KR20090056597A - Method and apparatus for calibrating the sound source signal acquired through the microphone array - Google Patents

Abstract

A method and an apparatus for correcting a sound source signal are provided to prevent a distortion of a sound source signal due to characteristic mismatching between individual microphones by correcting the sound source signal according to the calculated reference probability distribution. A probability distribution calculator calculates the probability distributions expressing the number of the sound source signals existing in each interval according to the size interval of the sound source signals. A probability distribution calculator includes an accumulator. The accumulator calculates and accumulates the probability distributions about the size of the sound source signals in each interval. A reference probability distribution calculator(211) calculates the reference probability distribution representing the probability distributions based on the calculated probability distributions. A signal corrector(212) corrects the sound source signals according to the calculated reference probability distribution. The reference probability distribution calculator calculates the reference probability distribution based on the accumulated probability distributions.

Description

Method and apparatus for calibrating the sound source signal acquired through the microphone array}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for correcting a sound source signal acquired through a microphone array. The present invention relates to a method and apparatus for correcting a characteristic difference of a sound source signal obtained in acquiring sound such as voice call, recording, and voice recognition in a digital information device such as a teleconference system.

The age has come to make it possible to make phone calls using mobile phones or to acquire sound through digital camcorders and recorders. Various digital devices, such as CE (consumer electronics) devices and mobile phones, use a microphone as a means for acquiring sound, and stereo sound using two or more channels instead of a single channel mono sound. In general, a microphone array including a plurality of microphones is used to implement the microphone.

The microphone array can combine multiple microphones to obtain additional properties regarding directivity such as the direction or position of the sound to be acquired as well as the sound itself. Directivity refers to increasing the sensitivity to the sound source signal emitted from the sound source located in a specific direction by using the time difference that the sound source signal reaches each of the plurality of microphones constituting the array. Thus, by acquiring sound source signals using such a microphone array, it is possible to emphasize or suppress the sound source signal input from a specific direction.

Adjusting the microphone array means adjusting the directional adjustment parameters such as the delay value of each of the microphones constituting the microphone array or the spacing between the microphones. In addition, in order to adjust the microphone array as intended by the user, a basic precondition that the microphone array must operate according to the physical characteristics of the individual microphones (meaning signal magnitude, phase, and frequency response, etc.) must be established.

The technical problem to be solved by the present invention is a sound source signal correction to solve the problem that the sound source signal is distorted due to the characteristic mismatch between the individual microphones constituting the microphone array, the sound source signals input through the microphone array in a digital device capable of sound acquisition To provide a method and apparatus.

In order to achieve the above technical problem, the sound source signal correction method according to the present invention comprises the steps of calculating the probability distributions expressed as the number of sound source signals present in each section for each of the magnitude section of the sound source signals obtained through the microphone array; Calculating a reference probability distribution representative of the calculated probability distributions; And correcting the sound source signals according to the calculated reference probability distribution.

In order to solve the other technical problem, the present invention provides a computer-readable recording medium recording a program for executing the sound source signal correction method described above on a computer.

In order to achieve the above technical problem, the sound source signal correction apparatus according to the present invention calculates a probability distribution that calculates probability distributions expressing the number of sound source signals present in each section as a probability for each of the magnitude sections of the sound source signals obtained through the microphone array. part; A reference probability distribution calculating unit for calculating a reference probability distribution representing the calculated probability distributions; And a signal correction unit for correcting the sound source signals according to the calculated reference probability distribution.

Hereinafter, various embodiments of the present invention will be described in detail with reference to the accompanying drawings. In describing the embodiments, the sound source will be used as a term meaning a source from which sound is emitted. In addition, sound pressure refers to a force exerted by sound energy using a physical quantity of pressure, and a sound pressure field conceptually expresses an area in which sound pressure affects a sound source.

1 is a block diagram illustrating a schematic idea of correcting a sound source signal in a problem situation to be solved by the present invention, and includes a microphone array 100, a sound source signal corrector 110, and a signal processor 120. Before describing each component, a problem situation to be solved by the present invention will be described.

As described above, microphone arrays are used to take advantage of the directional characteristics of sound, such as directivity. In general, the microphone array improves the amplitude by appropriately weighting each signal received in the microphone array in order to receive a target signal mixed with background noise with high sensitivity, thereby reducing noise when the desired target signal and the interference noise signal are different in direction. It serves as a filter that can be spatially reduced. This type of spatial filter is called a beamformer.

However, in order to process a sound source signal using such a beam former, the physical characteristics of each individual microphone constituting the microphone array must match those known to the user. That is, the magnitudes of the signals, phase and frequency responses of the signals acquired through the individual microphones must all match. This is because the factors for adjusting the microphone array are based on the physical characteristics of these individual microphones.

If the sound source signal is delayed or the size of the sound source signal is changed for a certain time to adjust the microphone array and the individual microphones constituting the microphone array are not adjusted according to the user's intention, the result is different from the user's intention. You will have no choice but to obtain a distorted sound source signal. For example, the individual microphones that make up the microphone array when trying to remove the unwanted ambient interference sound field side-lobe from the typical adaptive beamformer algorithm, generalized side-lobe canceller (GSC). If there is a mismatch between the features, this mismatch will cause signal leakage, resulting in a distortion of the sound source signal.

The characteristic mismatch between the microphones described above is largely due to the following two causes. First, there is a case due to an error generated during the manufacturing process of the microphone, and second, the physical characteristics of the microphone may change due to aging as time passes in the process of using the microphone. Both of these causes originate from the actual product manufacturing or use process, and inevitably occur in many cases. Accordingly, various embodiments of the present invention described below attempt to correct a sound source signal so that the sound acquisition device is insensitive to characteristic mismatch between microphones when these problems occur in a sound acquisition device having a microphone array. Looking at the schematic configuration of Figure 1 under such a problem situation is as follows.

The microphone array 100 obtains a sound source signal from the outside. The method of adjusting the microphone array 100 such as the direction of the sound source or the size of the sound source signal may be variously designed according to the situation and the purpose of implementing the embodiment of the present invention.

The sound source signal correcting unit 110 corrects characteristic mismatches of sound source signals obtained through the microphone array 100. The following two methods can be used to solve the distortion problem of the sound source signal caused by the characteristic mismatch between the individual microphones. First, when a sound source signal is input through the microphone array, there may be a method of correcting the respective sound source signals immediately input. Second, the degree of distortion in the process of processing the sound source signal input through the microphone array according to a specific purpose is There may be a way to calibrate. According to various embodiments of the present disclosure, the former, namely, a method of directly correcting the input signals itself, may be used. A detailed description follows from FIG. 2.

The signal processor 120 processes sound source signals corrected by the sound source signal corrector 110 according to a user's purpose. This may be freely designed according to various embodiments and environments in which the present invention will be implemented, such as conventional sound source signal processing, beam formers, and sound source position trackers.

2 is a block diagram showing a sound source signal correction apparatus according to an embodiment of the present invention, the microphone array 200, the reference probability distribution calculator 211, the signal correction unit 212 and the signal processor 220 Include. The microphone array 200 and the signal processor 220 have the same configuration as the microphone array 100 and the signal processor 120 described with reference to FIG. 1, and correspond to the configuration of a conventional sound source signal processing apparatus. Therefore, in a more strict sense, the sound source signal correcting apparatus according to the present embodiment includes a reference probability distribution calculating unit 211 and a signal correcting unit 212 included in the region 210 shown by a dotted line. Hereinafter, the sound source signal correction apparatus will be described in detail with respect to these two components.

The reference probability distribution calculator 211 calculates probability distributions expressing the number of sound source signals present in each section as a probability for each of the magnitude sections of the sound source signals acquired through the microphone array 200. In general, the probability distribution refers to a distribution state of random variables, and the probability values are mapped to events that may occur in a trial. In this case, the distribution state of the random variable may be calculated by counting the number of variables included in the interval obtained by dividing the variable by a constant width.

In the embodiments of the present invention, a probability distribution function (PDF) means dividing the magnitudes of sound source signals into predetermined sections and counting the number of sound source signals corresponding to each section and expressing them as probability. Typically, a graph that divides the variance into arbitrary intervals and displays the number of values (also called frequency) included in each interval is called a histogram, which will be used as a graph that visually displays the probability distribution. will be.

The magnitudes of the sound source signals, which are the basis for calculating the probability distribution, can be expressed in various ways, typically the magnitude of the signal in the time domain or the magnitude of the signal in the frequency domain. Could be.

The magnitude of the input signals in the time domain will be the amplitude of the signals themselves. Therefore, the probability distribution may be calculated by dividing the amplitude of the input signals into a predetermined size section and counting the number of signals included in each section. On the other hand, to obtain the magnitude of the input signal in the frequency domain is as follows.

First, digital signal processing of sound source signals input through the microphone array 200 is converted into a frequency domain through a fast Fourier transform for convenience of operation. In general, digital signal processing uses a convolution to input a signal into a corresponding system and to express a resultant output signal, which is divided into frames to finitely limit a given target signal. . A frame refers to a unit that divides a sound source signal into a predetermined section according to a change in time. Once the sound source signals are converted into the frequency domain frame by frame, the process of calculating the probability distribution is similar to that in the time domain. That is, the probability distribution may be calculated by dividing the magnitude of the signal in the frequency region into a constant magnitude section and counting the number of signals included in each section.

The reference probability distribution calculating unit 211 first calculates the distribution degree of the sound source signal for each of the above sections as a probability, and calculates a reference value representing the respective sound source signals from the calculated results. This reference value is a reference probability distribution, which is a measure for correcting the respective sound source signals in a later step.

Hereinafter, the process of calculating the reference probability distribution from the input sound source signals by the reference probability distribution calculator 211 will be described in more detail with reference to FIGS. 4, 5A, and 5B. In the following embodiments, the cumulative probability distribution obtained by accumulating the probability distributions from the smaller signal to the larger signal will be used as the mood probability distribution, and a normal cumulative probability distribution that normalizes the cumulative probability distribution to a specific value will be used. As the probability distributions described above, means for representing various probability distributions besides the cumulative probability distribution and the normal cumulative probability distribution may be used. Hereinafter, each specific meaning and configuration will be described.

4 is a diagram illustrating a method of calculating a cumulative probability distribution in a sound source signal correction apparatus according to an embodiment of the present invention. The first graph illustrates a histogram of a probability distribution of sound source signals input through a microphone array. The second graph illustrates a histogram that accumulates and normalizes the probability distribution of sound source signals. The horizontal axis of the graph represents a magnitude section of the sound source signal, and the vertical axis represents the number of sound source signals corresponding to each section as a probability distribution. In FIG. 4, it is assumed that the microphone array is composed of four microphones, and sound source signals input through the respective microphones are represented by four channels.

In the left graph of FIG. 4, although the input sound source signal is the same, it can be seen that the difference between the probability distributions for each channel appears. That is, it can be seen that the characteristic mismatch between the microphones appear. If these probability distributions are accumulated in the histogram by accumulating the signal from the smaller to the larger relative to the horizontal axis, a graph in which the signal magnitude increases as shown in FIG. 4 will be drawn. However, the cumulative probability distributions thus drawn do not coincide with each other due to the characteristic mismatch between the microphones described above. In other words, the cumulative results of the probability distributions are different. Therefore, a normalization process is performed to match these cumulative results. Here, normalization means that the maximum value is matched to a specific value that is a reference for easy calculation. In this embodiment, the maximum value of the accumulated probability distribution is set to 1 and the minimum value is set to 0. The histogram generated through the above process is shown in the second graph of FIG. 4. The second graph of FIG. 4 is a histogram showing a normalized cumulative probability distribution function, in which the vertical axes are all normalized to 1, and the probability distributions of the four channels gradually increase as the right proceeds. To form a graph.

5A and 5B are detailed block diagrams illustrating a configuration of calculating a reference cumulative probability distribution in the sound source signal correcting apparatus according to an embodiment of the present invention, and show a specific configuration for generating the histogram of FIG. .

In FIG. 5A, the reference cumulative probability distribution calculator includes a normal cumulative probability distribution calculator 510 and a representative value calculator 520.

As described above, the normal cumulative probability distribution calculator 510 calculates the normal cumulative probability distributions corresponding to the sound source signals from the signal distributions of the magnitude intervals of the sound source signals received through the microphone array. Referring to FIG. 5B, the normal cumulative probability distribution calculator 510 may be configured by an accumulator 511 and a normalizer 512. The accumulator 511 calculates and accumulates probability distributions of the magnitudes of the signals for each of the magnitude intervals of the sound source signals. The normalizer 512 normalizes the accumulated maximum values so that the accumulated maximum values coincide with each other.

The representative value calculator 520 calculates a representative value from the normal cumulative probability distributions calculated by the normal cumulative probability distribution calculator 510 and sets the representative cumulative probability distribution as the reference cumulative probability distribution. When a sound source signal is generated as many as the number of individual microphones constituting the microphone array, and a normal cumulative probability distribution corresponding to the generated sound source signal is calculated, a representative value representing them is calculated and set as a reference for subsequent sound source signal correction. . For example, assuming four separate microphones, four normal cumulative probability distributions will also be calculated, and one reference cumulative probability distribution will be calculated from these four normal cumulative probability distributions according to a specific rule. Here, the specific rule may be implemented in the form of an arbitrary function for calculating the representative value.

As a method of calculating the representative value, various methods of selecting a value representative of arbitrary values are generally available, and these methods can be easily derived by those skilled in the art. It is. Representative examples of the method of calculating the representative value are as follows.

First, the average value can be set as a representative value. That is, the average value of the magnitudes of the signals of the plurality of normal cumulative probability distributions is obtained and set as the reference cumulative probability distribution. Second, it can be set as a representative value by using a median. In this method, the normal cumulative probability distribution corresponding to the median of the signal magnitudes is set as the reference cumulative probability distribution in consideration of the distribution of the sound source signals. In addition to the above methods, a method of selecting one or more sound source signals from among the plurality of sound source signals to give more weight to the selected sound source signals and to calculate a representative value may be used.

The process of calculating the reference cumulative probability distribution in the reference probability distribution calculating unit 211 of FIG. 2 has been described above. Through this process, when a difference occurs between input sound source signals due to characteristic mismatch between a plurality of microphones, a reference cumulative probability distribution that serves as a reference for correcting the sound source signals may be obtained.

Next, the signal corrector 212 corrects sound source signals according to the reference cumulative probability distribution calculated by the reference probability distribution calculator 211. Accordingly, the signal correction unit 212 may exist as many as the number of individual microphones constituting the microphone array, and corrects each sound source signal (meaning a channel) with reference to a reference cumulative probability distribution. Hereinafter, a method of correcting sound source signals in the signal corrector 212 will be described in more detail with reference to FIGS. 6 and 7.

FIG. 6 is a detailed block diagram illustrating a configuration of correcting a sound source signal using a conversion function in the sound source signal correction device according to an embodiment of the present invention. The signal correction unit 620 is again a probability distribution converter 621. ).

First, the normal cumulative probability distribution calculator 610 calculates a normal cumulative probability distribution from sound source signals. This process is as described above with reference to FIG. Next, the probability distribution transformer 621 may correct the sound source signals corrected from the normal cumulative probability distributions calculated by the normal cumulative probability distribution calculator 610 by using a function that converts the sound source signals into the reference cumulative probability distribution. Create That is, the normal cumulative probability distribution is corrected according to the reference value. This correction process is performed through a sample-by-sample comparison process according to a reference cumulative probability distribution. The sample-to-sample comparison refers to converting a specific sample of the sound source signal before correction with a corresponding reference sample (meaning a value according to the reference cumulative probability distribution). Refer to FIG. 7 to help understand the above conversion process.

7 is a diagram illustrating a method of generating a corrected sound source signal in a probability distribution converter of a sound source signal correcting apparatus according to an embodiment of the present invention. 7 is a histogram of a reference cumulative probability distribution, and a graph shown by a dotted line is a histogram of a normal cumulative probability distribution corresponding to a sound source signal that is not currently corrected. The vertical axis of the graph is assumed to be normalized to a value from 0 to 1. Correcting a sound source signal means matching a normal cumulative probability distribution value for a specific sample with a reference cumulative probability distribution value.

For example, when the value of A on the horizontal axis of the graph indicates a specific value of the uncorrected sound source signal (meaning sample), the cumulative probability distribution corresponding to the value of A will be the value of C on the vertical axis. In this case, if the input value corresponding to the cumulative probability distribution is inversely estimated based on the C value, the corrected input signal will be the B value on the horizontal axis of the histogram of the reference cumulative probability distribution. In other words, the B value is a signal obtained by correcting the uncorrected microphone input signal A value.

In the above, the process of correcting the sound source signal by the signal corrector 212 of FIG. 2 has been described with reference to FIGS. 6 and 7. Hereinafter, the process of calculating the reference probability distribution and the process of correcting the sound source signal according to the calculated reference probability distribution in the sound source signal correction device 210 of FIG.

이라고 하면, 이들 마이크로폰 입력 신호들에 대응하는 정규 누적 확률 분포들은

와 같이 표현할 수 있다. 이 때, 정규 누적 확률 분포들은

이라는 조건을 만족한다. 이어서, 마이크로폰 입력 신호 각각의 누적 확률 분포로부터 기준 누적 확률 분포를 산출하면 다음의 수학식 1과 같이 표현된다.First, assuming that the number of individual microphones constituting the microphone array is k, k microphone input signals are obtained.

, The normal cumulative probability distributions corresponding to these microphone input signals

It can be expressed as In this case, the normal cumulative probability distributions

Satisfies the condition Subsequently, when the reference cumulative probability distribution is calculated from the cumulative probability distributions of the microphone input signals, the following equation 1 is expressed.

은 기준 누적 확률 분포에 따른 함수이고, f(·)는 대표값을 산출하는 함수를 의미한다. 앞서 도 5a에서 설명한 바와 같이 f(·)는 평균값이나 중앙값을 구하는 함수가 될 수 있을 것이다.here,

Is a function according to the reference cumulative probability distribution, and f (·) is a function for calculating a representative value. As described above with reference to FIG. 5A, f (·) may be a function for obtaining an average value or a median value.

Next, a process of correcting the microphone input signal to follow the reference cumulative probability distribution is expressed by Equation 2 below.

는 보정되지 않은 마이크로폰 입력 신호이고,

는 보정된 마이크로폰 입력 신호를 의미한다. 또한,

는 정규 누적 확률 분포에 따른 함수이고,

은 기준 누적 확률 분포에 따른 함수를 의미한다. 즉, 수학식 2는 앞서 도 7을 통해 설명한 바와 같이, 특정 입력 신호에 대한 정규 누적 확률 분포 값을 기준 누적 확률 분포 값에 일치시키는 입력 신호 보정 과정을 나타낸다.here,

Is the uncorrected microphone input signal,

Denotes the corrected microphone input signal. Also,

Is a function of the normal cumulative probability distribution,

Denotes a function according to the reference cumulative probability distribution. That is, Equation 2 illustrates an input signal correction process of matching a normal cumulative probability distribution value for a specific input signal with a reference cumulative probability distribution value, as described above with reference to FIG. 7.

Equation 2 is summarized as the following Equation 3 with respect to the corrected microphone input signal.

및

과 같은 확률 분포에 따른 함수들은 일종의 비선형 변환 함수들로서, 수학식 3을 통해 도 2의 신호 보정부(212)는 최종적으로 보정된 음원 신호를 생성하여 신호 처리부(220)에 공급한다.ideal

And

Functions according to probability distributions as described above are a kind of nonlinear transformation functions. The signal correction unit 212 of FIG. 2 generates a finally corrected sound source signal and supplies it to the signal processor 220 through Equation 3.

The signal processor 220 is a component that is typically provided in a digital signal processing device having a microphone array, and thus may be freely designed according to the environment in which the embodiments of the present invention are implemented.

The configuration and role of the sound source signal correction apparatus have been described in detail with reference to various embodiments of the present disclosure. According to the embodiments, the reference probability distribution is calculated based on the probability distributions of the sound source signals input through the microphone array, and the sound source signals are corrected accordingly, thereby making individual microphones constituting the microphone array in the digital device capable of sound source acquisition. The problem that the sound source signal is distorted due to the characteristic mismatch between them is solved. In addition, once the reference probability distribution is calculated, the input sound source signals can be corrected in real time, thereby enabling fast error compensation.

FIG. 3 is a block diagram illustrating a sound source signal correcting apparatus in which a storage unit is added according to another exemplary embodiment of the present invention, and the storage unit 315 is added to the sound source signal correcting apparatus 210 of FIG. 2.

Since the sound source signal correcting apparatus 310 of FIG. 3 includes a reference probability distribution calculating unit 311 and a signal correcting unit 312 as in FIG. 2, the configuration features of the sound source signal correction apparatus 310 will be described below.

The storage unit 315 stores the reference probability distribution calculated by the reference probability distribution calculator 311. As described above, the characteristic mismatch between the microphones is caused by an error in the manufacturing process of the microphone or aging due to the use of the microphones. Therefore, once a characteristic mismatch occurs, it is likely that such a problem will not change significantly in a short time. In other words, the property mismatch itself changes slowly over time. Therefore, once the reference cumulative probability distribution is calculated, it is not necessary to calculate it several times in a short time. Rather, calculating the reference probability every time the sound source signal is corrected will be a waste of system resources.

For this reason, the storage unit 315 stores the reference probability distribution so that it is not necessary to recalculate the calculated probability distribution. The storage unit 315 may be a recording device capable of recording data or a specific storage device connected via a network. Subsequently, the signal correction unit 312 reads the reference probability distribution stored in the storage unit 315 and corrects the sound source signals according to the read reference probability distribution. In FIG. 3, when a predetermined time elapses or when the characteristics of the microphones are changed while the sound source signal correcting apparatus 310 performs the correction, the reference probability distribution calculator 311 may newly calculate the reference probability distribution. By updating the newly calculated reference probability distribution in the storage unit 315, the characteristic mismatch between the changed microphones may be eliminated.

In this embodiment, the reference probability distribution is not repeatedly calculated, and the calculated reference probability distribution is stored and used for the correction of the sound source signal, thereby reducing unnecessary calculations. Since it can be omitted, fast sound source signal correction is possible.

8 is a flowchart illustrating a sound source signal correction method according to another embodiment of the present invention, and includes the following steps.

In step 810, probability distributions expressing the number of sound source signals present in each section as a probability for each of the magnitude sections of the sound source signals acquired through the microphone array are calculated. Probability distributions may be calculated through a normalization process of accumulating probability distributions of sound source signals and matching maximum values.

In operation 820, a reference probability distribution representing each probability distribution is calculated based on the probability distributions calculated in operation 810. The reference probability distribution can be obtained by calculating a representative value such as an average value or a median value from the calculated probability distributions and setting the reference probability distribution.

In operation 830, the sound source signals acquired in operation 810 are corrected according to the reference probability distribution calculated in operation 820. The corrected sound source signals can be obtained by inputting the probability distributions using an inverse function of a function that converts the sound source signals into the reference probability distribution calculated in step 810.

According to this embodiment, the problem that the sound source signal is distorted due to the characteristic mismatch between the individual microphones constituting the microphone array is eliminated, and the correction of the sound source signal is performed in real time according to the calculated reference cumulative probability distribution, thereby enabling fast error correction. Do.

Meanwhile, the present invention can be embodied as computer readable codes on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored.

Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disks, optical data storage devices, and the like, which are also implemented in the form of carrier waves (for example, transmission over the Internet). It includes. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion. And functional programs, codes and code segments for implementing the present invention can be easily inferred by programmers in the art to which the present invention belongs.

In the above, various embodiments of the present invention have been described. Those skilled in the art will understand that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

1 is a block diagram illustrating a schematic idea of correcting a sound source signal in a problem situation to be solved by the present invention.

2 is a block diagram illustrating a sound source signal correction apparatus according to an embodiment of the present invention.

3 is a block diagram illustrating an apparatus for compensating a sound source signal to which a storage unit is added according to another exemplary embodiment of the present invention.

4 is a diagram illustrating a method of calculating a cumulative probability distribution in a sound source signal correction apparatus according to an embodiment of the present invention.

5A and 5B are detailed block diagrams illustrating a configuration of calculating a reference cumulative probability distribution in a sound source signal correction apparatus according to an embodiment of the present invention.

FIG. 6 is a detailed block diagram illustrating a configuration of correcting a sound source signal using a conversion function in the sound source signal correction apparatus according to an embodiment of the present invention.

7 is a diagram illustrating a method of generating a corrected sound source signal in a probability distribution converter of a sound source signal correcting apparatus according to an embodiment of the present invention.

8 is a flowchart illustrating a sound source signal correction method according to another embodiment of the present invention.

Claims (13)

Calculating probability distributions representing the number of sound source signals existing in each section as a probability for each of the magnitude intervals of the sound source signals acquired through the microphone array; Calculating a reference probability distribution representing the probability distributions based on the calculated probability distributions; And And correcting the sound source signals according to the calculated reference probability distribution. The method of claim 1, Computing the probability distributions includes calculating and accumulating probability distributions for the magnitudes of the sound source signals for each section, The calculating of the reference probability distribution may include calculating the reference probability distribution based on the accumulated probability distributions. The method of claim 2, Calculating the probability distributions further comprises normalizing the accumulated probability distributions so that the maximum values of the accumulated probability distributions coincide with a predetermined value, The calculating of the reference probability distribution includes calculating the reference probability distribution based on the normalized cumulative probability distributions. The method of claim 1, The calculating of the reference probability distribution may include calculating an average value or a median value from the calculated probability distributions and setting the reference probability distribution as the reference probability distribution. The method of claim 1, And correcting the sound source signals comprises generating corrected sound source signals by inputting the probability distributions into an inverse function of a function for converting from the sound source signals to the reference probability distribution. The method of claim 1, Storing the calculated reference probability distribution in advance; The correcting of the sound source signals may include reading the prestored reference probability distribution and correcting the sound source signals according to the read reference probability distribution. A non-transitory computer-readable recording medium having recorded thereon a program for executing the method of claim 1. A probability distribution calculator for calculating probability distributions representing, as a probability, the number of sound source signals existing in each section for each of the magnitude sections of the sound source signals acquired through the microphone array; A reference probability distribution calculating unit for calculating a reference probability distribution representing the probability distributions based on the calculated probability distributions; And And a signal correction unit for correcting the sound source signals according to the calculated reference probability distribution. The method of claim 8, The probability distribution calculator includes an accumulator that calculates and accumulates probability distributions for the magnitudes of the sound source signals for each section. And the reference probability distribution calculator calculates the reference probability distribution based on the accumulated probability distributions. The method of claim 9, The probability distribution calculator further includes a normalization unit for normalizing the accumulated probability distributions so that the maximum values of the accumulated probability distributions coincide with a predetermined value. And the reference probability distribution calculator calculates the reference probability distribution based on the normalized cumulative probability distributions. The method of claim 8, And the reference probability distribution calculator calculates an average or median value from the calculated probability distributions and sets the reference probability distribution as the reference probability distribution. The method of claim 8, And the signal correcting unit includes a probability distribution converting unit configured to generate corrected sound source signals by inputting the probability distributions to an inverse function of a function for converting the sound source signals into the reference probability distribution. The method of claim 8, Further comprising a storage unit for storing the calculated reference probability distribution in advance, And the signal correcting unit reads the prestored reference probability distribution and corrects the sound source signals according to the read reference probability distribution.

Images