JP2006325170A - Acoustic signal converter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a converter of an original acoustic signal which reproduces natural spatial orientation with a headphone or an earphone. <P>SOLUTION: Then acoustic signal after conversion is prevented from being symmetrical even when the original acoustic signal having symmetry is converted by obtaining direct sound and reflected sound from a virtual sound source by simulation, and simultaneously adding a numerical sequence substantially regarded as the random number, of which the average value is 0 and the standard deviation is ≥0.1 msec at arrival time of the reflected sound to a virtual participant thereby giving deviation. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to acoustic signal listening using a headphone or earphone.

  In particular, the present invention converts an original sound signal so that when the sound signal converted by the headphone or earphone is received, the sound signal can be obtained in the same spatial localization as when the original sound signal is reproduced by a speaker. The present invention relates to a conversion device.

  Phenomenon that the sound field is perceived as being in the listener's head when listening to stereophonic signals with headphones or earphones on the assumption that they will be played back by two speakers placed in front of the listener's left and right In-head localization occurs. In order to obtain a spatial localization equivalent to that obtained when a stereo signal is reproduced by a speaker using a headphone or an earphone, it is necessary to appropriately convert the original sound signal, and many devices or conversion methods for that purpose have been proposed.

  Some conventional converters use a filter including a reflected wave effect measured using a dummy head in an actual listening room (see, for example, Patent Document 1). Moreover, there exists a thing using the direct sound and reflected sound calculated | required by simulation (for example, refer patent document 2).

First, a method of using a filter measured in an actual listening room will be described with reference to FIGS. 2 (a) and 2 (b). In FIG. 2A, 5 is a listening room, in which a dummy head 1 is installed. Microphones 2 and 3 are housed in portions corresponding to the left and right ears of the dummy head 1. In the listening room 5, a left front speaker 4 is installed as a sound source. By adding the signal Sl to the left front speaker 4 and detecting it with the left microphone 2, a filter K ′ _LL representing the conversion that occurs when the sound from the left front speaker 4 reaches the left ear is obtained. Similarly, K ′ _LR indicating the conversion that occurs when the sound from the left front speaker 4 reaches the right ear is obtained. Next, as shown in FIG. 2B, a stereo headphone including a left headphone 10 and a right headphone 11 is attached to the dummy head 1. By adding the signal Sl to the left headphone and detecting with the left microphone 2, the filter Thp from the left headphone to the left ear is obtained. Considering left-right symmetry, Thp measurement is sufficient on either the left or right side.

The filter K _LL is defined as in _Equation 1. Here, Thp ⁻¹ is a Thp inverse filter. When this is applied to the left signal Sl and applied to the left headphone 10, the sound reaching the left ear from the headphone is equivalent to K ′ _LL acting on Sl. That is, when the original signal S1 is reproduced by the left speaker, the sound reaches the left ear. If the same operation is performed on the right ear, the listener feels as if sound is being emitted from the left speaker 4. When there are a plurality of speakers, the same operation may be applied to the sound signal to be added to each speaker.

Next, a method using the reflected sound obtained by simulation will be described. The original sound signal is added to a reflected sound generation circuit using a delay element to generate a reflected sound signal. The direct sound signal and the reflected sound signal are converted by the phase circuit, the amplitude control circuit, the delay circuit, etc., and then added. The circuit is configured so that natural spatial localization and timbre can be obtained when listening with headphones.
JP-A-2-200000 JP 58-124395 A

  Among the conventional techniques described above, in the method using a filter measured in an actual listening room, the speaker 4 used for measurement needs to be an ideal converter. Even when the speaker 4 cannot be regarded as ideal, it is possible to correct the filter to some extent based on the conversion characteristics of the speaker measured in front of the speaker 4. However, as shown in FIG. 2, the direct sounds 6 and 7 from the speaker 4 are radiated on the axis of the speaker 4, whereas the reflected sounds 8 and 9 are generally not radiated on the axis. . A speaker is generally known to have a strong directivity, and the frequency dependence of the amplitude and phase in a direction away from the axis is very different from that on the axis. It is practically impossible to correct this for all the reflected sounds in view of the large number of reflected sounds.

  In addition, the acoustic characteristics of the listening room used for measurement, particularly the reflected sound generated in the listening room, greatly affects the sound received by the listener. For good listening, it is necessary to carefully adjust many parameters such as the size of the listening room and the reflectivity of the wall, and to properly arrange sound absorbing materials and objects that diffuse sound. This adjustment is difficult, and there is no real ideal listening room.

  As a result, the obtained filter is merely a filter when a specific speaker is arranged at a specific position in a specific listening room, and has to reflect the drawbacks of the room and the speaker. Also, it is extremely difficult to adjust the timbre and reverberation according to the listener's preference because it involves changing the characteristics of the listening room itself.

  The method used to convert the original sound signal by simulation relies on a combination of a simple phase circuit, amplitude control circuit, delay circuit, etc., to reflect the reflected sound, reflecting the actual complex reflected wave. It is not a thing. For this reason, it was unsatisfactory in terms of spatial orientation and reproduction of natural timbre.

  The present invention is intended to solve the problems of these conventional methods, and an apparatus for converting an original sound signal that makes it possible to reproduce natural spatial localization and timbre with a headphone or earphone. The purpose is to give.

  The first solving means according to the present invention provides a virtual sound source and a virtual sound for each of a direct wave and a plurality of reflected waves from each virtual sound source for a virtual listener and at least one virtual sound source arranged in a virtual room. A virtual listener is obtained from each virtual sound source by obtaining a filter between the left and right ears of the listener and moving and synthesizing the respective filters on the time axis based on the difference in arrival time between the direct wave and the reflected wave. Headphone or earphone reception that calculates the filter for the left and right ears of the sound, applies the filter from the virtual sound source to the left and right ears to the original sound signal corresponding to each virtual sound source, and then sums all the virtual sound sources This is an acoustic signal conversion device for use at this time, by converting the arrival time of at least some of the reflected waves at this time, even if the original acoustic signal having left-right symmetry is converted It is characterized in that the fruit is so as not symmetrical.

  In the second problem solving means, correction is performed using a headphone or an inverse filter of a filter between the earphone and the ear in any process of filter derivation in the acoustic signal converter.

  The third problem solving means is an acoustic signal conversion apparatus according to the first or second problem solving means in which the average value is 0 at the arrival time of each reflected wave as a method of giving a deviation to the arrival time of the reflected wave in the previous period. And a method of adding a sequence of substantially random numbers having a standard deviation of 0.1 ms or more is employed.

  Further, the fourth problem solving means is the acoustic signal converter according to the first, second or third problem solving means, wherein the arrival time is at least with respect to the reflected wave arriving 5-15 milliseconds later than the direct wave. The operation of shifting is performed.

  The operation of the first solving means is as follows. If the left-right symmetry of the reflected wave is lost by shifting the arrival time of the reflected wave, the listener can correctly identify the reflected wave. Therefore, a natural spatial localization is obtained as a result of the effective mechanism of correctly estimating the position of the sound source by comparing the reflected wave with the direct wave. Since the present invention calculates the reflected wave by simulation, the conversion result does not depend on the characteristics of a specific speaker or the acoustic characteristics of the room.

The action of the second solving means is to obtain a more natural timbre and spatial localization by adding correction according to headphone or earphone.

  In addition, the action of the third solving means is to obtain a natural spatial orientation by using random numbers when giving a deviation to the reflected wave, regardless of the virtual sound source arrangement of the original sound signal. It is in.

  The fourth solution is to obtain natural spatial localization and timbre by performing an operation to shift the arrival time with respect to a reflected wave that arrives at a delay of 5-15 msec from the direct wave. is there.

  As described above, the original sound signal is converted by the apparatus according to the present invention, and is received by a headphone or earphone, so that a natural space can be obtained regardless of the arrangement of the virtual sound source. Localization and timbre can be obtained.

  In addition, since the present invention obtains the reflected wave based on the simulation, it is possible to easily change the reproduced sound tone and reverberation according to the listener's preference by appropriately selecting the parameters used for the simulation. It also has the advantage of being.

  Embodiments of the present invention will be described below with reference to FIGS. 1 and 3 to 6. In the following description, it is assumed that the signal is discrete. However, the present invention is not limited to discrete signals. The filter is assumed to be expressed by an impulse response. Of course, a completely equivalent explanation can be made using a frequency response or a transfer function. Note that the head-related transfer function is also expressed by an impulse response.

In this embodiment, the head-related transfer function H ′ (θ, φ) measured in an anechoic room is used. This was measured as follows. In FIG. 3A, 1 is a dummy head having left and right microphones 2 and 3, 4 is a speaker, and 12 is an anechoic chamber. The speaker 4 is installed at an angle θ and an elevation angle φ in the horizontal direction with respect to the dummy head 1, and its axis is directed toward the dummy head. During the measurement, the distance D ₀ between the speaker 4 and the dummy head 1 is kept constant regardless of θ and ψ. The head-related transfer function H ′ (θ, φ) is measured with respect to various θ and φ by adding the impulse signal 1 to the speaker 4 and detecting with the left microphone 3. In consideration of the right and left object, it is sufficient to perform measurement on either the left or right microphone.

Next, a microphone (not shown) is installed on the axis of the speaker 4 in the anechoic chamber 12, and the response when the impulse signal 1 is applied to the speaker 4 is measured to obtain the filter Tsp of the speaker alone. . Further, this inverse filter is calculated as Tsp- ¹ . Both H ′ (θ, φ) and Tsp ⁻¹ are responses to the impulse signal and are FIR filters. By convolving these two FIR filters according to Equation 2, a set of head related transfer functions H (θ, φ) in which the response of the speaker 4 is corrected can be obtained. Since the axis of the speaker 4 always faces the dummy head, the deviation of the speaker 4 from the ideal converter can be completely corrected by this method.

  The simulation of the reflected wave was based on the virtual image method. This is a known technique for calculating a reflected wave on the assumption that sound is reflected by a wall with the same incident angle and reflection angle as a light beam. In FIG. 4, 13 is a virtual room, and 14 is the position of the virtual listener. It is assumed that left and right virtual speakers are installed in front of the listener in the room. Reference numeral 15 denotes the position of the left virtual speaker.

The reflected wave generated from the left virtual speaker and reaching the listener after being reflected by the left wall surface of the room is equal to the direct wave when the virtual speaker is at a virtual image position 16 that is plane-symmetric with respect to 15 with respect to the left wall surface. . In order to obtain the direction and propagation distance of multiple reflected waves, the position of the virtual image may be moved for each reflection. The number of times the reflected wave was reflected by the left and right wall surfaces was l, the number of times the reflected wave was reflected by the front and back wall surfaces was m, and the number of times the reflected waves were reflected by the upper and lower wall surfaces was n. The sign of 1 was positive when the reflected wave was first reflected on the right wall surface, and negative when the reflected wave was first reflected on the left wall surface. Similarly, when m is positive, it is assumed that the light is reflected first from the front wall, and when m is positive, it is assumed that the light is reflected first from the upper wall. By designating a set of these indices (l, m, n), the corresponding virtual image position is obtained, and the distance Dlmn until the reflected wave reaches the listener and the direction with respect to the listener can be calculated. For example, the reflected wave shown in FIG. 4 is represented by (l, m, n) = (− 1, 0, 0). (0, 0, 0) indicates a direct wave.

A filter H _LL indicating the response of the left ear when an impulse signal is applied to the virtual left speaker in the virtual room was obtained by the following procedure. For each virtual image specified by (l, m, n), the arrival direction of the reflected wave to the listener is determined, and H (θ, φ) in the closest direction is selected, and this is designated as HLlmn. Further, the distance Dlmn between the virtual image and the listener is obtained. Next, HLlmn is synthesized according to Equation 3.

In Equation 3, τ (t) is an operator that delays time by t, and S is the speed of sound. γ is the reflectance of the wall, and the reverberation time was adjusted by changing this. A reverberation time of 0.2-0.5 seconds is generally preferred, and here it was set to 0.25 seconds. Lp (D) is a low-pass filter that provides high-frequency attenuation due to air absorption when a sound wave propagates a distance D. D ₀ / Dlmn is a term that gives the attenuation of the sound wave according to the distance, and D ₀ is the distance between the speaker and the dummy head when H (θ, φ) is measured. Considering that the reflected wave attenuates to 1/1000 after the reverberation time has elapsed, the sum of Equation 3 is sufficient for a wave that reaches the listener by about 1.5 times the reverberation time.

Next, as shown in FIG. 3B, a stereo headphone was attached to the dummy head 1, and the filter Thp was measured by giving the impulse signal 1 to the left headphone 10. Then, the inverse filter Thp- ¹ was calculated. By convolving this with H _LL , a filter F _LL was obtained. If F _LL is convolved with the left channel sound signal and added to the left headphone, the same sound as if the original sound signal was added to the left speaker should be added to the left ear. FIG. 5 is an example of the calculated _FLL , and shows the vicinity of the direct wave 20a in an enlarged manner. The horizontal axis is the time expressed by the number of samples, and the vertical axis is an arbitrary scale. In this experiment, a sampling frequency of 44.1 kHz was used. Many direct waves are followed by indirect waves 20b, and a waveform close to an actual reflected wave is obtained.

In the case of sound waves that reach the right ear from the left speaker, Equation 4 is used instead of Equation 3. Here, HRlmn is obtained by replacing θ of HLlmn with 2π−θ. Obtain _{F LR} by convoluting this with ^{Thp -1.}

H ′ (θ, φ) used in this example is measured by a microphone mounted on the end of a dummy head with a chamber imitating the ear canal, and is corrected by convolving Thp ^−1. It is necessary. If H ′ (θ, φ) is measured with a microphone provided near the earlobe of the dummy head, correction by Thp ⁻¹ can be omitted. However, better spatial localization and timbre can be obtained by correct correction as in this embodiment.

Next, the left and right original acoustic signals S _L and S _R are converted into acoustic signals S _L ^hp and S _R ^hp to be added to the headphones by Equation 5. Here, F _RL and F _LL are filters corresponding to the right speaker.

The virtual room, virtual speakers, and virtual listeners were arranged symmetrically. Therefore, from the symmetry, F _RR = F _LL and F _RL = F _LR . However, when listening to a sound signal that has undergone such a completely symmetrical transformation with a headphone, the sound on the left and right is localized outside the head, but the sound that should be localized in the front center is localized in the head. It was. This is due to the following reason.

Assume that the original acoustic signal has a relationship of S _L = S _R. If you listen to this with the left and right speakers, the sound will be localized near the center of the two speakers. On the other hand, if this signal is converted by Equation 5 with F _LL = F _RR and F _RL = F _LR , S _L ^hp = S _R ^hp and the sound generated from the headphone is the same on the left and right. It is said that humans estimate the direction based on the difference between the arrival times of the left and right sounds and the difference in the frequency spectrum. However, if the left and right sounds are exactly the same, this method uses a symmetrical plane (the front of the listener). It is only possible to estimate that there is a sound source in the rear or overhead), and not an accurate direction. There is a theory that the direction is estimated by the difference of each frequency spectrum when the sound source is in the plane of symmetry, but there is no way to know the frequency spectrum of the original sound source, so how it changed Since there is no means for estimating the azimuth, it is not considered that the direction can be estimated by this method.

  A reverberant sound constituted by a large number of reflected waves is generally diffused, that is, reaches a listener uniformly from all directions. If so, it is expected that the reverberant sound has little relation to the position of the sound source. Therefore, if the difference in frequency spectrum between reverberant sound and direct sound is compared, it should be possible to estimate the position of the sound source in the plane of symmetry. Based on this hypothesis, it is important for the localization of natural sound images that the reverberant sound can be recognized correctly by humans. If the signals entering the left and right ears are the same, humans could not separate the reverberant sound and the direct sound, and as a result, the position of the sound source could not be estimated, resulting in in-head localization. Therefore, we considered that it would be possible to obtain correct localization if the reverberant sound was differentiated between left and right so that humans could separate the reverberant sound and the direct sound, and the following experiment was attempted.

(Experiment 1)
For each reflected sound, a variation based on a Gaussian distribution was added to the amplitude.

In Equation 6, the first term is a direct wave. The second term is the term of the reflected wave, and the sum is taken for the reflected wave. Glmn (1, δ) is a random number sequence having a Gaussian distribution with an average value of 1 and a standard deviation δ, and takes different values for different sets of (l, m, n). These were convolved with Thp ⁻¹ as F _LL and F _LR, and F _RR and F _RL were obtained in the same manner. The value of Glmn (1, δ) used at this time is naturally different from that used when F _LL and F _LR are obtained. Although δ was changed in the range of 0 to 0.4 and the original sound signal was converted using Equation 5, a listening experiment was performed, but the localization in the head was not eliminated.

(Experiment 2)

込むことによってＦ_ＬＲ、Ｆ_ＬＬ、Ｆ_ＲＲ、Ｆ_ＲＬを得た。原音響信号を数５に従って変換した後に

０．１ｍｓｅｃ以上の場合にスピーカーで再生した場合と遜色ない空間定位が得られた。しかし、得られた音色はやや不自然なものであった。

F _LR , F _LL , F _RR and F _RL were obtained. After converting the original sound signal according to Equation 5

In the case of 0.1 msec or more, the spatial localization comparable to the case of reproducing with a speaker was obtained. However, the obtained tone was somewhat unnatural.

(Experiment 3)
For each reflected sound, a variation based on a Gaussian distribution was added to the arrival time.

In Equation 8, Glmn (1, δ) is a random number sequence having a Gaussian distribution with an average value of 1 and a standard deviation δ, as before, and varies in the time at which the reflected wave arrives. After performing this operation, Thp- ¹ was convoluted to obtain F _LL and F _LR , and F _RR and F _RL were obtained in the same manner. Glmn (1, δ) for the right speaker is different from that of the left speaker as before. The original sound signal was converted according to Equation 5 and then auditioned with headphones. When δ is changed in the range of 0-0.2, a spatial localization comparable to that when δ = 0.005 or more is reproduced with a speaker is obtained, and the best value is in the range of δ = 0.1-0.15. Met.

  FIG. 6 illustrates the technique used in Experiments 1 to 3. In FIG. 6A, 13 is a virtual room, 15 and 17 are positions of left and right virtual speakers, and 14 is a position of a virtual listener. The sound wave generated by applying the impulse signal to the left virtual speaker is reflected on the left wall surface and reaches the virtual listener 18a, and the sound wave generated by applying the impulse signal to the right virtual speaker is reflected on the right wall surface. In this case, 19a is the virtual listener. It is assumed that the positions of the virtual room and the virtual speakers and listeners in the virtual room are symmetrical. At this time, in the calculation, the reflected wave 18a and the reflected wave 19a have the same shape and the same arrival time. As shown in 18b and 19b of FIG. 6 (b), Experiment 1 is the one in which the amplitudes of the left and right reflected waves are changed. In Experiments 2 and 3, as shown in 18c and 19c of FIG. 6 (c), The arrival times of the left and right waves are changed. In FIG. 6, for the sake of simplicity, only the case of the primary reflected wave is shown, but the same operation is performed for the multiple reflected wave.

  Since it was found that it is effective to vary the time that the reflected wave arrives, it was investigated which range of reflected waves would be most effective next.

(Experiment 4)
Similar to Experiment 3, for each reflected sound, the arrival time was varied based on the Gaussian distribution. However, in Experiment 3, all reflected waves were varied, whereas in Experiment 4, the reflected waves that reached the listener during a certain time Ta from the direct wave arrival time were not varied. As a result of trial listening with Ta changed, correct spatial localization was obtained if Ta was less than about 5 milliseconds.

(Experiment 5)
Contrary to Experiment 4, the arrival time varied only for the reflected wave reaching the listener during a certain time Tb from the direct wave arrival time, and the reflected wave reaching after that was not varied. . As a result of trial listening with varying Tb, correct spatial orientation was obtained if Ta was about 14 ms or longer.

  In summary, in order for the listener to correctly recognize the sound image that should be positioned at the center of the left and right speakers, the arrival time of the reflected waves that reach the left and right ears with a delay of about 5-14 msec from the direct wave is shifted. It was concluded that it was important to make the listener recognize the reflected wave. The necessary difference between the arrival times on the left and right is about 0.2 msec in the case of Experiment 2.

  The arrival time difference in Experiment 3 can be calculated as follows. In this experiment, the distance between the virtual sound source and the virtual listener was about 2.6 m. Therefore, the direct wave reaches the listener in about 7.5 milliseconds. From Experiment 4 and Experiment 5, it can be seen that the reflected wave delayed by about 10 milliseconds from the direct wave is most important for spatial localization. Since this wave takes 17.5 milliseconds to propagate, if a dispersion with a standard deviation of 0.005 is given to this wave, a dispersion of about 0.09 milliseconds is given to the arrival time. In this case, the arrival time difference is about 0.13 ms, which is approximately the same as Experiment 2.

  The method of giving the time difference between the arrival times of the left and right reflected waves may be given mechanically as in Experiment 2. However, as in Experiment 3, it is preferable to make the time difference between the reflected waves as regular as possible. It is done. This is because the influence of the standing wave in the room can be eliminated by giving variations in arrival time based on the random number sequence. In addition, if the time difference is given to all the reflected waves, a natural spatial localization can be obtained.

  As long as the rule that the arrival time of the reflected wave is different for the left and right ears is satisfied, the reflected wave may be freely corrected according to the purpose. For example, the reverberation time can be changed according to preference by changing the reflectance of the wall. In addition, the initial reflected wave that reaches within 5 milliseconds from the direct wave has a large influence on the timbre although the influence on the spatial localization is small. Therefore, it is possible to obtain a more natural timbre by making the amplitude of the initial reflected wave smaller than the value obtained by calculation. Furthermore, the reflected wave from the wall surface behind the virtual speaker may be reduced by providing directivity to the virtual speaker. In addition, since the reflected wave that is 15 ms or more away from the direct wave has little influence on the sound image localization, the calculation of the reflected wave in this range can be replaced with a simpler filter.

  Although the above embodiment is a case where the original sound signal has two channels, the present invention is not limited to this. FIG. 1 shows a case where a virtual speaker 21 is arranged in front of the listener 14. The reflected waves 18d and 19d from the left and right wall surfaces reach the listener at the same time in the calculation. Using the same method as in Experiment 3, if the arrival time is varied with a random number sequence having a Gaussian distribution for each reflected wave, the sound image is correctly displayed in front of the listener when the converted signal is received with headphones. To be localized. On the other hand, in the method of Experiment 2 in which the reflected wave is mechanically shifted at the same time, the arrival time cannot be shifted between the left and right reflected waves. That is, the method of giving a difference in arrival time with a random number sequence has a wider application range. When there are three or more virtual speakers, for example, when converting a surround playback signal, the above method may be applied to each virtual speaker.

The figure which shows the principle of this invention. The figure which shows how to obtain | require the filter by the conventional method. The figure which shows how to obtain | require the head-related transfer function used in the Example of this invention. The figure which shows the simulation method of the reflected wave in the Example of this invention. 4 is an example of a filter according to an embodiment of the invention. The figure which shows the experimental method performed in the Example of this invention.

Explanation of symbols

1 dummy head 2 left microphone 3 right microphone 4 speaker 5 listening room 6, 7 direct wave 8, 9 reflected wave 10 left headphone 11 right headphone 12 anechoic room 13 virtual room 14 virtual listener 15 virtual left speaker position 16 Virtual image position 17 Virtual right speaker position 18a, 18b, 18c, 18d Reflected wave 19a, 19b, 19c, 18d Reflected wave 20a Direct wave 20b Reflected wave 21 Virtual speaker position

Claims (4)

For a virtual listener and at least one virtual sound source placed in a virtual room,
A filter between the virtual sound source and the left and right ears of the virtual listener is obtained for each of the direct wave and the plurality of reflected waves from each virtual sound source, and the respective based on the difference between arrival times of the direct wave and the reflected wave. By moving and synthesizing these filters on the time axis, a filter from each virtual sound source to the left and right ears of the virtual listener is obtained, and the original sound signal corresponding to each virtual sound source is converted to the original sound signal from the virtual sound source to the left and right ears In the acoustic signal converter that operates the filter and then sums all virtual sound sources,
By changing the arrival time of at least some of the reflected waves, even if the original acoustic signal with left-right symmetry is converted, the conversion result does not become left-right symmetrical. Acoustic signal converter.   2. The acoustic signal converter according to claim 1, wherein correction is performed using an inverse filter of a headphone or a filter between an earphone and an ear in any one of the processes of deriving the filter.   A method of giving a deviation to the arrival time of the reflected wave in the previous period is to add a sequence of substantially random numbers having an average value of 0 and a standard deviation of 0.1 ms or more at the arrival time of each reflected wave. The acoustic signal converter according to claim 1 or 2.   The acoustic signal converter according to claim 1, 2 or 3, wherein an operation for shifting the arrival time is performed at least for a reflected wave that arrives at a delay of 5-15 milliseconds from a direct wave.

Images