JP2014160214A - Method, device and program for removing impulse response noise - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method, a device and a program for removing impulse response noise capable of removing non-signal noise included in measured impulse responses.SOLUTION: The method for removing impulse response noise includes: a non-signal noise detection step of detecting non-signal noise from a non-signal response beforehand measured in measurement space for measuring impulse responses; a probability density function calculation step of calculating a probability density function of the non-signal noise on the basis of the average and the variance of the non-signal noise; a confidence interval calculation step of calculating the range of amplitude indicating a predetermined confidence interval, on the basis of the probability density function; an average calculation step of calculating the average of beforehand measured multiple impulse responses; a non-response interval detection step of detecting amplitude included in the range of the amplitude of the confidence interval, among the average amplitude of the impulse responses, and detecting an interval including the detected amplitude as a non-response interval; and a non-response interval removal step of removing the non-response interval from the beforehand measured impulse responses.

Description

  The present invention relates to an impulse response noise removing method and apparatus for removing noise from an impulse response, and a program thereof.

  The impulse response is an output when a signal having an infinitesimal time width and an infinite magnitude called an impulse is input to a target system, that is, a system response. However, since an ideal impulse as described above is actually difficult to generate, the impulse response is usually measured using a sweep signal whose phase is shifted for each frequency. Specific examples of such an impulse response include various ones depending on the measurement target. For example, a head impulse response indicating acoustic transfer characteristics from a sound source position to an ear position in a free sound field, or a sound source in an acoustic space. An indoor impulse response indicating acoustic transmission characteristics from the position to the sound receiving point can be exemplified.

  Here, when the impulse response is measured, the response may include a non-response interval (hereinafter referred to as a non-response interval). For example, the head impulse response described above is strictly defined as “an impulse response as a sound propagation characteristic from the center position of both ears when there is no head in the free sound field to the ear canal entrance or the eardrum when the head is present”. However, in practice, it is impossible to place a sound source for measurement at the center position of both ears of the measurement subject. Therefore, normally, the head impulse response is measured by arranging the sound source at a position at a predetermined distance from the measurement subject's head. Therefore, at the beginning of the measured response, a delay portion corresponding to the time required for propagation of the distance between the measurement subject's head and the sound source is included as a non-response section.

  Also, when measuring the impulse response, the measurement time is often set with a margin for the actual response length. Therefore, as a result, no response corresponding to the above-described margin time also occurs in the section after the impulse response has converged. Since these non-response sections are before application of impulses or after convergence of responses, they should be essentially no signal (amplitude 0), but noise in the absence of a signal to be measured (for example, background noise in the acoustic field) , Background noise, etc., hereinafter referred to as “no-signal noise”), a signal corresponding to no-signal noise may be recorded in the no-response interval.

  Since such no-signal noise basically has an irregular amplitude, it is possible to cancel and suppress the no-signal noise, for example, by measuring the impulse response multiple times under the same conditions and averaging them. It is. For example, in Patent Document 1, a characteristic of a system included in an impulse response is calculated by such averaging, and a technique for removing no-signal noise from the impulse response by truncating the amplitude below a predetermined threshold value as 0. Has been proposed.

  As another viewpoint, Patent Document 2 proposes a technique for measuring the level of environmental noise (no signal noise) in advance to determine the level of a measurement signal and improving the S / N ratio.

JP 2012-128207 A (see FIG. 2) JP 2002-330500 A (see FIG. 8)

  However, the technique proposed in Patent Document 1 exhibits a certain noise suppression effect by averaging impulse responses and truncating with a threshold value. However, for some reason, a value peculiar to no-signal noise (for example, an unnaturally large amplitude value). ) Occurs, it is not canceled by the threshold value, and is not canceled out by averaging. Therefore, there is a problem that the signal remains as no-signal noise. Moreover, since the technique proposed in Patent Document 1 smoothes signal portions other than no-signal noise by averaging impulse responses, the characteristics of the system included in the impulse responses may not be obtained.

  Further, since the technique proposed in Patent Document 2 is pre-processing of impulse response measurement in the first place, there is a problem that no-signal noise included in the already measured impulse response cannot be removed.

  The present invention has been made in view of such points, and it is an object of the present invention to provide an impulse response noise removal method, an apparatus thereof, and a program thereof that can remove no-signal noise included in a measured impulse response. .

  In order to solve the above-mentioned problem, the impulse response noise removing method according to claim 1 is an impulse response noise removing method for removing no-signal noise included in a no-response section of an impulse response, comprising: a no-signal noise detection step; A probability density function calculation step, a confidence interval calculation step, an average calculation step, a no-response interval detection step, and a no-response interval removal step are included.

  The impulse response noise removing method for performing such a procedure is based on a response in a state where no impulse is applied, which is measured in advance in a measurement space in which the impulse response is measured by the no-signal noise detection means in the no-signal noise detection step. Detect no-signal noise in the measurement space. Thereby, the impulse response noise removing method can acquire a signal of no signal noise such as background noise and background noise in the measurement space. Next, in the impulse response noise removal method, in the probability density function calculation step, the probability density function calculation means calculates the no-signal noise from the average and variance of the amplitude in the no-signal noise distribution detected in the no-signal noise detection step. Calculate the probability density function. Next, in the impulse response noise removal method, in the confidence interval calculation step, the confidence interval calculation means estimates that no-signal noise is included in the measurement space from the probability density function calculated in the probability density function calculation step. An amplitude range indicating a predetermined confidence interval is calculated.

  Next, the impulse response noise removing method calculates an average of a plurality of impulse responses measured in advance in the average calculation step. Thereby, the impulse response noise removal method can suppress the no-signal noise by the averaging process even if the impulse response includes a specific value of the no-signal noise. Next, the impulse response noise removing method is the non-response interval detection step, wherein the non-response interval detection means uses the amplitude calculated in the confidence interval calculation step among the average amplitudes of the impulse responses calculated in the average calculation step. An amplitude included in the range is detected, and a section within the average of impulse responses including the amplitude is detected as a no-response section.

  And the impulse response noise removal method removes the signal of the non-response section detected in the non-response section detection step from the impulse response measured in advance by the non-response section removal means in the non-response section removal step. As a result, the impulse response noise removal method applies the no-response section to the impulse response as it is without averaging in the no-signal noise removal processing, and therefore smoothes the signal portion other than the no-signal noise due to averaging. It is possible to appropriately eliminate non-signal noise while preserving the characteristics of the system.

  The impulse response noise elimination method according to claim 2 is the impulse response noise elimination method according to claim 1, wherein the confidence interval calculation step is calculated by the confidence interval calculation means in the probability density function calculation step. An amplitude range indicating a predetermined confidence interval is calculated on the basis of the average of the amplitudes. As a result, the impulse response noise removal method can calculate a confidence interval that stochastically indicates the amplitude range in which no-signal noise is included in the measurement space with reference to the average of the amplitude in the distribution of no-signal noise.

  In order to solve the above-mentioned problem, an impulse response noise removing apparatus according to claim 3 is an impulse response noise removing apparatus for removing no-signal noise included in a no-response section of an impulse response, comprising: no-signal noise detecting means; Probability density function calculating means, confidence interval calculating means, average calculating means, no response interval detecting means, and no response interval removing means are provided.

  The impulse response noise removing apparatus having such a configuration is obtained by detecting no signal noise in the measurement space from a response in a state where no impulse is applied, which is measured in advance in the measurement space in which the impulse response is measured by the no signal noise detection unit. Is detected. Thereby, the impulse response noise removing apparatus can acquire a signal of no signal noise such as background noise and background noise in the measurement space. Further, the impulse response noise removing device calculates the probability density function of no signal noise from the average and variance of the amplitude in the distribution of no signal noise detected by the no signal noise detection means by the probability density function calculation means. Then, the impulse response noise removing device calculates a predetermined confidence interval estimated by the confidence interval calculation means to include no signal noise in the measurement space from the probability density function calculated by the probability density function calculation means. The amplitude range shown is calculated.

  In addition, the impulse response noise removing device calculates an average of a plurality of impulse responses measured in advance by an average calculating unit. Thereby, the impulse response noise removing device can suppress the no-signal noise by the averaging process even when the impulse response includes a no-signal noise having a specific value. Further, the impulse response noise removing device detects an amplitude included in the range of amplitudes calculated by the confidence interval calculation unit among the average amplitudes of the impulse responses calculated by the average calculation unit by the non-response interval detection unit. A section within the average of the impulse response including the amplitude is detected as a non-response section.

  Then, the impulse response noise removing device removes the signal in the non-response section detected by the non-response section detecting means from the impulse response measured in advance by the non-response section removing means. As a result, the impulse response noise removing apparatus applies a no-response section to an intact impulse response that is not subjected to averaging in the no-signal noise removal processing, and therefore smoothes signal portions other than no-signal noise due to averaging. It is possible to appropriately eliminate non-signal noise while preserving the characteristics of the system.

  In order to solve the above-mentioned problem, an impulse response noise elimination program according to claim 4 is a computer program for removing no-signal noise included in a no-response section of an impulse response. Means, confidence interval calculation means, average calculation means, no response interval detection means, and no response interval removal means.

  The impulse response noise elimination program having such a configuration is a non-signal noise in the measurement space from a response in a state where no impulse is applied, which is measured in advance in the measurement space in which the impulse response is measured by the no-signal noise detection unit. Is detected. Thereby, the impulse response noise removal program can acquire a signal of no signal noise such as background noise and background noise in the measurement space. Further, the impulse response noise removal program calculates a probability density function of no signal noise from the average and variance of amplitudes in the distribution of no signal noise detected by the no signal noise detection means by the probability density function calculation means. The impulse response noise removal program calculates a predetermined confidence interval, which is estimated by the confidence interval calculation means to include no signal noise in the measurement space from the probability density function calculated by the probability density function calculation means. The amplitude range shown is calculated.

  Further, the impulse response noise removal program calculates an average of a plurality of impulse responses measured in advance by an average calculation means. As a result, the impulse response noise removal program can suppress the no-signal noise by the averaging process even when the impulse response includes a specific value of the no-signal noise. The impulse response noise removal program detects the amplitude included in the range of the amplitude calculated by the confidence interval calculation means from the average amplitude of the impulse response calculated by the average calculation means by the non-response interval detection means. A section within the average of the impulse response including the amplitude is detected as a non-response section.

  Then, the impulse response noise removal program removes the non-response section detected by the non-response section detection means from the impulse response measured in advance by the non-response section removal means. As a result, the impulse response noise removal program applies the no-response section to the impulse response as it is without averaging in the no-signal noise removal processing, and therefore smoothes signal portions other than no-signal noise due to averaging. It is possible to appropriately eliminate non-signal noise while preserving the characteristics of the system.

  According to the first, third, and fourth aspects of the invention, when an impulse response is measured, a no-response section included in the response is detected, and no-signal noise (measurement target) mixed in the no-response section is detected. Noise in the absence of the signal) can be removed using the statistical nature of no-signal noise.

  According to the invention according to claim 2, the confidence interval of the no-signal noise calculated from the amplitude distribution of the no-signal noise before averaging with respect to the impulse response in which the no-signal noise having a specific value is suppressed by averaging. By applying, it is possible to detect the non-response interval more accurately as compared with the case where the confidence interval is applied to the impulse response as it is without averaging.

It is a block diagram which shows the whole structure of the impulse response noise removal apparatus which concerns on embodiment of this invention. It is the schematic for demonstrating the process of the probability density function calculation means and confidence interval calculation means of the impulse response noise removal apparatus which concerns on embodiment of this invention. It is the schematic for demonstrating the process of the average calculation means of the impulse response noise removal apparatus which concerns on embodiment of this invention. It is the schematic for demonstrating the process of the non-response area detection means of the impulse response noise removal apparatus which concerns on embodiment of this invention. It is the schematic for demonstrating the process of the non-response area removal means of the impulse response noise removal apparatus which concerns on embodiment of this invention. It is the schematic which shows an example (when replaced with amplitude 0) of the impulse response which removed the non-signal noise by the impulse response noise removal apparatus which concerns on embodiment of this invention. It is the schematic which shows the other example (in the case of truncation) of the impulse response which removed the no signal noise by the impulse response noise removal apparatus which concerns on embodiment of this invention. It is a flowchart which shows the operation | movement (impulse response noise removal method) of the impulse response noise removal apparatus which concerns on embodiment of this invention.

  An impulse response noise removing method and apparatus and program therefor according to an embodiment of the present invention will be described with reference to the drawings. In the following description, the impulse response noise removing device is first explained, and then the impulse response noise removing method and the impulse response noise removing program are explained. In the following description, the same configuration is given the same name and symbol, and detailed description is omitted.

[Impulse response noise elimination device]
The impulse response noise removing apparatus 1 removes no-signal noise from the measured impulse response. Specifically, as shown in FIG. 1, the impulse response noise removing apparatus 1 receives an impulse response before noise removal and a no-signal response from the outside, and in a no-response section statistically obtained from the no-signal response. Based on this, the no-signal noise is removed from the impulse response, and the impulse response after the noise removal is output to the outside. The type of impulse response input to the impulse response noise removing apparatus 1 is not particularly limited. For example, the head impulse response or the indoor impulse response described above is input.

  Here, the above-mentioned “no-signal noise” refers to noise generated in a state where a measurement signal (impulse) is not input in a system that is an impulse response measurement target, that is, an impulse response measurement space. It is shown that. The cause of the no-signal noise includes, for example, stationary noise such as an operation sound of an air conditioner inside and outside the measurement space, and sudden noise generated due to a synchronization deviation or the like in the measurement apparatus itself. Further, the above-mentioned “no signal response” indicates a response measured in a state where no impulse is applied in the measurement space. This no-signal response can be measured by collecting sound in a state where no measurement object is installed in the measurement space and no impulse is applied to the measurement object.

  Here, as shown in FIG. 1, the impulse response noise removing apparatus 1 includes a no-signal noise detection unit 10, a probability density function calculation unit 20, a confidence interval calculation unit 30, an average calculation unit 40, and a no-response interval detection. Means 50 and non-response section removing means 60 are provided. Hereinafter, each component of the impulse response noise removing apparatus 1 will be described.

  The no-signal noise detecting means 10 detects no-signal noise included in the no-signal response. As shown in FIG. 1, the no-signal noise detection means 10 receives a no-signal response measured in advance in a measurement space of an impulse response from a measurement device (not shown). Here, the no-signal response is a response that originally does not include any signal. Therefore, the no-signal noise detection means 10 detects the signal included in the no-signal response as it is as no-signal noise and outputs it to the probability density function calculation means 20 as shown in FIG. By such processing, the impulse response noise removing apparatus 1 can acquire a signal of no signal noise such as background noise and background noise in the measurement space.

The probability density function calculating means 20 calculates a probability density function of no signal noise. As shown in FIG. 1, no-signal noise is input to the probability density function calculating unit 20 from the no-signal noise detecting unit 10. Then, as shown in FIG. 2, the probability density function calculating means 20 calculates a probability density function of no signal noise corresponding to the distribution from the distribution of amplitude of the no signal noise, and distributes the amplitude of the no signal noise. And the probability density function are output to the confidence interval calculation means 30. Specifically, the probability density function calculating means 20 first calculates the average amplitude μ and variance σ ² in the distribution of no-signal noise, as shown in the following equation (1), and then calculates the average μ of the amplitude. Then, a probability density function f (x) of no-signal noise is calculated from the variance σ ² .

  The confidence interval calculation means 30 calculates a confidence interval of the amplitude of the no-signal noise. The “confidence interval” indicates a range of amplitude estimated to include no-signal noise in the measurement space. In other words, the confidence interval stochastically indicates an amplitude range including no-signal noise in the measurement space, and is represented by an amplitude range centered on an average of amplitudes in the probability density function as shown in FIG. It is.

  The confidence interval is usually expressed as “N% confidence interval” together with a confidence level N indicating the reliability of the confidence interval. Here, for example, “99% confidence interval” in the present invention indicates “a range of amplitude in which no signal noise is included with a probability of 99% in the measurement space”. In general, a value such as 95% or 99% is often used as the confidence level N, but this value is arbitrarily set according to the accuracy required for the impulse response model.

  The confidence interval calculation means 30 uses the amplitude distribution of no-signal noise as a population distribution, and the amplitude estimated to contain no-signal noise on the basis of the average amplitude in the probability density function as shown in FIG. Calculate the upper and lower limits. That is, as shown in FIG. 1, the confidence interval calculation means 30 receives the amplitude distribution of the no-signal noise and the probability density function from the probability density function calculation means 20. Then, the confidence interval calculation means 30 adds and subtracts a value obtained by multiplying the sample standard error by a t value predetermined for each confidence level N with respect to the sample average, as shown in the following equation (2). Thus, the confidence interval shown in FIG. 2 is calculated and output to the no-response interval detection means 50.

  N% confidence interval = sample mean ± t × sample standard error (2)

  Here, as described above, an arbitrarily set value of the confidence level N is used for N% in the above-described equation (2). In addition, the value of “t distribution table” determined in advance for each confidence level N is used as “t” in the above equation (2). In addition, “sample average” in the above equation (2) indicates a sample average of the amplitude of no-signal noise. As the sample average, specifically, the average μ of the amplitude in the distribution of no signal noise described above is used.

  In addition, “sample standard error” in the above equation (2) indicates the standard deviation of the sample mean, and specifically indicates the square root of the unbiased variance of no signal noise divided by the number of samples. ing. Unbiased variance is the sum of squares of the values obtained by subtracting the sample mean from each sample divided by (number of samples minus 1).

  The average calculating means 40 calculates the average of impulse responses. As shown in FIG. 1, a plurality of impulse responses before noise removal measured in advance in an impulse response measurement space are input to the average calculating means 40 as shown in FIG. Note that the measurement device used in the measurement of the impulse response is the same as the measurement device used in the measurement of the no-signal response described above and has the same sensitivity as the microphone in the acoustic measurement. Then, as shown in FIG. 3, the average calculating means 40 calculates the average of the impulse responses by adding and averaging the impulse responses, and outputs this to the no-response section detecting means 50.

  Here, FIG. 3 schematically shows the impulse response before noise removal before and after averaging in relation to the time axis and the amplitude. In the schematic diagram of the impulse response in FIG. 3, the horizontal axis represents time, and the length of the straight line extending in the vertical direction indicates the magnitude of the amplitude.

  The three impulse responses before averaging shown on the left side of FIG. 3 include no-signal noise having an irregular amplitude after the response start and response convergence. That is, the first, seventh to twelfth amplitudes from the left in the top impulse response before averaging, and the first, seventh to twelfth amplitudes from the left in the middle impulse response before averaging, The first to twelfth to twelfth amplitudes from the left in the bottom impulse response schematically represent the amplitude of no-signal noise.

  On the other hand, since no-signal noise can assume an irregular and normal distribution, when these impulse responses are added and averaged by the above-described average calculating means 40, as shown on the right side in FIG. The irregular amplitude included later is suppressed to some extent and approaches the average value (= 0). Therefore, it becomes easy to detect a silent section including no signal noise in the subsequent silent section detection process. By such processing, the impulse response noise removing apparatus 1 can suppress the no-signal noise by the averaging process even when the impulse response includes a non-signal noise having a specific value.

  The non-response section detecting means 50 detects a non-response section included in the impulse response. As shown in FIG. 1, the no-response interval detection means 50 receives the confidence interval from the confidence interval calculation means 30 and the average impulse response from the average calculation means 40. Next, as shown in FIG. 4, the no-response interval detection unit 50 has the amplitude indicated by the confidence interval calculated by the confidence interval calculation unit 30 among the average amplitudes of the impulse responses calculated by the average calculation unit 40. An amplitude included in the range is detected, and a section within the average of impulse responses including the amplitude is detected as a no-response section. Then, the no-response section detecting unit 50 outputs the no-response section, that is, a section estimated to contain no-signal noise to the no-response section removing unit 60.

  The no-response section removing means 60 removes no-signal noise by removing the signal in the no-response section in the impulse response. As shown in FIG. 1, the impulse response before averaging is input to the non-response section removing unit 60 from a measuring device (not shown), and the non-response section detecting unit 50 receives the non-response section. The above-mentioned “impulse response before averaging” indicates any one of a plurality of impulse responses measured under the same conditions by a measuring device (not shown). That is, for example, one of the three impulse responses (see FIG. 3) input to the average calculating unit 40 is input to the no-response section removing unit 60. Then, as shown in FIG. 5, the no-response section removing means 60 considers the signal included in the response section as the no-signal noise in the impulse response before averaging, removes the impulse response after the noise removal to the outside. Output. By such processing, the impulse response noise removing apparatus 1 applies the no-response section to the impulse response as it is without averaging in the no-signal noise removal processing. The smoothing of the signal portion can be avoided, and the no-signal noise can be appropriately removed while preserving the characteristics of the system.

  Here, the specific method of noise removal differs depending on the application of the impulse response after noise removal. For example, when the impulse response after noise removal is used as it is as an impulse response model, the no-response interval removing unit 60 replaces the amplitude of no-signal noise included in the no-response interval with 0, as shown in FIG. . Thereby, the delay part at the beginning of the impulse response can be included in the impulse response.

  In addition, when the non-response section removing unit 60 obtains an impulse response in a certain space, for example, when a reverse system that cancels the impulse response is obtained, it is included in the non-response section as shown in FIG. The amplitude of the no-signal noise is replaced with 0, and the delay portion at the beginning of the impulse response and the no-response section after the impulse response convergence are truncated. As a result, it is possible to efficiently obtain the inverse system by omitting a useless delay portion at the beginning of the impulse response.

  The impulse response noise removing apparatus 1 having the above-described configuration detects a no-response section included in the response when measuring the impulse response, and detects no-signal noise (a signal to be measured is included in the no-response section). Noise in the absence state) can be removed using the statistical nature of no-signal noise. Further, the impulse response noise removing apparatus 1 is configured to provide a confidence interval of no-signal noise calculated from an amplitude distribution of the no-signal noise before averaging, with respect to an impulse response in which the no-signal noise having a specific value is suppressed by averaging. By applying, it is possible to detect the non-response interval more accurately as compared with the case where the confidence interval is applied to the impulse response as it is without averaging.

  As described above, since the impulse response noise removing apparatus 1 can remove the no-signal noise included in the impulse response, the measured impulse response is used for an application in which, for example, slight mixing of no-signal noise has a large effect. Even if it is a case, a stable process can be performed.

[Impulse response noise elimination method]
Hereinafter, an operation of the impulse response noise removing apparatus, that is, an impulse response noise removing method will be described with reference to FIG. 8 (refer to FIG. 1 as appropriate). The impulse response noise elimination method includes a no-signal noise detection step, a probability density function calculation step, a confidence interval calculation step, an average calculation means, a no-response interval detection step, and a no-response interval elimination step. These steps are performed in order.

  In the impulse response noise removing method, first, in the no-signal noise detection step, the no-signal noise detection means 10 detects no-signal noise included in the no-signal response (step S1). Next, in the impulse response noise removing method, in the probability density function calculating step, the probability density function calculating means 20 calculates the probability density function of no signal noise using the above-described equation (1) (step S2). Next, in the impulse response noise removing method, in the confidence interval calculation step, the confidence interval calculation means 30 calculates the confidence interval of the amplitude of the no-signal noise using the above-described equation (2) (step S3).

  Next, in the impulse response noise removal method, the average calculation unit 40 calculates the average of a plurality of impulse responses in the average calculation step (step S4). Next, in the impulse response noise removing method, in the no-response interval detection step, the no-response interval detection means 50 determines the average amplitude of the impulse response within the confidence interval as the amplitude of the no-signal noise. Assuming that there is a section, the section including the amplitude is detected as a non-response section (step S5). In the impulse response noise removing method, in the no-response interval removing step, the no-response interval removing means 60 removes the signal included in the response interval as no-signal noise in the impulse response before averaging (step S6). ), The process is terminated. The impulse response noise removing method removes no-signal noise from the impulse response by the procedure as described above.

[Impulse response noise elimination program]
The impulse response noise removing apparatus 1 described above can be realized by operating a general computer with a program that functions as each of the above-described units and units. This program can be distributed via a communication line, or can be written on a recording medium such as a CD-ROM for distribution.

  In other words, the impulse response noise removal program uses a computer to remove the no-signal noise included in the no-response section of the impulse response, the no-signal noise detection means 10, the probability density function calculation means 20, and the confidence interval calculation means. 30 may be configured to function as the average calculating unit 40, the no-response section detecting unit 50, and the no-response section removing unit 60.

  As mentioned above, the impulse response noise removing method and the apparatus thereof and the program thereof according to the present invention have been specifically described by the mode for carrying out the invention, but the gist of the present invention is not limited to these descriptions. It should be construed broadly based on the claims. Needless to say, various changes and modifications based on these descriptions are also included in the spirit of the present invention.

DESCRIPTION OF SYMBOLS 1 Impulse response noise removal apparatus 10 No signal noise detection means 20 Probability density function calculation means 30 Confidence interval calculation means 40 Average calculation means 50 No response area detection means 60 No response area removal means

Claims (4)

An impulse response noise removal method for removing no-signal noise included in a non-response section of an impulse response,
A no-signal noise detection step for detecting no-signal noise in the measurement space from a response in a state where no impulse is applied, which is measured in advance in the measurement space for measuring the impulse response by the no-signal noise detection means,
A probability density function calculating step of calculating a probability density function of the no-signal noise from an average and variance of amplitudes in the distribution of the no-signal noise detected in the no-signal noise detection step by a probability density function calculating means;
A confidence interval calculation means calculates an amplitude range indicating a predetermined confidence interval, which is estimated to include the no-signal noise in the measurement space, from the probability density function calculated in the probability density function calculation step. A confidence interval calculation step,
An average calculating step of calculating an average of a plurality of impulse responses measured in advance by an average calculating means;
The non-response section detecting means detects an amplitude included in the range of the amplitude calculated in the confidence section calculation step from the average amplitude of the impulse response calculated in the average calculation step, and the amplitude is included. A no-response interval detection step of detecting an interval within the average of the impulse response as the no-response interval;
A no-response interval removing step of removing the no-response interval detected in the no-response interval detection step from the impulse response measured in advance by the no-response interval removing means,
Including an impulse response noise elimination method.   In the confidence interval calculation step, the confidence interval calculation means calculates an amplitude range indicating a predetermined confidence interval based on an average of amplitudes in the probability density function calculated in the probability density function calculation step. The impulse response noise removing method according to claim 1. An impulse response noise removing device that removes no-signal noise included in a no-response section of an impulse response,
A no-signal noise detection means for detecting no-signal noise in the measurement space from a response in a state where no impulse is applied, which is measured in advance in the measurement space for measuring the impulse response,
A probability density function calculating means for calculating a probability density function of the no signal noise from the mean and variance of the amplitude in the distribution of the no signal noise detected by the no signal noise detecting means;
A confidence interval calculation means for calculating a range of amplitude indicating a predetermined confidence interval, which is estimated from the probability density function calculated by the probability density function calculation means to include the no-signal noise in the measurement space; ,
An average calculating means for calculating an average of a plurality of impulse responses measured in advance;
Among the average amplitudes of the impulse responses calculated by the average calculation means, an amplitude included in the amplitude range calculated by the confidence interval calculation means is detected, and an interval within the average of the impulse responses including the amplitude is detected. Non-response section detecting means for detecting as the non-response section,
A non-response section removing means for removing the non-response section detected by the non-response section detecting means from the impulse response measured in advance;
An impulse response noise removing apparatus comprising: In order to remove the no-signal noise included in the no-response section of the impulse response,
A no-signal noise detection means for detecting no-signal noise in the measurement space from a response measured in advance in the measurement space for measuring the impulse response, in a state where no impulse is applied;
A probability density function calculating means for calculating the probability density function of the no-signal noise from the mean and variance of the amplitude in the distribution of the no-signal noise detected by the no-signal noise detecting means;
A confidence interval calculation means for calculating a range of amplitude indicating a predetermined confidence interval, which is estimated from the probability density function calculated by the probability density function calculation means to include the no-signal noise in the measurement space;
An average calculating means for calculating an average of a plurality of impulse responses measured in advance;
Among the average amplitudes of the impulse responses calculated by the average calculation means, an amplitude included in the amplitude range calculated by the confidence interval calculation means is detected, and an interval within the average of the impulse responses including the amplitude is detected. A no-response section detecting means for detecting as a no-response section,
A non-response section removing means for removing the non-response section detected by the non-response section detecting means from the impulse response measured in advance;
Impulse response noise removal program to function as.

Images