JP5376173B2 - Radiation pattern estimation method, apparatus and program thereof - Google Patents

Description

  The present invention relates to a method, apparatus, and program for estimating a radiation directivity characteristic of a sound source having directivity.

  A directional sound source, for example, a voice spoken by a human, can be clearly heard when heard in front, but it is difficult to clearly hear the utterance content from the back of the head. This is because a voice uttered by a human has directional characteristics, and a strong sound is produced in the front direction and a weak sound is produced in the occipital region. In addition, the voice that is heard is clearly seen in the front, but it can be heard in the back of the head. This is because it is difficult to hear the high sound range and the low sound range is easy to hear. This phenomenon indicates that the sound directivity has frequency dependency.

If the radiation directivity characteristics of a sound source having such directivity can be accurately grasped, for example, the realistic sensation of a sound source such as computer graphics can be improved. The radiation directivity characteristics of the sound source are the observation signal Y _i (w) observed by a plurality of microphones 2 _i arranged so as to surround the sound source 1 as shown in FIG. Using the source signal S (w) of the sound source, for example, H _i (w) = Y _i (w) / S (w) is obtained. This H _i (w) is a radiation directivity characteristic that does not include a reflection component. Here, i represents the radiation direction, w represents the frequency, and Y, S, and H are generally complex numbers.

  However, since the anechoic chamber used for measurement is expensive, it is preferable that the radiation directivity characteristics can be measured in a normal room. Therefore, Non-Patent Document 1 discloses a method of extracting the radiation characteristic even from a state in which the radiation directivity characteristic of the sound source S (w) is unknown in a room with reverberation. The method is as follows.

First, as shown in FIG. 2, a plurality of microphones are arranged at positions away from the wall of the room, and the sound Y _i (w) radiated from the sound source S (w) is observed. Then, S (w) is estimated from the observed microphone signal Y _i (w) using a dereverberation technique (for example, Non-Patent Document 2). From the estimated S (w) and Y _i (w), a transfer function G _i (w) from the sound source 1 to the microphone 2 _i is obtained as G _i (w) = Y _i (w) / S (w). As shown in FIG. 3, G _i (w) includes the reverberation characteristics of the room and the reflected sound component from the wall. The horizontal axis in FIG. 3 is time, the vertical axis is the amplitude of the microphone signal, and G _i (w) is inverse Fourier transformed into a signal in the time domain, which is called an impulse response. From this impulse response, only the first half L <(P) that is not affected by reverberation or reflection is cut out and subjected to Fourier transform to obtain a radiation directivity H _i (w) that is not affected by reverberation or reflection.

Okamoto et al., `` Fundamental study on sound source radiation pattern extraction using a surrounding microphone '', IEICE Tech., EA200-48 (2009-08) pp.31-36. Okamoto et al., “High accuracy of linear prediction blind reverberation removal by whitening filter of observation signal”, Sound lecture (Spring), pp.675-676, Mar, 2009.

  As a method of measuring the radiation directivity characteristics of a sound source, two conventional methods for measuring in an anechoic room and in a room with reverberation were shown, but both methods tried to find radiation characteristics in all directions. In this case, it is necessary to prepare a large number of microphones and obtain an observation signal, which is troublesome.

  The present invention has been made in view of such problems. A radiation directivity characteristic estimation method, an apparatus, and a program for estimating a radiation directivity characteristic based on observation signals observed with a smaller number of microphones than the conventional method are provided. The purpose is to provide.

  The radiation directivity characteristic estimation method according to the present invention is a radiation directivity characteristic estimation method for estimating the radiation directivity characteristic of a sound source having a perimeter length of Q. The sound collection process, the minimum observation point calculation process, and the interpolation point calculation And a radiation directivity characteristic estimation process. In the sound collection process, an observation signal from the sound source is obtained for each necessary frequency using sound collection signals of L microphones arranged at a predetermined interval around the sound source. In the process of calculating the minimum number of observation points, the number obtained by subtracting 1 from the length Q of the circumference of the sound source divided by a wavelength equal to or less than one-eighth of the wavelength for each frequency is obtained as the minimum number of observation points M for each necessary frequency. . In the interpolation point number calculation process, a value obtained by adding 1 to the value obtained by dividing the number L of microphones by the minimum number of observation points M + 1 is obtained as the number N of observation points used for interpolation for interpolating the radiation directivity characteristics. In the radiation directivity characteristic estimation process, interpolation is performed using observation signals of the number N of observation points used for interpolation to estimate the radiation directivity characteristics at positions between observation points.

  According to the radiation directivity characteristic estimation method of the present invention, the change in the radiation directivity characteristic of a frequency having a wavelength larger than the peripheral length Q of the sound source is gentle in all directions, and the radiation directivity characteristic having a wavelength smaller than the peripheral length Q The radiation directivity characteristics are estimated by changing the number of observation signals used for interpolation according to the frequency. Therefore, there is no need to arrange microphones in all directions of the desired radiation directivity, and the radiation directivity can be accurately estimated with fewer microphones than in the conventional method.

Conventional method, shows an example of the arrangement of the sound source 1 and the microphone 2 _i which measures the radiation directivity characteristic of the sound source 1 in an anechoic chamber. It shows an example of the arrangement of the sound source 1 and the microphone 2 _i of the method in a room with a reverberation disclosed in Non-Patent Document 1 and dereverberation measuring the radiation patterns. The figure which shows the example of the observation signal of the time domain which shows the part in connection with the radiation | emission characteristic of a sound source, and the part in connection with a reflection and reverberation component. It is a figure which shows the radiation directivity characteristic of a different frequency, (a) shows the example of the radiation directivity characteristic of a low frequency, (b) is a figure which shows the example of the radiation directivity characteristic of a high frequency. The figure which shows a mode that the amplitude is changing linearly in the range below 1/8 of the wavelength of the waveform of a sine wave. The figure which shows the function structural example of the radiation directivity characteristic estimation apparatus 100 of this invention. The figure which shows the operation | movement flow of the radiation directivity characteristic estimation apparatus 100. When measuring the directional characteristics of radiation source 1 with 16 microphones, shows an example of the arrangement of the sound source 1 and the microphone 2 _i. The figure which shows the function structural example of the radiation directivity characteristic estimation apparatus 100 'of this invention. The figure which shows the operation | movement flow of radiation | emission directivity characteristic estimation apparatus 100 '.

  Embodiments of the present invention will be described below with reference to the drawings. The same reference numerals are given to the same components in a plurality of drawings, and the description will not be repeated. Prior to the description of the embodiments, the basic idea of the present invention will be described.

[Basic idea of the present invention]
The basic idea of the present invention is that, when interpolating the radiation directivity characteristics between observation points, pay attention to the difference between the low-frequency and high-frequency wavelengths and determine the number of known observation signals used in the surroundings for interpolation. It is estimated by changing.

  FIG. 4 qualitatively shows the relationship between the circumference Q of the sound source and the radiation directivity characteristics. FIG. 4A shows an example of the radiation directivity characteristic of a wavelength larger than the length Q around the sound source, that is, a frequency equal to or lower than the frequency corresponding to the wavelength of the length Q around the sound source. FIG. 4B shows an example of radiation directivity characteristics of a wavelength smaller than the length Q around the sound source, that is, a frequency equal to or higher than the frequency corresponding to the wavelength of the length Q around the sound source. Both exhibit radiation directivity in polar coordinates.

  The sound has a longer wavelength at lower frequencies and a shorter wavelength at higher frequencies. When the size of the sound source is smaller than the wavelength, it is considered that the change due to the direction of the radiation directivity characteristic of the sound from the sound source is small (FIG. 4A). This is clear from the experience of hearing low-frequency sounds with the same sound pressure when moving and listening. On the other hand, the radiation directivity characteristic of a sound with a high frequency when the size of the sound source is larger than the wavelength varies greatly depending on the direction (FIG. 4B).

  If the sound wave is decomposed for each frequency, the waveform of each frequency is a sine wave. When the sine wave of each frequency of the radiation directivity characteristic is also observed at an interval of 1/8 or less of the wavelength, the change of the characteristic is considered to be gradual. In FIG. 5, the change of 1/8 wavelength of the sine wave is shown by a broken line. The horizontal axis in FIG. 5 is time, and the vertical axis is amplitude. When a sine wave is viewed at an interval equal to or less than one-eighth of the wavelength, there is no change in inclination, and there is little error even if linear approximation is performed.

  In the present invention, the radiation directivity characteristic is estimated by paying attention to the above two characteristics. In the sound source with the radius r, the change of the radiation directivity characteristic below the frequency corresponding to the eighth wavelength of the length Q around the sound source is gentle. Therefore, if the frequency is equal to or lower than the frequency f shown in the equation (1), the observation point for observing the radiation directivity characteristic of the sound source may be one point on the circumference of the radius R.

  Here, c is the speed of sound and is about 340 m.

  On the other hand, since the wavelength becomes shorter as the frequency becomes higher, the change in the radiation directivity characteristic of the sound source becomes severe. In order to observe at an interval of 1/8 wavelength, the minimum number M of observation points is expressed by the equation (2). It is necessary to increase as represented by. Since −1 has the same starting point (0) and end point (2πr), the number is reduced.

  Note that 8 is a value determined because the change in the characteristics of the sound wave is gentle, so any value that is a real number equal to or greater than 8 is acceptable. However, if the value is simply increased, M also increases, so that the value is determined by the balance with the number L of microphones.

  Thus, the minimum number of observation points M is a value that varies with the frequency f. When the number of observation points is constant, the minimum observation point number M for estimating the radiation directivity characteristic increases as the frequency f increases. The radiation directivity characteristic between the observation points is obtained from the two observation signals sandwiching the estimated position.

  On the other hand, when the frequency f is low, the minimum number of observation points M is in a decreasing relationship. However, there are more observation signals than the minimum number M of observation points. Although it is possible to estimate the radiation directivity characteristics from a small number of M observation signals, it is conceivable that the observation signals include noise. Therefore, it is possible to reduce the estimation error of the radiation directivity characteristic by interpolating including observation signals at observation points other than two adjacent points.

  The number N of points used for this interpolation can be formulated as shown in Equation (3).

  When the minimum number of observation points M is equal to or greater than the actual number of observation points L (number of microphones), it is sufficient to interpolate using observation signals at two adjacent observation points, and the radiation directivity characteristics can be obtained with a small number of microphones. It can be estimated accurately. Details will be described later.

FIG. 6 shows a functional configuration example of the radiation directivity characteristic estimation apparatus 100 of the present invention. FIG. 7 shows the operation flow. The radiation directivity characteristic estimation apparatus 100 includes L microphones 2 ₁ to 2 _L , a sound collection unit 10, a minimum observation point number calculation unit 20, an interpolation point calculation unit 30, and a radiation directivity characteristic estimation unit 40. To do.

FIG. 8 shows an example of the arrangement of the sound source 1 and L microphones. FIG. 8 shows an example in which, for example, ₁₆ microphones 2 ₁ to 2 ₁₆ are arranged on the circumference of a predetermined radius R around the sound source 1 having a radius r. The sound collection signals collected by the 16 microphones are input to the sound collection unit 10. The number of microphones is not limited to 16. In addition, although the example in which the sound source 1 is arranged at equal intervals with a central angle of 22.5 ° has been shown, it is not necessary to have equal intervals. Further, as shown in FIG. 2, the microphones may be arranged in a square shape with the sound source 1 as the center. The number L of microphones is a number equal to or greater than M determined by Expression (2), based on the length Q around the sound source 1 and the upper limit frequency f _max of the radiation directivity to be obtained. For example, when the upper limit frequency f _max is 1280 Hz, L = 32.

  The sound collection unit 10 performs frequency analysis on the L sound collection signals and obtains an observation signal from the sound source 1 for each necessary frequency (step S10). The sound collection unit 10 performs digital signal processing. In the figure, an A / D converter or the like for converting the sound collection signal, which is an analog signal, into a digital signal is omitted.

  The sound collection unit 10 includes a frequency analysis unit 11 and a storage unit 12. Sixteen sound collection signals are subjected to frequency analysis for each frequency required by the frequency analysis unit 11, and the analyzed observation signals are stored in the storage unit 12. To be recorded. For the frequency analysis means 11, for example, fast Fourier transform or the like is used. The frequency analysis means 11 performs frequency analysis in units of the number of samples from which a desired frequency resolution can be obtained (step S11). The observation signal subjected to frequency analysis is stored in the storage unit 12 (step S12).

  The minimum observation point number calculation unit 20 obtains the number obtained by subtracting 1 from the number obtained by dividing the circumference Q of the sound source 1 by a wavelength equal to or less than one-eighth of the wavelength for each frequency as the minimum observation point number M for each frequency. (Step S20). The interpolation point calculation unit 30 obtains a value obtained by adding 1 to the value obtained by dividing the number L of microphones by the minimum number of observation points M + 1 as the number N of observation points used for interpolation for interpolating the radiation directivity characteristic (step S30). The peripheral length Q of the sound source 1 and the number L of microphones may be set as constants in advance or may be given from the outside.

  The radiation directivity estimator 40 interpolates using the observation signals of the number N of observation points used for interpolation, and estimates the radiation characteristic of the position between the observation points (step S40). The radiation directivity characteristic estimation unit 40 reads the frequency analysis result of the observation points used for interpolation stored in the storage unit 12 and estimates the radiation characteristic of the position between the observation points. The estimated radiation characteristic may be a power for each frequency, or may be a power obtained by inverse Fourier transforming the estimated frequency component into a time domain signal.

  Next, the operation of the radiation directivity characteristic estimation unit 40 will be described in more detail with a specific example. When the radius r of the sound source 1 is r = 0.17 m and the number L of microphones is L = 32, the minimum observation point number calculation unit 20 and the interpolation point number calculation unit 30 have the minimum observation point number M and the number N of observation points used for interpolation. Is calculated as shown in Table 1.

周波数f＝80Hzの場合、最低観測点数ＭはＭ＝１である。これは、放射指向特性が方向によって変動しないためである。したがって、例えば5°方向の放射指向特性を推定する場合、0°の観測点の観測信号と同じ値にして良い。しかし、その0°の観測信号に雑音が含まれている可能性もある。一方、観測信号は３２個存在するので、推定する位置を含むその方向の１６個の観測信号の例えば平均値を、その位置の放射指向特性の値とする。このように複数の観測信号を平均するので、放射指向特性の推定精度を向上させることができる。

When the frequency f = 80 Hz, the minimum observation point number M is M = 1. This is because the radiation directivity characteristic does not vary depending on the direction. Therefore, for example, when estimating the radiation directivity characteristic in the 5 ° direction, it may be the same value as the observation signal at the 0 ° observation point. However, there is a possibility that the observed signal at 0 ° contains noise. On the other hand, since 32 observation signals exist, for example, an average value of 16 observation signals in the direction including the estimated position is set as a value of the radiation directivity characteristic at the position. Since the plurality of observation signals are averaged in this way, the estimation accuracy of the radiation directivity characteristics can be improved.

  When the frequency f = 160 Hz, the minimum number of observation points M is M = 3. As the frequency is slightly higher, the minimum number of observation points M required for interpolation increases. In this case, for example, an average value of three observation signals including the estimated position may be used as an estimated value of the radiation directivity characteristic at the position, but in order to improve estimation accuracy, eight observations including the estimated position are included. Let the average value of the signal be the estimated value.

  When the frequency f = 1280 Hz, the minimum number of observation points M is M = 31. For the radiation directivity characteristics between the observation points, the average value of the two observation signals sandwiching the position between them is an estimated value.

  As described above, in the low frequency region, that is, the frequency region where the actual number of observation points is larger than the minimum number of observation points M, interpolation can be performed using more observation signals than two adjacent points. Is equal to or greater than the number of actual observation points L, it can be understood that interpolation may be performed using observation signals at two adjacent points.

  Although the method for obtaining the estimated value by the average value has been described, the estimated value may be obtained by another method. For example, an approximate straight line may be obtained by the least square method, and the approximate straight line may be used for linear interpolation. When N ≧ 4, the radiation directivity characteristic at the estimated position may be estimated by spline interpolation. Spline interpolation is a well-known interpolation method that performs interpolation with a smooth curve.

  In addition, although the radiation directivity estimation apparatus 100 did not touch on the removal of the reverberation component, the reverberation removing means is provided in the sound collection unit 10 to remove the reflection / reverberation component of the room as described in the prior art. You may make it do. FIG. 9 shows a functional configuration example of the radiation directivity characteristic estimation apparatus 100 ′ provided with the dereverberation means 13 for extracting the signal of only the first half of the impulse response. The operation flow is shown in FIG. The radiation directivity estimation apparatus 100 ′ is different from the radiation directivity characteristic estimation apparatus 100 only in the sound collection unit 10 ′ provided with the dereverberation means 13. The operation of the dereverberation means 13 is the same as that already described in the prior art.

  As described above, the radiation directivity characteristic estimation method of the present invention pays attention to the difference between the low and high wavelengths when interpolating the radiation directivity characteristics between observation points, and knows the surroundings used for interpolation. This changes the number of observation signals used. According to this method, in order to know the radiation directivity characteristics of a sound source, it was necessary to prepare many observation points, but the radiation directivity characteristics in all directions were estimated from a small number of observation points. It is possible and efficient.

  When the processing steps in the above method are realized by a computer, the processing contents of the functions that each step should have are described by a program. Then, by executing this program on the computer, the processing means in each process is realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the computer-readable recording medium, for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used. Specifically, for example, as a magnetic recording device, a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only). Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.

  The program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in a recording device of a server computer and transferring the program from the server computer to another computer via a network.

  Each means may be configured by executing a predetermined program on a computer, or at least a part of these processing contents may be realized by hardware.

Claims (6)

A radiation directivity characteristic estimation method for estimating a radiation directivity characteristic of a sound source having a circumference length of Q,
A sound collection process for obtaining an observation signal from the sound source for each necessary frequency using sound collection signals of L microphones arranged at predetermined intervals around the sound source,
The lowest observation point number calculation unit obtains a number obtained by subtracting 1 from the number obtained by dividing the circumference Q of the sound source by a wavelength equal to or less than one-eighth of the wavelength for each frequency as the minimum observation point number M for each frequency The process of calculating the minimum number of observation points,
An interpolation point number calculating step for obtaining a value obtained by adding 1 to a value obtained by dividing the number L of the microphones by the minimum number of observation points M + 1 as the number N of observation points used for interpolation for interpolating the radiation directivity characteristics;
A radiation directivity estimation section in which a radiation directivity estimation section interpolates using observation signals of the number N of observation points used for the interpolation to estimate the radiation directivity characteristics at positions between the observation points;
A radiation directivity characteristic estimation method including: In the radiation directivity characteristic estimation method according to claim 1,
The radiation directivity characteristic estimation method, wherein the radiation directivity characteristic estimation process is a process of estimating a radiation characteristic at a position between the observation points by linear interpolation. In the radiation directivity characteristic estimation method according to claim 1,
The radiation directivity estimation method, wherein the radiation directivity estimation process is a process of estimating the radiation characteristics at positions between the observation points by spline interpolation when the number N ≧ 4. In the radiation directivity characteristic estimation method according to any one of claims 1 to 3,
The above sound collection process
A radiation directivity characteristic estimation method comprising a dereverberation step of extracting a signal of only the first half of an impulse response. A radiation directivity characteristic estimation apparatus for estimating a radiation directivity characteristic of a sound source having a circumference length of Q,
L microphones arranged at predetermined intervals around the sound source;
A sound collection unit that receives L sound collection signals of the microphone as input and obtains an observation signal from the sound source for each necessary frequency;
A minimum observation point number calculation unit for obtaining a number obtained by subtracting 1 from the number obtained by dividing the wavelength of the observation frequency by a wavelength equal to or less than one-eighth of the wavelength as the minimum observation point number M;
An interpolation point calculation unit for obtaining the number N of observation points used for interpolation for interpolating the radiation directivity characteristics of observation points as a value obtained by adding 1 to a value obtained by dividing the number L of the microphones by the minimum number of observation points M + 1;
A radiation directivity estimation unit that estimates the radiation directivity at positions between the observation points by interpolating the radiation characteristics of the observation points using observation signals of the number N of observation points used for the interpolation;
A radiation directivity characteristic estimation apparatus comprising:   A radiation directivity characteristic estimation method program for causing a computer to execute the radiation directivity characteristic estimation method according to any one of claims 1 to 4.

Images