprising: a noise suppression device corresponding to a reference microphone different from the reference microphone of the noise suppression device, arranged within the same moving body, having common suppression information which is information related to noise suppression, and suppressing noise using the common suppression information; using the suppression information of a noise suppression device corresponding to an area in front of the area corresponding to the noise suppression device with respect to the traveling direction of the moving body; a noise suppression device.

2. A noise suppression system including N noise suppression devices that perform active noise control, wherein: the N noise suppression devices respectively correspond to N reference microphones arranged within the same moving body; the noise suppression device generates a cancellation signal for suppressing noise using the sound collection signal of the corresponding reference microphone and a model; the noise suppression system includes a suppression information sharing unit that receives suppression information from a certain noise suppression device and outputs, to another noise suppression device, the suppression information to be used by the other noise suppression device among the suppression information; using the suppression information of the certain noise suppression device corresponding to an area in front with respect to the traveling direction of the moving body by the other noise suppression device corresponding to an area in the rear; a noise suppression system.

3. The noise suppression system according to Claim 2, wherein: the noise suppression device has a determination unit that determines whether active noise control operates stably using the cancellation signal reproduced by a cancellation speaker; the suppression information includes: (i) the sound collection signal of the reference microphone (ii) the cancellation signal reproduced by the cancellation speaker (iii) the determination result of the determination unit (iv) the model used when generating the cancellation signal from the sound collection signal of the reference microphone including at least any one of; a noise suppression system.

4. A noise suppression method using a noise suppression device that performs active noise control, comprising: a noise suppression device corresponding to a reference microphone different from the reference microphone of the noise suppression device, arranged within the same moving body, having common suppression information which is information related to noise suppression, and suppressing noise using the common suppression information; using the suppression information of a noise suppression device corresponding to an area in front of the area corresponding to the noise suppression device with respect to the traveling direction of the moving body; a noise suppression method.

5. A noise suppression method using a noise suppression system including N noise suppression devices that perform active noise control, wherein the N noise suppression devices respectively correspond to N reference microphones arranged in the same moving body, the noise suppression device generates a cancellation signal for suppressing noise by using the sound collection signal of the corresponding reference microphone and a model, the noise suppression system includes a suppression information sharing unit that receives suppression information from a certain noise suppression device and outputs, to other noise suppression devices, suppression information among the suppression information that is used in the other noise suppression devices, suppression information of the certain noise suppression device corresponding to a front area is used by the other noise suppression device corresponding to a rear area with respect to the traveling direction of the moving body, Noise suppression method. **Claim 6**: A noise suppression device that performs active noise control, a noise suppression device corresponding to a reference microphone different from the reference microphone of the noise suppression device, arranged in the same moving body, having common suppression information that is information related to noise suppression, suppressing noise by using the common suppression information, predicting a sound collection signal to be collected in the future by the reference microphone of the noise suppression device by using a sound collection signal collected by a reference microphone different from the reference microphone of the noise suppression device, and generating a cancellation signal for suppressing noise by using the predicted sound collection signal, Noise suppression device. **Claim 7**: A noise suppression method using a noise suppression device that performs active noise control, a noise suppression device corresponding to a reference microphone different from the reference microphone of the noise suppression device, arranged in the same moving body, having common suppression information that is information related to noise suppression, suppressing noise by using the common suppression information, predicting a sound collection signal to be collected in the future by the reference microphone of the noise suppression device used in the noise suppression method by using a sound collection signal collected by a reference microphone different from the reference microphone of the noise suppression device used in the noise suppression method, and generating a cancellation signal for suppressing noise by using the predicted sound collection signal, Noise suppression method. **Claim 8** A program for causing a computer to function as the noise suppression device according to Claim 1 or Claim 6.

Images