Audio signal downmixing method, audio signal downmixing device, program

Abstract

The present invention provides an audio signal downmixing method and apparatus that estimate the phase difference spectrum of two channel signals with less computational processing than conventional methods and using processing suitable for fixed-point arithmetic. [Solution] In an audio signal downmixing device, if u(k) and v(k) are the real and imaginary parts of the product Y(k) of the complex conjugate ￣X2(k) of the frequency spectrum X1(k) and frequency spectrum X2(k), respectively, the phase difference spectrum estimation unit selects one of several representative values ​​of phase difference spectra, which are values ​​on the circumference of the unit circle in the complex plane and have different argument angles in the complex plane, based on a combination of signs indicating whether u(k) is positive or negative and signs indicating whether v(k) is positive or negative, and obtains it as the phase difference spectrum φ(k).

Description

Technical Field

[0001] The present invention relates to a technique for obtaining a phase difference spectrum of signals of two channels in order to mix, encode, or process the signals of the two channels using the relationship between the signals of the two channels.

Background Art

[0002] As a technique for obtaining the phase difference spectrum of sound signals of two channels, there is a technique described in Patent Document 1. What is mainly described in Patent Document 1 is a technique for mixing sound signals of multiple channels to obtain one sound signal. Specifically, a value representing the magnitude of the correlation between the input sound signals of two channels and which of the input sound signals of the two channels is leading are obtained, and the input sound signal of the leading channel among the input sound signals of the two channels is included more greatly as the value representing the magnitude of the correlation is larger. A technique for obtaining a downmix signal by weighted addition of the input sound signals of two channels is described. Patent Document 1 describes a technique for obtaining the time difference between the input sound signals of two channels in order to obtain which of the input sound signals of the two channels is leading. As an example of the technique for obtaining the time difference between the input sound signals of two channels, the phase difference spectrum in the frequency domain of the input sound signals of the two channels is obtained, and the inverse Fourier transform of each candidate time difference given to the phase difference spectrum is performed to obtain a phase difference signal for each time difference. A technique for obtaining the time difference with the largest phase difference signal among the candidate time differences as the time difference between the input sound signals of the two channels is described. According to this technique, by using the phase difference spectrum of each frequency of the sound signals of two channels, the time difference between the sound signals of the two channels can be obtained so as to be less affected by the harmonic structure and pitch components of the sound signals. That is, the technique for obtaining the phase difference spectrum of the sound signals of two channels described in Patent Document 1 is a useful technique in applications such as obtaining the time difference between the sound signals of two channels, obtaining which of the sound signals of the two channels is leading, and mixing, encoding, or processing signals using the relationship between any two of these sound signals. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] International Publication No. 2021 / 181974 [Overview of the project] [Problems that the invention aims to solve]

[0004] To obtain the phase difference spectrum of two-channel signals using the technology described in Patent Document 1, it is necessary to divide each complex spectrum by the square root of the sum of the squares of the real and imaginary parts for each frequency. Because the range of possible values ​​for the sum of squares is large, if the process of obtaining the sum of squares is performed on a general-purpose or dedicated processor, it is necessary to perform power-intensive floating-point arithmetic or fixed-point arithmetic, which involves a large amount of computational processing including additional operations such as digit alignment. Furthermore, performing square root calculations and divisions on a processor requires, for example, about 30 times the processing power of addition, subtraction, and multiplication. Therefore, the technology for obtaining the phase difference spectrum of two-channel signals described in Patent Document 1 has the problem of high power consumption and / or high computational processing load when implemented on a processor. In other words, the technology for obtaining the phase difference spectrum of two-channel signals described in Patent Document 1 has the problem of high computational processing load and being unsuitable for fixed-point arithmetic. The present invention aims to provide an audio signal downmixing technique that uses a method for estimating the phase difference spectrum of two channel signals with less computational processing than conventional methods and with processing suitable for fixed-point arithmetic. [Means for solving the problem]

[0005] One aspect of the present invention is an audio signal downmixing method comprising: a phase difference spectrum estimation step of estimating a phase difference spectrum φ(k) of the frequency spectrum X1(k) of an input signal of a first channel and a phase difference spectrum X2(k) of an input signal of a second channel for a frequency k; an inter-channel relationship information acquisition step of obtaining a value representing the correlation between the input sound signal of the first channel and the input sound signal of the second channel and leading channel information, which is information indicating which channel, the first channel or the second channel, is preceding, using the estimated phase difference spectrum φ(k) for the frequency k; and a downmixing step of obtaining a downmix signal from the input sound signal of the first channel and the input sound signal of the second channel using the value representing the correlation and the leading channel information, wherein u(k), Let v(k) be the real and imaginary parts of the product Y(k) of the complex conjugates ￣X2(k) of the frequency spectra X1(k) and X2(k), respectively. The phase difference spectrum estimation step involves selecting one of several representative values ​​of phase difference spectra, which lie on the circumference of the unit circle in the complex plane and have different argument angles in the complex plane, based on a combination of signs indicating whether u(k) is positive or negative and signs indicating whether v(k) is positive or negative, to obtain the phase difference spectrum φ(k). [Effects of the Invention]

[0006] According to the present invention, the phase difference spectrum of two channels of signals used in audio signal downmixing can be estimated with less computational processing than conventional methods and using processing suitable for fixed-point arithmetic. [Brief explanation of the drawing]

[0007] [Figure 1] This is a block diagram showing the audio signal downmixing device 100 according to the first and second embodiments. [Figure 2] This is a flowchart illustrating the processing of the audio signal downmixing device 100 in the first and second embodiments. [Figure 3] This figure illustrates the representative values ​​for the first example of the phase difference spectrum estimation unit 122. [Figure 4] This figure illustrates the representative values ​​of the first quadrant in the second example of the phase difference spectrum estimation unit 122. [Figure 5] This is a block diagram of the inter-channel relationship information estimation device 120 according to the third embodiment. [Figure 6] This is a flowchart showing the processing of the inter-channel relationship information estimation device 120 of the third embodiment. [Figure 7] This is a block diagram of the phase difference spectrum estimation device 200 according to the fourth embodiment. [Figure 8] This is a flowchart showing the processing of the phase difference spectrum estimation device 200 according to the fourth embodiment. [Figure 9] This is a block diagram of the signal encoding device 300 according to the fifth embodiment. [Figure 10] This is a flowchart showing the processing of the signal encoding device 300 according to the fifth embodiment. [Figure 11] This is a block diagram of the signal processing device 400 according to the sixth embodiment. [Figure 12] This is a flowchart showing the processing of the signal processing device 400 according to the sixth embodiment. [Figure 13] This figure shows an example of the functional configuration of a computer that implements each device in the embodiments of the present invention. [Modes for carrying out the invention]

[0008] <First Embodiment> In the first embodiment, the phase difference spectrum estimation process of the present invention is described in which the process is applied to an audio signal downmixing device that performs downmixing processing considering the relationship between a first channel input audio signal and a second channel input audio signal in order to obtain a monaural signal useful for signal processing such as encoding processing.

[0009] Two-channel audio signals that are the target of signal processing such as encoding are often digital audio signals obtained by AD conversion of sounds picked up by a left-channel microphone and a right-channel microphone placed in a certain space. In this case, the input to the signal processing device such as encoding is a first-channel input audio signal, which is a digital audio signal obtained by AD conversion of the sound picked up by the left-channel microphone placed in the space, and a second-channel input audio signal, which is a digital audio signal obtained by AD conversion of the sound picked up by the right-channel microphone placed in the space. These first-channel input audio signals and second-channel input audio signals often contain sounds emitted by each sound source present in the space, with the difference (so-called arrival time difference) between the arrival time from the sound source to the left-channel microphone and the arrival time from the sound source to the right-channel microphone. Taking this into consideration, the audio signal downmixing device of the first embodiment performs downmixing processing that takes into account the relationship between the first-channel input audio signal and the second-channel input audio signal so that a monaural signal useful for signal processing such as encoding can be obtained. The following describes the audio signal downmixing device of the first embodiment.

[0010] The audio signal downmixing device 100 of the first embodiment includes an inter-channel relationship information estimation unit 120 and a downmixing unit 130, as shown in Figure 1. The audio signal downmixing device 100 obtains and outputs a downmix signal, described later, from an input 2-channel stereo time-domain audio signal in frame units of a predetermined time length, for example, 20 ms. The input to the audio signal downmixing device 100 is a 2-channel stereo time-domain audio signal, which is, for example, a digital audio signal obtained by AD conversion after sound such as speech or music is picked up by two microphones, a digital decoded audio signal obtained by encoding and decoding the aforementioned digital audio signal, and a digital signal processed audio signal obtained by signal processing the aforementioned digital audio signal, and consists of a first channel input audio signal and a second channel input audio signal. The downmix signal, which is a time-domain monaural audio signal obtained by the audio signal downmixing device 100, is input to at least an audio signal encoding device that encodes the downmix signal and at least an audio signal processing device that signals the downmix signal. If the number of samples per frame is T, the audio signal downmixing device 100 receives the first channel input audio signals x1(1), x1(2), ..., x1(T) and the second channel input audio signals x2(1), x2(2), ..., x2(T) on a frame-by-frame basis, and the audio signal downmixing device 100 downmixes the signal x on a frame-by-frame basis. M (1), x M (2), ..., x M The system obtains (T) and outputs it. Here, T is a positive integer; for example, if the frame length is 20ms and the sampling frequency is 32kHz, then T is 640. The audio signal downmixing device 100 performs the processing shown in steps S120 and S130 in Figure 2 for each frame.

[0011] [Channel relationship information estimation unit 120] The inter-channel relationship information estimation unit 120 receives the first channel input sound signal and the second channel input sound signal input to the sound signal downmixing device 100. The inter-channel relationship information estimation unit 120 obtains and outputs an inter-channel correlation value γ and preceding channel information from the first channel input sound signal and the second channel input sound signal (step S120). Specifically, the processing in step S120 consists of the processing from steps S121 to S123 shown in Figure 2. As shown in Figure 1, the inter-channel relationship information estimation unit 120 includes a Fourier transform unit 121, a phase difference spectrum estimation unit 122, and an inter-channel relationship information acquisition unit 123. The Fourier transform unit 121 performs step S121, the phase difference spectrum estimation unit 122 performs step S122, and the inter-channel relationship information acquisition unit 123 performs step S123.

[0012] Leading channel information indicates which channel, the first or the second, contains the same sound signal first. For example, it corresponds to the information of which microphone—the left channel microphone or the right channel microphone—is positioned in a given space to receive sound from the main sound source in that space first. If the same sound signal is contained in the first channel input signal first, it is said that the first channel is leading or the second channel is trailing. If the same sound signal is contained in the second channel input signal first, it is said that the second channel is leading or the first channel is trailing. Leading channel information indicates which channel, the first or the second, is leading. The inter-channel correlation value γ is a correlation value that takes into account the time difference between the first channel input sound signal and the second channel input sound signal. That is, the inter-channel correlation value γ represents the magnitude of the correlation between the sample sequence of the leading channel's input sound signal and the sample sequence of the trailing channel's input sound signal, which is positioned τ samples later than that sample sequence. This τ will also be referred to as the inter-channel time difference below. The preceding channel information and the inter-channel correlation value γ represent the relationship between the first channel input sound signal and the second channel input sound signal, and therefore can also be considered inter-channel relationship information.

[0013] [Fourier transform section 121] The Fourier transform unit 121 performs a Fourier transform on each of the first channel input sound signals x1(1), x1(2), ..., x1(T) and the second channel input sound signals x2(1), x2(2), ..., x2(T) as shown in equations (1-1) and (1-2) below, thereby obtaining the frequency spectra X1(k) and X2(k) at each frequency k from 0 to T-1 (step S121).

number

number

[0014] The frequency spectra X1(k) and X2(k) obtained by the Fourier transform unit 121 at each frequency k from 0 to T-1 are output from the Fourier transform unit 121 and input to the phase difference spectrum estimation unit 122.

[0015] [Phase difference spectrum estimation unit 122] First, let's explain the prior art. In Patent Document 1, the inter-channel relationship information estimation unit 120, after step S121, uses the frequency spectra X1(k) and X2(k) at each frequency k obtained by equations (1-1) and (1-2) to obtain the phase difference spectrum φ(k) at each frequency k by the following equation (1-3).

number

[0016] When the process of obtaining the phase difference spectrum φ(k) at each frequency k using equation (1-3) is performed by the processor, it is assumed that the following equation (1-4), which is equivalent to equation (1-3), is used. ￣X2(k) is the complex conjugate of X2(k). Note that the superscript "￣" should ideally be placed directly above "X2(k)", but due to the constraints of notation in the specification, it is written as "￣X2(k)".

number

[0017] The process of obtaining the phase difference spectrum φ(k) at each frequency k using equation (1-4) includes, for example, a first process of calculating the product of the complex conjugate φ(k) of the frequency spectrum X1(k) and the frequency spectrum X2(k), a second process of calculating |X1(k)| and |X2(k)|, a third process of calculating the product of |X1(k)| and |X2(k)|, and a fourth process of dividing the product obtained in the first process by the product obtained in the third process. Here, |X1(k)| is the real part X1(k) of the frequency spectrum X1(k), as shown in equation (1-5A) below. real and the imaginary part X1(k) imag It is the square root of the sum of the squares of . Similarly, |X2(k)| is the real part of the frequency spectrum X2(k), as shown in equation (1-5B) below. real and the imaginary part X2(k) imag It is the square root of the sum of the squares. In other words, the second operation involves two square root operations.

number

[0018] In a normal processor, as exemplified in the ITU-T arithmetic conversion standard, one operation of square root operation or division requires about 30 times the number of clock cycles of one operation of multiplication and addition. Therefore, the process of obtaining the phase difference spectrum φ(k) at each frequency k by Equation (1-4) involves the second process and the fourth process, resulting in a large amount of arithmetic processing. Although it is possible to complete the square root operation once by combining the second process and the third process, in this case, it is necessary to calculate the product of the value of the dimension of the energy of X1(k) and the value of the dimension of the energy of X2(k). Since the value of the dimension of energy has the magnitude of the square of the waveform value, the value of the product of the values of the dimension of energy has the magnitude of the fourth power of the waveform value. A value with the magnitude of the fourth power of the waveform value can be calculated without special processing for floating-point operations with high power consumption, but for fixed-point operations with low power consumption, additional processing such as digit alignment is required because the range of possible values is limited. That is, the process of obtaining the phase difference spectrum φ(k) at each frequency k by Equation (1-4) has the problems of a large amount of arithmetic processing and being unsuitable for fixed-point operations. Therefore, as will be described below, the phase difference spectrum estimation unit 122 estimates the phase difference spectrum of the signals of two channels with less arithmetic processing than before and a process suitable for fixed-point operations.

[0019] In the following, as shown in the following Equation (1-6A), the product of the frequency spectrum X1(k) of the first channel, which is the numerator of the right side of Equation (1-4), and the complex conjugate \(\overline{X2(k)}\) of the frequency spectrum X2(k) of the second channel is defined as Y(k). As shown in the following Equation (1-6B), the real part Y(k) real of Y(k) is defined as u(k), and as shown in the following Equation (1-6C), the imaginary part Y(k) imag of Y(k) is defined as v(k) for explanation.

Equation

[0020] As can be seen from the fact that in the right-hand term of equation (1-4), the product of the complex conjugate of frequency spectra X1(k) and X2(k), ￣X2(k), is divided by the product of |X1(k)| and |X2(k)|, the phase difference spectrum φ(k) lies on the circumference of the unit circle in the complex plane. Therefore, the complex value of a point on the complex plane whose argument is the same as Y(k) and which lies on the circumference of the unit circle in the complex plane is the phase difference spectrum φ(k).

[0021] [[First example of the phase difference spectrum estimation unit 122]] As described above, the arguments of Y(k) and the phase difference spectrum φ(k) on the complex plane are the same, so Y(k) and the phase difference spectrum φ(k) are in the same quadrant of the complex plane. Therefore, the phase difference spectrum estimation unit 122 of the first example selects one of the representative values ​​of the phase difference spectra of each predetermined quadrant based on which quadrant Y(k) is in, and obtains it as the phase difference spectrum φ(k) (step S122-A).

[0022] Specifically, the phase difference spectrum estimation unit 122 obtains a predetermined representative value of the phase difference spectrum of the first quadrant as the phase difference spectrum φ(k) if Y(k) is in the first quadrant of the complex plane, obtains a predetermined representative value of the phase difference spectrum of the second quadrant as the phase difference spectrum φ(k) if Y(k) is in the second quadrant of the complex plane, obtains a predetermined representative value of the phase difference spectrum of the third quadrant as the phase difference spectrum φ(k) if Y(k) is in the third quadrant of the complex plane, and obtains a predetermined representative value of the phase difference spectrum of the fourth quadrant as the phase difference spectrum φ(k) if Y(k) is in the fourth quadrant of the complex plane.

[0023] The representative values ​​for each quadrant are predetermined and stored in the representative value storage unit 1221 within the phase difference spectrum estimation unit 122. The representative values ​​of the phase difference spectrum for each quadrant are the estimated values ​​of the phase difference spectrum for each quadrant, and are, for example, as shown in Figure 3, the complex values ​​of points that lie on the circumference of the unit circle in the complex plane and whose argument in the complex plane is the median of the range of argument angles for each quadrant.

[0024] The argument range for the first quadrant is from 0 to π / 2. Therefore, the representative value for the first quadrant is, for example, the value of a point on the circumference of the unit circle on the complex plane with an argument of π / 4, specifically, the value whose real part is cos(π / 4) and imaginary part is sin(π / 4). The argument range for the second quadrant is from π / 2 to π. Therefore, the representative value for the second quadrant is, for example, the value of a point on the circumference of the unit circle on the complex plane with an argument of 3π / 4, specifically, the value whose real part is cos(3π / 4) and imaginary part is sin(3π / 4). The argument range for the third quadrant is from π to 3π / 2. Therefore, the representative value for the third quadrant is, for example, the value of a point on the circumference of the unit circle on the complex plane with an argument of 5π / 4, specifically, the value whose real part is cos(5π / 4) and imaginary part is sin(5π / 4). Since the argument range for the fourth quadrant is from 3π / 2 to 2π, the representative value for the fourth quadrant is, for example, the value of a point on the circumference of the unit circle on the complex plane with an argument of 7π / 4, specifically the value whose real part is cos(7π / 4) and imaginary part is sin(7π / 4).

[0025] The quadrant of the complex plane in which Y(k) lies can be determined by the combination of the signs of u(k) and v(k). Specifically, if both the signs of u(k) and v(k) are positive, Y(k) lies in the first quadrant of the complex plane; if the sign of u(k) is negative and the sign of v(k) is positive, Y(k) lies in the second quadrant; if both the signs of u(k) and v(k) are negative, Y(k) lies in the third quadrant; and if the sign of u(k) is positive and the sign of v(k) is negative, Y(k) lies in the fourth quadrant of the complex plane. Therefore, the phase difference spectrum estimation unit 122 should obtain a predetermined representative value of the phase difference spectrum in the first quadrant as the phase difference spectrum φ(k) if the signs of u(k) and v(k) are both positive values; obtain a predetermined representative value of the phase difference spectrum in the second quadrant as the phase difference spectrum φ(k) if the sign of u(k) is negative and the sign of v(k) is positive; obtain a predetermined representative value of the phase difference spectrum in the third quadrant as the phase difference spectrum φ(k) if the signs of u(k) and v(k) are both negative values; and obtain a predetermined representative value of the phase difference spectrum in the fourth quadrant as the phase difference spectrum φ(k) if the sign of u(k) is positive and the sign of v(k) is negative. Furthermore, if the sign indicating whether u(k) and v(k) are positive or negative is included as a single bit at a predetermined position (for example, the leading bit) in the bit strings of u(k) and v(k) represented by a predetermined number of bits, the phase difference spectrum estimation unit 122 can obtain the phase difference spectrum φ(k) by making a determination based only on the two bits: the single bit at the predetermined position in u(k) and the single bit at the predetermined position in v(k).

[0026] Of course, instead of the combination of the signs of u(k) and v(k), the quadrant of the complex plane in which Y(k) lies can also be determined by the combination of whether u(k) and v(k) are positive or negative. Therefore, the phase difference spectrum estimation unit 122 may obtain a representative value of the phase difference spectrum of the predetermined first quadrant as the phase difference spectrum φ(k) if both u(k) and v(k) are positive, obtain a representative value of the phase difference spectrum of the predetermined second quadrant as the phase difference spectrum φ(k) if u(k) is negative and v(k) is positive, obtain a representative value of the phase difference spectrum of the predetermined third quadrant as the phase difference spectrum φ(k) if both u(k) and v(k) are negative, and obtain a representative value of the phase difference spectrum of the predetermined fourth quadrant as the phase difference spectrum φ(k) if u(k) is positive and v(k) is negative. Of course, the phase difference spectrum estimation unit 122 may determine which quadrant of the complex plane Y(k) is in by using the sign of either u(k) or v(k) and whether the other is positive or negative.

[0027] Furthermore, if Y(k) lies on the boundary line of a quadrant in the complex number plane, the phase difference spectrum estimation unit 122 can assume that Y(k) lies in one of the quadrants on either side of the boundary line and proceed with step S122-A. That is, if Y(k) lies on the boundary line of a quadrant in the complex number plane, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum of one of the quadrants on either side of the boundary line as the phase difference spectrum φ(k). When Y(k) lies on the boundary line of a quadrant in the complex number plane, the predetermined representative value of the phase difference spectrum of one of the quadrants on either side of the boundary line can be predetermined and stored in the phase difference spectrum estimation unit 122. Specifically, if Y(k) lies on the boundary line between the first and second quadrants, that is, if u(k) is 0 and v(k) is a positive value, the phase difference spectrum estimation unit 122 can obtain either a predetermined representative value of the phase difference spectrum of the first quadrant or a predetermined representative value of the phase difference spectrum of the second quadrant as the phase difference spectrum φ(k). Similarly, if Y(k) lies on the boundary line between the second and third quadrants, that is, if u(k) is a negative value and v(k) is 0, the phase difference spectrum estimation unit 122 can obtain either a predetermined representative value of the phase difference spectrum of the second quadrant or a predetermined representative value of the phase difference spectrum of the third quadrant as the phase difference spectrum φ(k). Similarly, if Y(k) lies on the boundary line between the third and fourth quadrants, that is, if u(k) is 0 and v(k) is a negative value, the phase difference spectrum estimation unit 122 can obtain either a predetermined representative value of the phase difference spectrum of the third quadrant or a predetermined representative value of the phase difference spectrum of the fourth quadrant as the phase difference spectrum φ(k). Similarly, if Y(k) lies on the boundary line between the fourth and first quadrants, that is, if u(k) is a positive value and v(k) is 0, the phase difference spectrum estimation unit 122 can obtain either a predetermined representative value of the phase difference spectrum of the fourth quadrant or a predetermined representative value of the phase difference spectrum of the first quadrant as the phase difference spectrum φ(k).

[0028] [[Modified example of the first phase difference spectrum estimation unit 122]] In addition to step S122-A, the phase difference spectrum estimation unit 122 may also obtain a predetermined representative value of the phase difference spectrum when Y(k) is on the boundary line of a quadrant in the complex plane as the phase difference spectrum φ(k) (step S122-A2). Specifically, when Y(k) is on the boundary line between the first and second quadrants, that is, when u(k) is 0 and v(k) is a positive value, the phase difference spectrum estimation unit 122 may obtain a predetermined representative value of the phase difference spectrum when Y(k) is on the boundary line between the first and second quadrants as the phase difference spectrum φ(k). Similarly, when Y(k) lies on the boundary line between the second and third quadrants, that is, when u(k) is negative and v(k) is 0, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) lies on the boundary line between the second and third quadrants as the phase difference spectrum φ(k). Similarly, when Y(k) lies on the boundary line between the third and fourth quadrants, that is, when u(k) is 0 and v(k) is negative, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) lies on the boundary line between the third and fourth quadrants as the phase difference spectrum φ(k). Similarly, when Y(k) lies on the boundary line between the fourth quadrant and the first quadrant, that is, when u(k) is positive and v(k) is 0, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) lies on the boundary line between the fourth quadrant and the first quadrant as the phase difference spectrum φ(k).

[0029] Each representative value of the phase difference spectrum when it lies on the boundary line of a quadrant is predetermined and stored in the representative value storage unit 1221 within the phase difference spectrum estimation unit 122. When Y(k) lies on the boundary line between the first and second quadrants, the representative value of the phase difference spectrum is, for example, the value of a point on the circumference of a unit circle with an argument of π / 2 on the complex plane, with a real part of 0 and an imaginary part of 1. When Y(k) lies on the boundary line between the second and third quadrants, the representative value of the phase difference spectrum is, for example, the value of a point on the circumference of a unit circle with an argument of π on the complex plane, with a real part of -1 and an imaginary part of 0. When Y(k) lies on the boundary line between the third and fourth quadrants, the representative value of the phase difference spectrum is, for example, the value of a point on the circumference of a unit circle with an argument of 3π / 2 on the complex plane, with a real part of 0 and an imaginary part of -1. When Y(k) lies on the boundary between the fourth and first quadrants, the representative value of the phase difference spectrum is, for example, the value at a point on the circumference of the unit circle in the complex plane with an argument of 0, where the real part is 1 and the imaginary part is 0.

[0030] [[Second example of the phase difference spectrum estimation unit 122]] The phase difference spectrum estimation unit 122 in the first example has a maximum error of π / 4 in the phase difference spectrum angle. The phase difference spectrum estimation unit 122 in the second example estimates the phase difference spectrum with less error than the phase difference spectrum estimation unit 122 in the first example. The phase difference spectrum estimation unit 122 in the second example selects one of the representative values ​​of the phase difference spectrum of the real axis half region of each quadrant and the representative value of the phase difference spectrum of the imaginary axis half region of each quadrant, based on which quadrant Y(k) is in and whether Y(k) is in the real axis half region or the imaginary axis half region of that quadrant, and obtains it as the phase difference spectrum φ(k) (step S122-B).

[0031] Specifically, the phase difference spectrum estimation unit 122 obtains a representative value of the phase difference spectrum of a predetermined half-region on the real axis side of the first quadrant of the complex plane as the phase difference spectrum φ(k) if Y(k) is in the real axis side half-region of the first quadrant of the complex plane; obtains a representative value of the phase difference spectrum of a predetermined half-region on the imaginary axis side of the first quadrant as the phase difference spectrum φ(k) if Y(k) is in the real axis side half-region of the second quadrant of the complex plane as the phase difference spectrum φ(k) if Y(k) is in the real axis side half-region of the second quadrant of the complex plane; and obtains a representative value of the phase difference spectrum of a predetermined half-region on the imaginary axis side of the second quadrant as the phase difference spectrum φ(k) if Y(k) is in the imaginary axis side half-region of the second quadrant of the complex plane. If Y(k) is in the real-axis half of the third quadrant of the complex plane, the representative value of the phase difference spectrum of the predetermined real-axis half of the third quadrant is obtained as the phase difference spectrum φ(k). If Y(k) is in the imaginary-axis half of the third quadrant of the complex plane, the representative value of the phase difference spectrum of the predetermined imaginary-axis half of the third quadrant is obtained as the phase difference spectrum φ(k). If Y(k) is in the real-axis half of the fourth quadrant of the complex plane, the representative value of the phase difference spectrum of the predetermined real-axis half of the fourth quadrant is obtained as the phase difference spectrum φ(k). If Y(k) is in the imaginary-axis half of the fourth quadrant of the complex plane, the representative value of the phase difference spectrum of the predetermined imaginary-axis half of the fourth quadrant is obtained as the phase difference spectrum φ(k).

[0032] Representative values ​​of the phase difference spectrum for each region are predetermined and stored in the representative value storage unit 1221 within the phase difference spectrum estimation unit 122. Since the representative values ​​of the phase difference spectrum for each region are estimated values ​​of the phase difference spectrum for each region, for example, for the first quadrant, as shown in Figure 4, they are the complex values ​​of points that lie on the circumference of the unit circle in the complex plane and whose argument in the complex plane is the median value of the range of argument angles for each region.

[0033] Since the argument angle range for the real-axis half of the first quadrant is from 0 to π / 4, a representative value of the phase difference spectrum for the real-axis half of the first quadrant is, for example, the value of a point on the circumference of a unit circle on the complex plane with an argument angle of π / 8, specifically the value where the real part is cos(π / 8) and the imaginary part is sin(π / 8). Since the argument angle range for the imaginary-axis half of the first quadrant is from π / 4 to π / 2, a representative value of the phase difference spectrum for the imaginary-axis half of the first quadrant is, for example, the value of a point on the circumference of a unit circle on the complex plane with an argument angle of 3π / 8, specifically the value where the real part is cos(3π / 8) and the imaginary part is sin(3π / 8).

[0034] Since the argument angle range for the real-axis half of the second quadrant is from 3π / 4 to π, a representative value of the phase difference spectrum for the real-axis half of the second quadrant is, for example, the value of a point on the circumference of the unit circle on the complex plane with an argument angle of 7π / 8, specifically the value where the real part is cos(7π / 8) and the imaginary part is sin(7π / 8). Since the argument angle range for the imaginary-axis half of the second quadrant is from π / 2 to 3π / 4, a representative value of the phase difference spectrum for the imaginary-axis half of the second quadrant is, for example, the value of a point on the circumference of the unit circle on the complex plane with an argument angle of 5π / 8, specifically the value where the real part is cos(5π / 8) and the imaginary part is sin(5π / 8).

[0035] Since the argument angle range for the real-axis half of the third quadrant is from π to 5π / 4, a representative value of the phase difference spectrum for the real-axis half of the third quadrant is, for example, the value of a point on the circumference of the unit circle in the complex plane with an argument angle of 9π / 8, specifically the value where the real part is cos(9π / 8) and the imaginary part is sin(9π / 8). Since the argument angle range for the imaginary-axis half of the third quadrant is from 5π / 4 to 3π / 2, a representative value of the phase difference spectrum for the imaginary-axis half of the third quadrant is, for example, the value of a point on the circumference of the unit circle in the complex plane with an argument angle of 11π / 8, specifically the value where the real part is cos(11π / 8) and the imaginary part is sin(11π / 8).

[0036] Since the argument angle range for the real-axis half of the fourth quadrant is from 7π / 4 to 2π, a representative value of the phase difference spectrum for the real-axis half of the fourth quadrant is, for example, the value of a point on the circumference of the unit circle with an argument angle of 15π / 8 on the complex plane, specifically the value where the real part is cos(15π / 8) and the imaginary part is sin(15π / 8). Since the argument angle range for the imaginary-axis half of the fourth quadrant is from 3π / 2 to 7π / 4, a representative value of the phase difference spectrum for the imaginary-axis half of the fourth quadrant is, for example, the value of a point on the circumference of the unit circle with an argument angle of 13π / 8 on the complex plane, specifically the value where the real part is cos(13π / 8) and the imaginary part is sin(13π / 8).

[0037] Regardless of which quadrant Y(k) is in, if the absolute value of the real part of Y(k) |u(k)| is greater than the absolute value of the imaginary part of Y(k) |v(k)|, then Y(k) is in the real-axis half of the quadrant. If the absolute value of the real part of Y(k) |u(k)| is less than the absolute value of the imaginary part of Y(k) |v(k)|, then Y(k) is in the imaginary-axis half of the quadrant. In other words, whether Y(k) is in the real-axis half of the quadrant or the imaginary-axis half of the quadrant can be determined by which of |u(k)| or |v(k)| is greater. Therefore, the phase difference spectrum estimation unit 122 of the second example only needs to obtain one of the representative values ​​of the phase difference spectrum of the real axis half of each quadrant and the representative value of the phase difference spectrum of the imaginary axis half of each quadrant as the phase difference spectrum φ(k) based on which quadrant Y(k) is in and which of the absolute value of the real part of Y(k) |u(k)| and the absolute value of the imaginary part of Y(k) |v(k)| is larger.

[0038] Specifically, the phase difference spectrum estimation unit 122 obtains a representative value of the phase difference spectrum for a predetermined half-region on the real axis side of the first quadrant as the phase difference spectrum φ(k) if Y(k) is in the first quadrant of the complex plane and |u(k)| is greater than |v(k)|, and if Y(k) is in the first quadrant of the complex plane and |u(k)| is less than |v(k)|, it obtains a representative value of the phase difference spectrum for a predetermined half-region on the imaginary axis side of the first quadrant. The phase difference spectrum φ(k) is obtained, and if Y(k) is in the second quadrant of the complex plane and |u(k)| is greater than |v(k)|, the representative value of the phase difference spectrum of the predetermined half of the second quadrant on the real axis side is obtained as the phase difference spectrum φ(k), and if Y(k) is in the second quadrant of the complex plane and |u(k)| is less than |v(k)|, the representative value of the phase difference spectrum of the predetermined half of the second quadrant on the imaginary axis side is obtained as the phase difference spectrum If Y(k) is in the third quadrant of the complex plane and |u(k)| is greater than |v(k)|, the representative value of the phase difference spectrum of the predetermined half of the real axis region of the third quadrant is obtained as the phase difference spectrum φ(k). If Y(k) is in the third quadrant of the complex plane and |u(k)| is less than |v(k)|, the representative value of the phase difference spectrum of the predetermined half of the imaginary axis region of the third quadrant is obtained as the phase difference spectrum φ(k). The phase difference spectrum estimator 122 can determine which quadrant of the complex plane Y(k) is in, in the same way as the phase difference spectrum estimator 122 in the first example.In other words, the phase difference spectrum estimation unit 122 can determine which quadrant of the complex plane Y(k) is in based on the sign of u(k) or whether u(k) is positive or negative, and the sign of the imaginary part v(k) or whether the imaginary part v(k) is positive or negative, for example, based on the combination of the signs of u(k) and v(k), or the combination of whether u(k) and v(k) are each positive or negative.

[0039] Furthermore, if a single bit at a predetermined position in the bit strings of u(k) and v(k), which are represented by a predetermined number of bits, represents whether the value is positive or negative, and multiple bits at other positions in the bit strings represent absolute values, then the phase difference spectrum estimation unit 122 can obtain the absolute value of u(k)| |u(k)| and the absolute value of v(k)| |v(k)| by either extracting the multiple bits representing the absolute values ​​of u(k) and v(k), or, if at least one of u(k) and v(k) is negative, replacing the bit representing the negative value with the bit representing the positive value.

[0040] Furthermore, if Y(k) lies on the boundary line of a quadrant in the complex number plane, the phase difference spectrum estimation unit 122 can perform step S122-B by assuming that Y(k) is in one of the quadrants flanking the boundary line, similar to the phase difference spectrum estimation unit 122 in the first example. That is, if Y(k) lies on the boundary line of a quadrant in the complex number plane, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum on the boundary line side of one of the quadrants flanking the boundary line as the phase difference spectrum φ(k). When Y(k) lies on the boundary line of a quadrant in the complex number plane, the predetermined representative value of the phase difference spectrum on the boundary line side of one of the quadrants flanking the boundary line can be predetermined and stored in the phase difference spectrum estimation unit 122. Specifically, when Y(k) lies on the boundary line between the first and second quadrants, that is, when u(k) is 0 and v(k) is a positive value, the phase difference spectrum estimation unit 122 can obtain either a representative value of the phase difference spectrum of the predetermined half-region on the imaginary axis side of the first quadrant or a representative value of the phase difference spectrum of the predetermined half-region on the imaginary axis side of the second quadrant as the phase difference spectrum φ(k). Similarly, when Y(k) lies on the boundary line between the second and third quadrants, that is, when u(k) is a negative value and v(k) is 0, the phase difference spectrum estimation unit 122 can obtain either a representative value of the phase difference spectrum of the predetermined half-region on the real axis side of the second quadrant or a representative value of the phase difference spectrum of the predetermined half-region on the real axis side of the third quadrant as the phase difference spectrum φ(k). Similarly, if Y(k) lies on the boundary line between the third and fourth quadrants, that is, if u(k) is 0 and v(k) is a negative value, the phase difference spectrum estimation unit 122 can obtain either a representative value of the phase difference spectrum of the imaginary axis half of the third quadrant (which is predetermined) or a representative value of the phase difference spectrum of the imaginary axis half of the fourth quadrant (which is predetermined) as the phase difference spectrum φ(k).Similarly, if Y(k) lies on the boundary line between the fourth quadrant and the first quadrant, that is, if u(k) is positive and v(k) is 0, the phase difference spectrum estimation unit 122 can obtain either a representative value of the phase difference spectrum of the predetermined half of the fourth quadrant on the real axis side or a representative value of the phase difference spectrum of the predetermined half of the first quadrant on the real axis side as the phase difference spectrum φ(k).

[0041] Furthermore, if Y(k) lies on the boundary line between the real-axis half and the imaginary-axis half of a given quadrant, the phase difference spectrum estimation unit 122 can assume that Y(k) is located in either of the regions straddling the boundary line of that quadrant and proceed with step S122-B. If the phase difference spectrum estimation unit 122 makes a decision based on |u(k)| and |v(k)|, if |u(k)| and |v(k)| are the same value, it can either obtain a representative value of the phase difference spectrum of the predetermined real-axis half region as the phase difference spectrum φ(k), similar to the case where |u(k)| is greater than |v(k)|, or obtain a representative value of the phase difference spectrum of the predetermined imaginary-axis half region as the phase difference spectrum φ(k), similar to the case where |u(k)| is less than |v(k)|. In other words, the phase difference spectrum estimation unit 122 can either perform step S122-B by reinterpreting the above-mentioned "when |u(k)| is greater than |v(k)|" as "when |u(k)| is greater than or equal to |v(k)|", or by reinterpreting the above-mentioned "when |u(k)| is less than |v(k)|" as "when |u(k)| is less than or equal to |v(k)|". The choice of which reinterpretation to perform can be predetermined and stored in the phase difference spectrum estimation unit 122.

[0042] Specifically, if Y(k) is in the first quadrant of the complex plane and |u(k)| and |v(k)| are the same value, the phase difference spectrum estimation unit 122 can obtain either a representative value of the phase difference spectrum for the predetermined half of the first quadrant on the real axis side or a representative value of the phase difference spectrum for the predetermined half of the first quadrant on the imaginary axis side as the phase difference spectrum φ(k). Similarly, if Y(k) is in the second quadrant of the complex plane and |u(k)| and |v(k)| are the same value, the phase difference spectrum estimation unit 122 can obtain either a representative value of the phase difference spectrum for the predetermined half of the second quadrant on the real axis side or a representative value of the phase difference spectrum for the predetermined half of the second quadrant on the imaginary axis side as the phase difference spectrum φ(k). Similarly, if Y(k) is in the third quadrant of the complex plane and |u(k)| and the value |v(k)| are the same, the phase difference spectrum estimation unit 122 can obtain the phase difference spectrum φ(k) from either a representative value of the phase difference spectrum in the predetermined half of the third quadrant on the real axis side or a representative value of the phase difference spectrum in the predetermined half of the third quadrant on the imaginary axis side. Similarly, if Y(k) is in the fourth quadrant of the complex plane and |u(k)| and |v(k)| are the same, the phase difference spectrum estimation unit 122 can obtain the phase difference spectrum φ(k) from either a representative value of the phase difference spectrum in the predetermined half of the fourth quadrant on the real axis side or a representative value of the phase difference spectrum in the predetermined half of the fourth quadrant on the imaginary axis side.

[0043] [[A modified example of the second phase difference spectrum estimation unit 122]] In addition to step S122-B, the phase difference spectrum estimation unit 122 may, when Y(k) lies on the boundary line of a quadrant in the complex plane, obtain a predetermined representative value of the phase difference spectrum when Y(k) lies on the boundary line of a quadrant as the phase difference spectrum φ(k), similar to the phase difference spectrum estimation unit 122 in the modified example of the first example (step S122-B2).

[0044] In addition to step S122-B, or in addition to step S122-B and step S122-B2, the phase difference spectrum estimation unit 122 may obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) lies on the boundary line between the real axis half and the imaginary axis half of a quadrant, as the phase difference spectrum φ(k) (step S122-B3). That is, if the phase difference spectrum estimation unit 122 makes a determination based on |u(k)| and |v(k)|, and |u(k)| and |v(k)| are the same value, it may obtain a predetermined representative value of the phase difference spectrum for the case where |u(k)| and |v(k)| are the same value, as the phase difference spectrum φ(k).

[0045] Specifically, if Y(k) is in the first quadrant of the complex plane and |u(k)| and |v(k)| have the same value, the phase difference spectrum estimation unit 122 should obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) is in the first quadrant of the complex plane and |u(k)| and |v(k)| have the same value as the phase difference spectrum φ(k). Similarly, if Y(k) is in the second quadrant of the complex plane and |u(k)| and |v(k)| have the same value, the phase difference spectrum estimation unit 122 should obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) is in the second quadrant of the complex plane and |u(k)| and |v(k)| have the same value as the phase difference spectrum φ(k). Similarly, if Y(k) is in the third quadrant of the complex plane and |u(k)| and |v(k)| have the same value, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) is in the third quadrant of the complex plane and |u(k)| and |v(k)| have the same value as the phase difference spectrum φ(k). Similarly, if Y(k) is in the fourth quadrant of the complex plane and |u(k)| and |v(k)| have the same value, the phase difference spectrum estimation unit 122 can obtain a predetermined representative value of the phase difference spectrum for the case where Y(k) is in the fourth quadrant of the complex plane and |u(k)| and |v(k)| have the same value as the phase difference spectrum φ(k).

[0046] A predetermined representative value of the phase difference spectrum when Y(k) lies on the boundary line between the real-axis half and the imaginary-axis half of each quadrant, that is, the representative value of the phase difference spectrum when |u(k)| and |v(k)| are the same for each quadrant, is predetermined and stored in the representative value storage unit 1221 within the phase difference spectrum estimation unit 122. The representative value of the phase difference spectrum when Y(k) lies on the boundary line between the real-axis half and the imaginary-axis half of the first quadrant, that is, the representative value of the phase difference spectrum when Y(k) is in the first quadrant of the complex plane and |u(k)| and |v(k)| are the same, is, for example, the value of a point on the circumference of a unit circle with an argument of π / 4 on the complex plane, specifically, a value whose real part is cos(π / 4) and imaginary part is sin(π / 4). A representative value of the phase difference spectrum when Y(k) lies on the boundary line between the real axis half and the imaginary axis half of the second quadrant, that is, when Y(k) is in the second quadrant of the complex plane and |u(k)| and |v(k)| are the same value, is, for example, the value of a point on the circumference of the unit circle in the complex plane with an argument of 3π / 4, specifically the value whose real part is cos(3π / 4) and imaginary part is sin(3π / 4). A representative value of the phase difference spectrum when Y(k) lies on the boundary between the real-axis half and the imaginary-axis half of the third quadrant, that is, when Y(k) is in the third quadrant of the complex plane and |u(k)| and |v(k)| are the same value, is, for example, the value of a point on the circumference of the unit circle in the complex plane with an argument of 5π / 4, specifically the value where the real part is cos(5π / 4) and the imaginary part is sin(5π / 4). A representative value of the phase difference spectrum when Y(k) lies on the boundary line between the real-axis half and the imaginary-axis half of the fourth quadrant, that is, when Y(k) is in the fourth quadrant of the complex plane and |u(k)| and |v(k)| are the same value, is, for example, the value of a point on the circumference of the unit circle in the complex plane with an argument of 7π / 4, specifically the value where the real part is cos(7π / 4) and the imaginary part is sin(7π / 4).

[0047] [[Third example of the phase difference spectrum estimation unit 122]] The declination of the phase difference spectrum estimated by the phase difference spectrum estimation unit 122 of the second example has an error of up to π / 8. In the second example, each quadrant is divided into two regions, i.e., the region on the real axis side and the region on the imaginary axis side, and the declination range of the region of Y(k) corresponding to each representative value of the phase difference spectrum is π / 4. However, in order to reduce the error of the declination of the estimated phase difference spectrum, each quadrant may be divided into three or more regions, and the declination range of the region of Y(k) corresponding to each representative value of the phase difference spectrum may be made narrower. The phase difference spectrum estimation unit 122 of the third example can estimate the phase difference spectrum φ(k) with less error than the phase difference spectrum estimation unit 122 of the second example when N, which is an integer of 2 or more, is set as the number of divisions of each quadrant and N is 3 or more. Hereinafter, it will be described assuming that n is each integer from 1 to 4N.

[0048] When the declination θ of Y(k) is greater than (n - 1)π / 2N and less than nπ / 2N (that is, when (n - 1)π / 2N < θ < nπ / 2N), the phase difference spectrum estimation unit 122 of the third example obtains the representative value of the predetermined phase difference spectrum in the case of (n - 1)π / 2N < θ < nπ / 2N as the phase difference spectrum φ(k) (step S122-C). Each representative value of the phase difference spectrum is stored in the representative value storage unit 1221 that is predetermined and exists in the phase difference spectrum estimation unit 122. The representative value of the phase difference spectrum in the case of (n - 1)π / 2N < θ < nπ / 2N is, for example, the value of a point on the circumference of the unit circle whose declination on the complex number plane is (2n - 1)π / 4N. Specifically, the real part is cos((2n - 1)π / 4N) and the imaginary part is sin((2n - 1)π / 4N). The declination (2n - 1)π / 4N on the complex number plane is the median value of the declination range from (n - 1)π / 2N to nπ / 2N on the complex number plane.

[0049] In the first quadrant, when the argument θ of Y(k) is greater than (n-1)π / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is greater than the product of the absolute value of the real part of Y(k) |u(k)| and tan((n-1)π / 2N). When the argument θ of Y(k) is less than nπ / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is smaller than the product of the absolute value of the real part of Y(k) |u(k)| and tan(nπ / 2N). Therefore, the phase difference spectrum estimation unit 122 should obtain a value as the phase difference spectrum φ(k) when Y(k) is in the first quadrant, and |v(k)| is greater than the product of |u(k)| and tan((n-1)π / 2N), and |v(k)| is smaller than the product of |u(k)| and tan(nπ / 2N) (i.e., when Y(k) is in the first quadrant and |u(k)|×tan((n-1)π / 2N)<|v(k)|<|u(k)|×tan(nπ / 2N)). In this case, the real part is cos((2n-1)π / 4N) and the imaginary part is sin((2n-1)π / 4N). Naturally, instead of comparing |u(k)| multiplied by its tangent value with |v(k)|, one may use the reciprocal of the tangent value to compare |u(k)| with the product of |v(k)| and its reciprocal. This also applies to subsequent comparisons.

[0050] In the second quadrant, when the argument θ of Y(k) is greater than (n-1)π / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is smaller than the product of the absolute value of the real part of Y(k) |u(k)| and |tan((n-1)π / 2N)|. When the argument θ of Y(k) is less than nπ / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is larger than the product of the absolute value of the real part of Y(k) |u(k)| and |tan(nπ / 2N)|. Therefore, the phase difference spectrum estimation unit 122 should obtain a value as the phase difference spectrum φ(k) where the real part is cos((2n-1)π / 4N) and the imaginary part is sin((2n-1)π / 4N) when Y(k) is in the second quadrant and |v(k)| is smaller than the product of |u(k)| and |tan((n-1)π / 2N)|, and |v(k)| is larger than the product of |u(k)| and |tan(nπ / 2N)| (i.e., when Y(k) is in the second quadrant and |u(k)|×|tan((n-1)π / 2N)|>|v(k)|>|u(k)|×|tan(nπ / 2N)|). Of course, instead of comparing |u(k)| multiplied by the absolute value of the tangent with |v(k)|, one may use the absolute value of the reciprocal of the tangent and compare |u(k)| with the product of |v(k)| multiplied by the absolute value of the reciprocal of the tangent. This also applies to subsequent comparisons.

[0051] In the third quadrant, when the argument θ of Y(k) is greater than (n-1)π / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is greater than the product of the absolute value of the real part of Y(k) |u(k)| and |tan((n-1)π / 2N)|. When the argument θ of Y(k) is less than nπ / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is smaller than the product of the absolute value of the real part of Y(k) |u(k)| and |tan(nπ / 2N)|. Therefore, the phase difference spectrum estimation unit 122 should obtain a value as the phase difference spectrum φ(k) when Y(k) is in the third quadrant, and |v(k)| is greater than the product of |u(k)| and |tan((n-1)π / 2N)|, and |v(k)| is smaller than the product of |u(k)| and |tan(nπ / 2N)| (i.e., when Y(k) is in the third quadrant, and |u(k)|×|tan((n-1)π / 2N)|<|v(k)|<|u(k)|×|tan(nπ / 2N)|), where the real part is cos((2n-1)π / 4N) and the imaginary part is sin((2n-1)π / 4N).

[0052] In the fourth quadrant, when the argument θ of Y(k) is greater than (n-1)π / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is smaller than the product of the absolute value of the real part of Y(k) |u(k)| and |tan((n-1)π / 2N)|. When the argument θ of Y(k) is less than nπ / 2N, the absolute value of the imaginary part of Y(k) |v(k)| is larger than the product of the absolute value of the real part of Y(k) |u(k)| and |tan(nπ / 2N)|. Therefore, the phase difference spectrum estimation unit 122 should obtain a value as the phase difference spectrum φ(k) where the real part is cos((2n-1)π / 4N) and the imaginary part is sin((2n-1)π / 4N) when Y(k) is in the fourth quadrant, and |v(k)| is smaller than the product of |u(k)| and |tan((n-1)π / 2N)|, and |v(k)| is larger than the product of |u(k)| and |tan(nπ / 2N)| (i.e., when Y(k) is in the fourth quadrant, and |u(k)|×|tan((n-1)π / 2N)|>|v(k)|>|u(k)|×|tan(nπ / 2N)|).

[0053] The phase difference spectrum estimation unit 122 may determine in which quadrant of the complex plane Y(k) lies in the same manner as the phase difference spectrum estimation unit 122 in the first example. That is, the phase difference spectrum estimation unit 122 may determine in which quadrant of the complex plane Y(k) lies based on, for example, the sign of u(k) or whether u(k) is a positive value or a negative value, and the sign of the imaginary part v(k) or whether the imaginary part v(k) is a positive value or a negative value, i.e., based on the combination of the sign of u(k) and the sign of v(k), or the combination of whether each of u(k) and v(k) is a positive value or a negative value.

[0054] When Y(k) lies on the boundary line of the region, the phase difference spectrum estimation unit 122 may perform step S122-C by regarding Y(k) as being in either one of the regions sandwiching the boundary line. That is, when the argument θ of Y(k) is greater than (n - 1)π / 2N and less than or equal to nπ / 2N (i.e., when (n - 1)π / 2N < θ ≤ nπ / 2N), the phase difference spectrum estimation unit 122 may obtain the representative value of the predetermined phase difference spectrum for the case of (n - 1)π / 2N < θ ≤ nπ / 2N as the phase difference spectrum φ(k), or when the argument θ of Y(k) is greater than or equal to (n - 1)π / 2N and less than nπ / 2N (i.e., when (n - 1)π / 2N ≤ θ < nπ / 2N), the phase difference spectrum estimation unit 122 may obtain the representative value of the predetermined phase difference spectrum for the case of (n - 1)π / 2N ≤ θ < nπ / 2N as the phase difference spectrum φ(k).

[0055] Specifically, the phase difference spectrum estimation unit 122 estimates the value of cos((2n-1)π / 4N) with a real part of cos((2n-1)π / 4N) and an imaginary part of sin((2n-1)π / 4N) when Y(k) is in the first quadrant or on the boundary line between the first and second quadrants, and |u(k)|×tan((n-1)π / 2N)<|v(k)|≦|u(k)|×tan(nπ / 2N) as the phase difference spectrum. If we obtain a vector φ(k) and Y(k) is in the second quadrant or on the boundary line between the second and third quadrants, and |u(k)|×|tan((n-1)π / 2N)|>|v(k)|≧|u(k)|×|tan(nπ / 2N)|, then the value of which the real part is cos((2n-1)π / 4N) and the imaginary part is sin((2n-1)π / 4N) is the phase difference spectrum φ( If we obtain Y(k) as k, and Y(k) is in the third quadrant or on the boundary between the third and fourth quadrants, and |u(k)|×|tan((n-1)π / 2N)|<|v(k)|≦|u(k)|×|tan(nπ / 2N)|, then the phase difference spectrum φ(k) is obtained as a value with real part cos((2n-1)π / 4N) and imaginary part sin((2n-1)π / 4N). If Y(k) is in the fourth quadrant or on the boundary line between the fourth and first quadrants, and |u(k)|×|tan((n-1)π / 2N)|>|v(k)|≧|u(k)|×|tan(nπ / 2N)|, then the phase difference spectrum φ(k) can be obtained by finding a value whose real part is cos((2n-1)π / 4N) and imaginary part is sin((2n-1)π / 4N).

[0056] Alternatively, the phase difference spectrum estimation unit 122 calculates the value of the phase difference spectrum if Y(k) is in the first quadrant or on the boundary line between the fourth quadrant and the first quadrant, and |u(k)|×tan((n-1)π / 2N)≦|v(k)|<|u(k)|×tan(nπ / 2N), where the real part is cos((2n-1)π / 4N) and the imaginary part is sin((2n-1)π / 4N). If we obtain a phase difference spectrum φ(k) such that Y(k) is in the second quadrant or on the boundary line between the first and second quadrants, and |u(k)|×|tan((n-1)π / 2N)|≧|v(k)|>|u(k)|×|tan(nπ / 2N)|, then the value of the phase difference spectrum φ(k) has a real part of cos((2n-1)π / 4N) and an imaginary part of sin((2n-1)π / 4N). If Y(k) is in the third quadrant or on the boundary line between the second and third quadrants, and |u(k)|×|tan((n-1)π / 2N)|≦|v(k)|<|u(k)|×|tan(nπ / 2N)|, then the phase difference spectrum φ(k) is obtained with a real part of cos((2n-1)π / 4N) and an imaginary part of sin((2n-1)π / 4N). If Y(k) is in the fourth quadrant or on the boundary line between the third and fourth quadrants, and |u(k)|×|tan((n-1)π / 2N)|≧|v(k)|>|u(k)|×|tan(nπ / 2N)|, then the phase difference spectrum φ(k) can be obtained by finding a value whose real part is cos((2n-1)π / 4N) and imaginary part is sin((2n-1)π / 4N).

[0057] [[Third example of modified phase difference spectrum estimation unit 122]] In addition to step S122-C, the phase difference spectrum estimation unit 122 may also obtain a predetermined representative value of the phase difference spectrum when Y(k) is on the boundary line of the region as the phase difference spectrum φ(k) (step S122-C2). That is, the phase difference spectrum estimation unit 122 may also obtain a predetermined representative value of the phase difference spectrum when the argument angle θ of Y(k) is nπ / 2N as the phase difference spectrum φ(k). Specifically, the phase difference spectrum estimation unit 122 may obtain a value as the phase difference spectrum φ(k) when |u(k)|×tan(nπ / 2N)=|v(k)| has a real part of cos(nπ / 2N) and an imaginary part of sin(nπ / 2N).

[0058] [[Fourth example of the phase difference spectrum estimation unit 122]] In the fourth example, we will explain an example in which the phase difference spectrum is estimated using binary search within the quadrant. However, for convenience, we will also include the case in which no search is performed within the quadrant. In the following, P is the number of times the binary search is performed and is a predetermined integer of 0 or more. For example, in the fourth example, the phase difference spectrum estimation unit 122 does not perform a search within the quadrant if P=0 (i.e., does not perform any binary search), performs one binary search if P=1, and performs two binary searches if P=2. P may be a different value for each frequency k, or it may be the same value for all frequencies.

[0059] In the fourth example, the phase difference spectrum estimation unit 122 searches for which quadrant Y(k) is located in. If P=0, it obtains a predetermined representative value of the phase difference spectrum for the quadrant in which Y(k) is located as the phase difference spectrum φ(k). If P≠0, it performs a binary search of the range of the argument angle P times for the quadrant in which Y(k) is located to identify the range of the argument angle in which Y(k) is located, and obtains a predetermined representative value of the phase difference spectrum for the identified range of the argument angle as the phase difference spectrum φ(k) (step S122-D). Each representative value of the phase difference spectrum is predetermined and stored in the representative value storage unit 1221 located within the phase difference spectrum estimation unit 122.

[0060] The representative value of the phase difference spectrum for each quadrant is, for example, the complex value of a point on the circumference of the unit circle where the argument of the complex number plane is the median of the range of argument angles for that quadrant. Specifically, it is a value where the real part is the cosine of the median of the range of argument angles for that quadrant and the imaginary part is the sine of the median of the range of argument angles for that quadrant. The representative value of the phase difference spectrum for each range of argument angles is, for example, the complex value of a point on the circumference of the unit circle where the argument of the complex number plane is the median of the range. Specifically, it is a value where the real part is the cosine of the median of the range of argument angles for that quadrant and the imaginary part is the sine of the median of the range of argument angles for that quadrant.

[0061] In each quadrant, the frequency distribution of the phase difference spectrum's argument angle may be biased depending on the relationship and frequency of the signals from the two channels. Therefore, it is possible to set a representative value for the phase difference spectrum while taking into account the bias in the frequency distribution of the argument angle. That is, it is not necessary for the representative value of the phase difference spectrum for each quadrant to be the complex value of a point on the circumference of the unit circle where the argument angle in the complex plane is the median of the range of the quadrant's argument angle. Rather, the representative value of the phase difference spectrum for each quadrant should be the complex value of a predetermined point on the circumference of the unit circle where the argument angle in the complex plane is within the range of the quadrant's argument angle. Specifically, the real part should be the cosine of the representative value of the argument angle within the quadrant's argument angle range, and the imaginary part should be the sine of the representative value of the argument angle within the quadrant's argument angle range. Similarly, it is not necessary for the representative value of the phase difference spectrum for each range of the argument to be the complex value of a point on the circumference of the unit circle where the argument in the complex plane is the median of the range of the argument. Rather, the representative value of the phase difference spectrum for each range of the argument can be any complex value of a predetermined point on the circumference of the unit circle where the argument in the complex plane is within the range of the argument. Specifically, it is a value where the real part is the cosine of the representative value of the argument in the range of the argument and the imaginary part is the sine of the representative value of the argument in the range of the argument.

[0062] For example, if the digital audio signals obtained by AD conversion of the sound picked up by a left channel microphone and a right channel microphone placed in a certain space are the first channel input audio signal and the second channel input audio signal, and the voice spoken by a person present in that space is included in the first channel input audio signal and the second channel input audio signal with a so-called arrival time difference, then the phase difference spectrum is distributed biased towards the real axis on the circumference of the unit circle in the complex plane at low frequencies, and distributed almost uniformly on the circumference of the unit circle in the complex plane without biasing towards a particular angle at medium to high frequencies. From this, for example, the value of the angle of the complex plane of the representative value of the phase difference spectrum of each quadrant should be a value closer to the real axis than the median of the range of the quadrant's angle of determination if the frequency is below a predetermined threshold or less than that threshold, and if the frequency is otherwise (i.e., if the frequency is higher than or equal to the threshold), then the angle of the complex plane should be the median of the range of the quadrant's angle of determination. Alternatively, for example, the argument values ​​in the complex plane of the representative values ​​of the phase difference spectra for each quadrant may be such that, at lower frequencies, the argument in the complex plane is closer to the real axis than the median of the quadrant's argument range, and at higher frequencies, the argument in the complex plane is closer to the median of the quadrant's argument range than the real axis.

[0063] Similarly, for example, the argument value in the complex plane of the representative value of the phase difference spectrum for each range of the argument may be such that, when the frequency is below or less than a predetermined threshold, the argument in the complex plane is closer to the real axis than the median of the range of the argument, and when the frequency is otherwise (i.e., when the frequency is higher or greater than the threshold), the argument in the complex plane is the median of the range of the argument. Alternatively, for example, the argument value in the complex plane of the representative value of the phase difference spectrum for each range of the argument may be such that the argument in the complex plane is closer to the real axis than the median of the range of the argument as the frequency decreases, and the argument in the complex plane is closer to the median of the range of the argument than the real axis as the frequency increases.

[0064] The thresholds mentioned above should be predetermined, for example, so that frequencies below or less than approximately 500 Hz are below or less than the threshold. Furthermore, the thresholds mentioned above should be defined for the sample numbers (sample indices) assigned sequentially from the low-frequency side. Therefore, for example, if the frame length is 20 ms, whether the sampling frequency is 32 kHz and phase difference spectra are obtained for virtually 320 frequencies, whether the sampling frequency is 48 kHz and phase difference spectra are obtained for virtually 480 frequencies, or whether the sampling frequency is 16 kHz and phase difference spectra are obtained for virtually 160 frequencies, that is, regardless of the sampling frequency, if the threshold is 10, when the index is less than or equal to the threshold of 10, the argument value of the complex number plane of the representative value of the phase difference spectrum should be closer to the real axis than the median of the argument range, and when the index is greater than the threshold of 10, the argument value of the complex number plane of the representative value of the phase difference spectrum should be the median of the argument range. Similarly, for example, if the frame length is twice 20ms, which is 40ms, the threshold should be set to 20, and if the frame length is half of 20ms, which is 10ms, the threshold should be set to 5.

[0065] In the specific example described later in step S122-D, the argument of the complex number plane is used as the representative value of the range of the argument of each quadrant, and the absolute value of the tangent, the value of the cosine, and the value of the sine of the representative value of each range of the argument. Therefore, it is preferable that the absolute value of the tangent, the value of the cosine, and the value of the sine of the range of the argument of each quadrant and the representative value of the argument of each range are calculated in advance and stored in the representative value storage unit 1221. Of course, the representative value storage unit 1221 may store complex values ​​in which the real part is the cosine and the imaginary part is the sine, instead of the aforementioned values ​​of the cosine and sine. Also, if the representative value of the phase difference spectrum of each quadrant is the median of the range of the argument of the complex number plane, and the phase difference spectrum estimation unit 122 does not use the absolute value of the tangent of the representative value of the range of the argument of each quadrant, then the absolute value of the tangent of the representative value of the range of the argument of each quadrant does not need to be stored in the representative value storage unit 1221.

[0066] A specific example of step S122-D performed by the phase difference spectrum estimation unit 122 will be explained in the following steps S122-D1 to S122-D6.

[0067] The phase difference spectrum estimation unit 122 first sets p=0 to determine which quadrant of the complex plane Y(k) is in, and obtains a representative value of the argument within the range of the argument in the quadrant in which Y(k) exists (step S122-D1). The determination of which quadrant of the complex plane Y(k) is in can be performed in the same way as in the first example of the phase difference spectrum estimation unit 122. That is, the phase difference spectrum estimation unit 122 can determine which quadrant of the complex plane Y(k) is in based on the sign of u(k) or whether u(k) is positive or negative, and the sign of the imaginary part v(k) or whether the imaginary part v(k) is positive or negative, for example, based on the combination of the signs of u(k) and v(k), or the combination of whether u(k) and v(k) are each positive or negative.

[0068] The phase difference spectrum estimation unit 122, after step S122-D1, if p=P (i.e., P=0), obtains a complex value as the phase difference spectrum φ(k), which is a predetermined representative value of the phase difference spectrum and is the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in step S122-D1. That is, the real part is the cosine of the representative value of the argument obtained in step S122-D1 and the imaginary part is the sine of the representative value of the argument obtained in step S122-D1 (step S122-D2). If P=0, the phase difference spectrum estimation unit 122 terminates the process in step S122-D in step S122-D2. Note that if the phase difference spectrum estimation unit 122 terminates the process in step S122-D in step S122-D2, the same result as the first example of the phase difference spectrum estimation unit 122 will be obtained.

[0069] The phase difference spectrum estimation unit 122, after step S122-D1, if p=P is not true (i.e., p≠P), adds 1 to p and uses that value as the new p (i.e., 1 as the new p), obtaining the range of the argument angle in the quadrant where Y(k) exists as the search range for the next step, and also obtaining the absolute value of the tangent of the representative value of the argument angle obtained in step S122-D1 (i.e., the representative value of the argument angle in the search range for the next step) (step S122-D3).

[0070] The phase difference spectrum estimation unit 122, after step S122-D3 or step S122-D6 described later, determines that if the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in the previous process (i.e., step S122-D3 or step S122-D6 described later) by |u(k)| is greater than |v(k)|, it determines that Y(k) exists in the real axis range of the search range and obtains a representative value of the argument angle of the real axis range of the search range. If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in the previous process by |u(k)| is less than |v(k)|, it determines that Y(k) exists in the imaginary axis range of the search range and obtains a representative value of the argument angle of the imaginary axis range of the search range (step S122-D4).

[0071] Of course, instead of comparing |v(k)| with the product of |u(k)| and the absolute value of the cotangent of the representative angle of the search range, we could also compare |u(k)| with the product of |v(k)| and the absolute value of the cotangent of |v(k)| and the representative angle of the search range. In other words, the phase difference spectrum estimation unit 122 may, after step S122-D3 or step S122-D6 described later, determine that if |u(k)| is greater than the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the previous process by |v(k)|, determine that Y(k) exists in the real axis range of the search range and obtain a representative value of the argument angle of the real axis range of the search range. If |u(k)| is less than the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the previous process by |v(k)|, determine that Y(k) exists in the imaginary axis range of the search range and obtain a representative value of the argument angle of the imaginary axis range of the search range. In this case, since the argument of the complex plane is used, the absolute value of the cotangent of the representative value of the range of the argument is also used. Therefore, it is desirable that the absolute value of the cotangent of the representative value of each range of the argument is calculated in advance and stored in the representative value storage unit 1221. The phase difference spectrum estimation unit 122 should, in the immediately preceding step S122-D3 and the step S122-D6 described later, obtain the absolute value of the cotangent of the representative value of the argument of the search range of the next step instead of the absolute value of the tangent of the representative value of the argument of the search range of the next step.

[0072] Furthermore, the real axis range of the search range refers to the real axis range of the search range when the search range in the complex plane is bisected by a line whose argument is the representative value, and the imaginary axis range of the search range refers to the imaginary axis range of the search range when the search range in the complex plane is bisected by a line whose argument is the representative value. If the representative value of the argument of the search range is the median of the argument of the search range, then the real axis range of the search range refers to the real axis half of the search range when the search range in the complex plane is bisected by a line whose argument is the median, and the imaginary axis range of the search range refers to the imaginary axis half of the search range when the search range in the complex plane is bisected by a line whose argument is the median.

[0073] Furthermore, if the process immediately preceding step S122-D4 is step S122-D3, and the representative value of the argument obtained in step S122-D1 is the median value of the argument, then the absolute value of the tangent of the representative value of the search range obtained in the previous process is always 1. Therefore, the phase difference spectrum estimation unit 122 may use "when |u(k)| is greater than |v(k)|" instead of "when |u(k)| is greater than |v(k)|", or "when |u(k)| is less than |v(k)|" instead of "when |u(k)| is less than |v(k)|", and in the previous process, step S122-D3, it is not necessary to obtain the absolute value of the tangent of the representative value of the argument in the quadrant where Y(k) exists. In other words, when the phase difference spectrum estimation unit 122 is performed after step S122-D3, when the representative value of the argument obtained in step S122-D1 is the median value of the argument, if |u(k)| is greater than |v(k)|, it determines that Y(k) exists in the half range on the real axis side of the search range obtained in step S122-D3, and obtains a representative value of the argument for the half range on the real axis side of the search range obtained in step S122-D3. If |u(k)| is less than |v(k)|, it determines that Y(k) exists in the half range on the imaginary axis side of the search range obtained in step S122-D3, and obtains a representative value of the argument for the half range on the imaginary axis side of the search range obtained in step S122-D3. When the next step is performed after step S122-D6, If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in step S122-D6 by |u(k)| is greater than |v(k)|, it is determined that Y(k) exists in the real axis range of the search range obtained in step S122-D6, and the representative value of the argument angle of the real axis range of the search range obtained in step S122-D6 is obtained. If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in step S122-D6 by |u(k)| is less than |v(k)|, it is determined that Y(k) exists in the imaginary axis range of the search range obtained in step S122-D6, and the process of obtaining the representative value of the argument angle of the imaginary axis range of the search range obtained in step S122-D6 should be performed as step S122-D4.

[0074] If p=P, the phase difference spectrum estimation unit 122, after step S122-D4, obtains a complex value as the phase difference spectrum φ(k) (step S122-D5). This complex value is a predetermined representative value, and the argument of the complex plane is the representative value of the argument obtained in step S122-D4. In other words, the real part is the cosine of the representative value of the argument obtained in step S122-D4 and the imaginary part is the sine of the representative value of the argument obtained in step S122-D4. If p=P, the phase difference spectrum estimation unit 122 terminates the process in step S122-D in step S122-D5.

[0075] The phase difference spectrum estimation unit 122, after step S122-D4, if p=P is not true (i.e., p≠P), adds 1 to p to obtain a new value for p, and uses the range of the argument angle within the range where Y(k) determined in step S122-D4 exists as the search range for the next step. It also obtains the absolute value of the tangent of the representative value of the argument angle obtained in step S122-D4 (i.e., the representative value of the argument angle in the search range for the next step) (step S122-D6). The phase difference spectrum estimation unit 122 then performs step S122-D4 after step S122-D6.

[0076] In step S122-D1, if Y(k) lies on the boundary line of a quadrant in the complex plane, the phase difference spectrum estimation unit 122 may process the data by assuming that Y(k) is in one of the quadrants on either side of the boundary line, similar to the phase difference spectrum estimation unit 122 in the first and second examples. Similarly, if Y(k) lies on the boundary line of the binary search for the range of the argument, the phase difference spectrum estimation unit 122 may process the data by assuming that Y(k) is in one of the ranges on either side of the boundary line. Specifically, in step S122-D4, the phase difference spectrum estimation unit 122 may change the condition from "when the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle in the search range by |u(k)| is greater than |v(k)|" to "when the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle in the search range by |u(k)| is greater than or equal to |v(k)|" or from "when the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle in the search range by |u(k)| is less than |v(k)|" to "when the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle in the search range by |u(k)| is less than or equal to |v(k)|". Similarly, when using the cotangent, the phase difference spectrum estimation unit 122 may, in step S122-D4, replace "when |u(k)| is greater than the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle in the search range by |v(k)|" with "when |u(k)| is greater than or equal to the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle in the search range by |v(k)|" or replace "when |u(k)| is less than the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle in the search range by |v(k)|" with "when |u(k)| is less than or equal to the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle in the search range by |v(k)|".

[0077] [[A modified example of the fourth instance of the phase difference spectrum estimation unit 122]] If Y(k) lies on the boundary line of a quadrant in the complex number plane, the phase difference spectrum estimation unit 122 may obtain a predetermined representative value of the phase difference spectrum when Y(k) lies on the boundary line of a quadrant as the phase difference spectrum φ(k), similar to the phase difference spectrum estimation unit 122 in the modified example of the first example. Specifically, in step S122-D1, the phase difference spectrum estimation unit 122 may also determine whether Y(k) lies on the boundary line of a quadrant in the complex number plane, and if Y(k) lies on the boundary line of a quadrant in the complex number plane, it may obtain a predetermined representative value of the phase difference spectrum when Y(k) lies on the boundary line of a quadrant as the phase difference spectrum φ(k), similar to the phase difference spectrum estimation unit 122 in the modified example of the first example, and then terminate step S122-D.

[0078] Similarly, if Y(k) lies on the boundary line of the binary search for the range of argument angles, the phase difference spectrum estimation unit 122 may obtain a predetermined representative value of the phase difference spectrum when Y(k) lies on the boundary line as the phase difference spectrum φ(k). Specifically, in step S122-D4, the phase difference spectrum estimation unit 122 may also determine whether the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angles of the search range obtained in the previous process by |u(k)| is the same as |v(k)|. If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angles of the search range obtained in the previous process by |u(k)| is the same as |v(k)|, the unit may obtain a complex value as the phase difference spectrum φ(k) where the real part is the cosine of the representative value of the argument angles of the search range obtained in the previous process and the imaginary part is the sine of the representative value of the argument angles of the search range obtained in the previous process, and then terminate step S122-D. Of course, in step S122-D4, the phase difference spectrum estimation unit 122 may, instead of determining whether the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in the previous process by |u(k)| is the same as |v(k)|, also determine whether |u(k)| is the same as the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the previous process by |v(k)|. If |u(k)| is the same as the value obtained by multiplying the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the previous process by |v(k)|, it may obtain a complex value as the phase difference spectrum φ(k) where the real part is the cosine of the representative value of the argument angle of the search range obtained in the previous process and the imaginary part is the sine of the representative value of the argument angle of the search range obtained in the previous process, and then terminate step S122-D.

[0079] [[Fifth example of the phase difference spectrum estimation unit 122]] In the first to fourth examples, the phase difference spectrum estimation unit 122 had a pre-associated relationship between the real and imaginary parts of the product of the complex conjugates of the frequency spectrum of the first channel and the frequency spectrum of the second channel, which represents the argument of the product on the complex plane, for each of the representative values ​​of the multiple phase difference spectra. However, this pre-association is not required. This example will be explained as the fifth example. In the explanation of the fifth example, Q is a predetermined integer of 2 or greater.

[0080] The representative value storage unit 1221 of the phase difference spectrum estimation unit 122 stores Q predetermined candidate values ​​for phase difference spectra. The Q predetermined candidate values ​​for phase difference spectra are values ​​that lie on the circumference of the unit circle in the complex number plane, and their arguments on the complex number plane are different from each other. The Q predetermined candidate values ​​for phase difference spectra may be arranged at equal intervals on the circumference of the unit circle in the complex number plane, or they may be arranged at unequal intervals on the circumference of the unit circle in the complex number plane, with a higher frequency range of arguments taking into account bias in the frequency distribution of the phase difference spectrum's argument angles. Also, similar to the fourth example, Q candidate values ​​for phase difference spectra may be predetermined for each frequency or frequency range, taking into account the differences in the bias of the frequency distribution of the phase difference spectrum's argument angles for each frequency. For example, the Q candidate phase difference spectrum values ​​may be arranged on the circumference of the unit circle in the complex plane such that the closer the argument angle is to the real axis, the closer they are to the real axis, if the frequency is below or less than a predetermined threshold. If the frequency is otherwise (i.e., higher than or equal to the threshold), they may be arranged at equal intervals on the circumference of the unit circle in the complex plane. Alternatively, for example, the Q candidate phase difference spectrum values ​​may be arranged such that the bias of the argument angle from the equal intervals toward the real axis increases as the frequency decreases, and the bias of the argument angle from the equal intervals toward the real axis decreases as the frequency increases. The threshold is the same as in the fourth example.

[0081] The phase difference spectrum estimation unit 122 selects a candidate phase difference spectrum φ(k) from Q predetermined candidate phase difference spectrum values ​​for each frequency k, where the argument on the complex plane of the product Y(k) of the complex conjugate ￣X2(k) of the frequency spectrum X1(k) of the first channel and the frequency spectrum X2(k) of the second channel is closest to the argument on the complex plane of Y(k) (step S122-E).

[0082] Let q be an integer between 1 and Q, let φ(q) be a candidate value for the phase difference spectrum, and let φ(q) be the real part of φ(q).real Let the imaginary part of φ(q) be φ(q) imag Let θ(φ(q)) be the argument of φ(q) in the complex plane. Then tanθ(φ(q)) = φ(q). imag / φ(q) real Therefore, if we let θ(Y(k)) be the argument of Y(k) in the complex plane, then tanθ(Y(k)) = v(k) / u(k). Thus, for each frequency k, the phase difference spectrum estimation unit 122 can select the tanθ(φ(q)) closest to tanθ(Y(k)) from among the Q tangents of tanθ(φ(1)) to tanθ(φ(Q)), and obtain the phase difference spectrum φ(k) corresponding to the selected tanθ(φ(q)). However, in order to avoid generating a large amount of computational processing due to the division required to calculate tanθ(Y(k))=v(k) / u(k), specifically, for example, the representative value storage unit 122 of the phase difference spectrum estimation unit 122 stores tanθ(φ(q)) in advance in association with candidate values ​​φ(q) for each phase difference spectrum, and the phase difference spectrum estimation unit 122 can then obtain the phase difference spectrum φ(k) corresponding to the tanθ(φ(q)) that gives the smallest value of |u(k)×tanθ(φ(q))-v(k)| for each frequency k. Of course, the representative value storage unit 122 of the phase difference spectrum estimation unit 122 may also store in advance the reciprocal of tanθ(φ(q)), which is cotθ(φ(q)), in association with the candidate value φ(q) of each phase difference spectrum, and the phase difference spectrum estimation unit 122 may obtain the phase difference spectrum φ(k) corresponding to the cotθ(φ(q)) that has the smallest value of |u(k)-v(k)×cotθ(φ(q))| for each frequency k. The value of tanθ(φ(q)) or cotθ(φ(q)) used by the phase difference spectrum estimation unit 122 in the above-mentioned processing may also be stored in the representative value storage unit 1221.

[0083] In the fifth example, unlike the representative values ​​in the first to fourth examples, the correspondence between the argument of the product of the complex conjugates of the frequency spectra of the first and second channels on the complex plane has not been established beforehand. Therefore, the values ​​stored in the representative value storage unit 1221 are referred to as candidate values ​​for the phase difference spectrum. However, since it is ultimately possible to identify the correspondence between the candidate values ​​in the fifth example and the argument of the product of the complex conjugates of the frequency spectra of the first and second channels on the complex plane, there is no problem in calling them representative values, just like the representative values ​​in the first to fourth examples.

[0084] [[Summary of Phase Difference Spectrum Estimation Unit 122 (Step S122)]] As explained in the first to fifth examples and the modifications of the first to fourth examples, the phase difference spectrum estimation unit 122 essentially obtains one candidate value of a plurality of predetermined phase difference spectrum candidates as the phase difference spectrum φ(k) for each frequency k, based on the relationship between the real part u(k) and the imaginary part v(k) of the product Y(k) of the complex conjugate ￣X2(k) of the frequency spectrum X1(k) of the first channel and the frequency spectrum X2(k) of the second channel. Here, the plurality of predetermined candidate values ​​of the phase difference spectrum are values ​​that lie on the circumference of the unit circle in the complex number plane, and their arguments on the complex number plane are different from each other. In the first to fourth examples and their modifications, each candidate value of the phase difference spectrum is pre-associated with a range of arguments on the complex number plane of the product of the complex conjugates of the frequency spectrum of the first channel and the frequency spectrum of the second channel. In the first to fourth examples and their variations, these multiple predetermined candidate values ​​for phase difference spectra and the corresponding range of the argument angles described above are stored in the representative value storage unit 1221. In the first to third examples and their variations, the phase difference spectrum estimation unit 122, for each frequency k, uses the relationship between the real part u(k) and the imaginary part v(k) of Y(k), which represent the argument angle of Y(k) in the complex plane, to select one candidate value from among the multiple predetermined candidate values ​​for phase difference spectra, in which the argument angle of Y(k) in the complex plane falls within the range of the argument angle of the product of the complex conjugates of the frequency spectra of the first channel and the second channel, and obtains it as the phase difference spectrum φ(k). Furthermore, in the fourth example and this modified example, the phase difference spectrum estimation unit 122 uses the relationship between the real part u(k) and the imaginary part v(k) of Y(k), which represent the argument angle of Y(k) in the complex plane, for each frequency k, to perform a binary search for the range of the argument angle in the quadrant in which Y(k) exists, thereby identifying the range of the argument angle in which Y(k) exists, and obtains a predetermined candidate value of the phase difference spectrum as the phase difference spectrum φ(k) for the identified range of the argument angle.

[0085] For example, the first example described above corresponds to a case where there are four representative values ​​for a predetermined candidate phase difference spectrum, and each representative value of the phase difference spectrum corresponds to one of the four quadrants from the first to the fourth quadrant in the complex plane of the complex conjugate product of the frequency spectrum of the first channel and the frequency spectrum of the second channel. In the first example, for each frequency k, the phase difference spectrum estimation unit 122 uses the relationship between the value of the real part u(k) and the value of the imaginary part v(k) of Y(k), which represent the complex plane of the argument of Y(k), as a combination of whether u(k) is positive or negative and whether v(k) is positive or negative, to obtain the representative value of the corresponding quadrant from among the four predetermined representative values ​​of the phase difference spectrum as the phase difference spectrum φ(k).

[0086] For example, the second example described above corresponds to a case where eight predetermined candidate values ​​for the phase difference spectrum are representative values, and each representative value of the phase difference spectrum is associated with one of eight ranges, which are the ranges of the argument of the product of the complex conjugates of the frequency spectrum of the first channel and the frequency spectrum of the second channel on the complex plane, where the argument is smaller or larger in each quadrant. In the second example, for each frequency k, the phase difference spectrum estimation unit 122 uses the relationship between the value of the real part u(k) and the value of the imaginary part v(k) of Y(k), which represent the argument of Y(k) on the complex plane, specifically whether u(k) is positive or negative and whether v(k) is positive or negative, and which of the absolute values ​​of u(k)| |u(k)| and v(k)| is larger, to obtain a representative value of the corresponding range from among the eight predetermined representative values ​​of the phase difference spectrum as the phase difference spectrum φ(k).

[0087] For example, the third example described above corresponds to a case where there are 4N representative values ​​for a predetermined candidate phase difference spectrum, and each representative value of the phase difference spectrum corresponds to one of 4N ranges obtained by dividing the argument of the product of the complex conjugates of the frequency spectrum of the first channel and the frequency spectrum of the second channel into N parts in the complex plane. In the third example, for each frequency k, the phase difference spectrum estimation unit 122 uses the relationship between the value of the real part u(k) and the value of the imaginary part v(k) of Y(k), which represent the argument of Y(k) in the complex plane, by considering whether u(k) is positive or negative and whether v(k) is positive or negative, and by multiplying either |u(k)| or |v(k)| by a predetermined value and then determining which is larger, to obtain a representative value for the corresponding range from among the 4N predetermined representative values ​​of the phase difference spectrum as the phase difference spectrum φ(k).

[0088] For example, the fourth example described above involves identifying the quadrant in which Y(k) exists and performing a binary search for the range of the argument within that quadrant. In the fourth example, the phase difference spectrum estimation unit 122 identifies the quadrant in which Y(k) exists for each frequency k using the combination of whether u(k) is positive or negative and whether v(k) is positive or negative. By multiplying either |u(k)| or |v(k)| by a predetermined value and then performing a binary search using which is larger, the unit obtains a representative value for the corresponding range from among several predetermined representative values ​​of phase difference spectra as the phase difference spectrum φ(k).

[0089] For example, the fifth example described above is one in which each candidate value of the phase difference spectrum is not pre-associated with the range of the argument angle on the complex plane of the product of the complex conjugates of the frequency spectrum of the first channel and the frequency spectrum of the second channel, and the selection of the most plausible candidate value is pre-associated with each candidate value of the phase difference spectrum. In the fifth example, the representative value storage unit 1221 of the phase difference spectrum estimation unit 122 stores candidate values ​​φ(q) and tanθ(φ(q)) for each integer q between 1 and Q from 1, and the phase difference spectrum estimation unit 122 obtains the phase difference spectrum φ(k) as the φ(q) corresponding to the tanθ(φ(q)) that gives the smallest value of |u(k)×tanθ(φ(q))-v(k)| for each frequency k.

[0090] The phase difference spectrum φ(k) for each frequency k from 0 to T-1 obtained by the phase difference spectrum estimation unit 122 is output from the phase difference spectrum estimation unit 122 and input to the inter-channel relationship information acquisition unit 123.

[0091] [Inter-channel relationship information acquisition unit 123] The inter-channel relationship information acquisition unit 123 acquires predetermined τ max from τ min up to (for example, τ max is a positive number, τ min The number of candidate samples τ (where τ is a negative number) cand Regarding this, the sequence obtained from the phase difference spectrum φ(0) to φ(T-1) is inversely Fourier transformed to τ max from τ min Number of candidate samples up to τ cand Regarding the phase difference signal ψ(τ cand ) is obtained, and the phase difference signal ψ(τ cand The correlation value γ is the absolute value of ). cand The maximum value of is obtained and output as the inter-channel correlation value γ, and τ when the correlation value is at its maximum value is obtained. cand If the value is positive, information indicating that the first channel is leading is obtained and output as leading channel information, and the τ when the correlation value is at its maximum is obtained. cand If the value is negative, information indicating that the second channel is preceding is obtained and output as preceding channel information (step S123). The following describes in detail an example of the processing of the inter-channel relationship information acquisition unit 123.

[0092] The inter-channel relationship information acquisition unit 123 first determines a predetermined τ max from τ min up to (for example, τ max is a positive number, τ min The number of candidate samples τ (where τ is a negative number) cand Regarding this, the sequence of phase difference spectra φ(0) to φ(T-1) input from the phase difference spectrum estimation unit 122 is inversely transformed as shown in equation (1-7) below, τ max from τ min Number of candidate samples up to τ cand Regarding the phase difference signal ψ(τ cand ) obtain.

number

[0093] The predetermined number of candidate samples is τ max from τ min It may be any integer value up to τ max from τ min It may include fractional or decimal values ​​between the given number and τ. max from τ min It is not necessary to include any integer value between the given value and τ. max =-τ min It may be so, or it may not be so. Assuming we are dealing with an input audio signal where it is unknown which channel is preceding, τ max Let τ be a positive number. min It is best to treat it as a negative number.

[0094] The phase difference signal ψ(τ) obtained from equation (1-7) cand The absolute value of ) represents a kind of correlation corresponding to the likelihood of the time difference between the first channel input sound signals x1(1), x1(2), ..., x1(T) and the second channel input sound signals x2(1), x2(2), ..., x2(T). Therefore, the inter-channel relationship information acquisition unit 123 calculates each candidate sample number τ cand Phase difference signal ψ(τ) cand The absolute value of ) is the correlation value γcand It is used as follows: That is, the inter-channel relationship information acquisition unit 123 uses the phase difference signal ψ(τ) obtained by equation (1-7). cand The correlation value γ is the absolute value of ). cand The maximum value of is obtained and output as the inter-channel correlation value γ, and τ when the correlation value is at its maximum value is obtained. cand If the value is positive, information indicating that the first channel is leading is obtained and output as leading channel information, and the τ when the correlation value is at its maximum is obtained. cand If the value is negative, information indicating that the second channel is leading is obtained and output as leading channel information. The inter-channel relationship information acquisition unit 123 obtains the value of τ when the correlation value is at its maximum. cand If the value is 0, information indicating that the first channel is leading may be obtained and output as leading channel information, or information indicating that the second channel is leading may be obtained and output as leading channel information, but it is preferable to obtain and output information indicating that neither channel is leading as leading channel information. The inter-channel relationship information acquisition unit 123 obtains the correlation value γ cand As a phase difference signal ψ(τ cand Instead of using the absolute value of ) as is, for example, each τ cand Regarding the phase difference signal ψ(τ cand τ for the absolute value of ) cand A normalized value may be used, such as the relative difference with respect to the average of the absolute values ​​of the phase difference signals obtained for each of the multiple candidate samples before and after. In other words, the inter-channel relationship information acquisition unit 123 uses each τ cand Regarding this, a predetermined positive number τ range Using this, the average value is obtained by the following equation (1-8), and the obtained average value ψ c (τ cand ) and phase difference signal ψ(τ cand The normalized correlation obtained by formula (1-9) below using ) is γ cand It may be used as such.

number

number

[0095] Furthermore, the normalized correlation value obtained by equation (1-9) is a value between 0 and 1, and τ cand The inter-channel time difference is so close to 1 that it seems plausible, and τ cand This value exhibits properties that are so close to zero that it is unlikely to represent an inter-channel time difference.

[0096] The inter-channel correlation value γ and leading channel information obtained by the inter-channel relationship information acquisition unit 123 are output from the inter-channel relationship information acquisition unit 123 and input to the downmix unit 130.

[0097] [Downmix section 130] The downmix unit 130 receives the first channel input sound signal input to the sound signal downmix device 100, the second channel input sound signal input to the sound signal downmix device 100, the inter-channel correlation value γ output by the inter-channel relationship information estimation unit 120, and the preceding channel information output by the inter-channel relationship information estimation unit 120. The downmix unit 130 weights and adds the first channel input sound signal and the second channel input sound signal so that the input sound signal of the preceding channel of the second channel input sound signal is included in the downmix signal as the inter-channel correlation value γ is larger, and outputs a downmix signal (step S130).

[0098] For example, if the absolute value or normalized value of the correlation coefficient is used for the inter-channel correlation value as in the example described above in the section explaining the inter-channel relationship information estimation unit 120, then the inter-channel correlation value γ input from the inter-channel relationship information estimation unit 120 is a value between 0 and 1. Therefore, the downmix unit 130 weights and adds the first channel input sound signal x1(t) and the second channel input sound signal x2(t) for each corresponding sample number t using the weight determined by the inter-channel correlation value γ to obtain the downmix signal x M(t) is sufficient. For example, the downmix unit 130 will use x if the preceding channel information is information indicating that the first channel is preceding, that is, if the first channel is preceding. M (t) = ((1+γ) / 2) × x1(t) + ((1-γ) / 2) × x2(t), if the preceding channel information indicates that the second channel precedes it, that is, if the second channel precedes it, then x M Let (t) = ((1-γ) / 2) × x1(t) + ((1+γ) / 2) × x2(t), and the downmix signal x M The goal is to obtain (t). When the downmix unit 130 obtains a downmix signal in this way, the smaller the channel correlation value γ is, that is, the smaller the correlation between the first channel input sound signal and the second channel input sound signal, the closer the downmix signal is to the signal obtained by averaging the first channel input sound signal and the second channel input sound signal. The larger the interchannel correlation value γ is, that is, the larger the correlation between the first channel input sound signal and the second channel input sound signal, the closer the downmix signal is to the input sound signal of the preceding channel among the first channel input sound signal and the second channel input sound signal.

[0099] Furthermore, if neither channel is preceding, the downmix unit 130 should weight-add the first channel input sound signal and the second channel input sound signal so that they are included in the downmix signal with equal weight, thereby obtaining and outputting the downmix signal. In other words, if the preceding channel information indicates that neither channel is preceding, the downmix unit 130 should, for example, weight-add the first channel input sound signal and the second channel input sound signal to obtain the downmix signal, specifically, for each sample number t, the first channel input sound signal x1(t) and the second channel input sound signal x2(t) are averaged to x M (t) = (x1(t) + x2(t)) / 2 is downmixed to the signal x M (t) is a good choice.

[0100] <Second Embodiment> The audio signal downmixing device 100 of the second embodiment is configured such that the inter-channel relationship information acquisition unit 123 assigns weights to each frequency to the phase difference signal ψ(τ) compared to the audio signal downmixing device 100 of the first embodiment. cand The system is modified to obtain such a result, and the accuracy of the phase difference spectrum estimation obtained by the phase difference spectrum estimation unit 122 is made dependent on the weight for each frequency. The differences between the audio signal downmixing device 100 of the second embodiment and the audio signal downmixing device 100 of the first embodiment will be explained below.

[0101] The channel relationship information acquisition unit 123 of the second embodiment acquires the number of candidate samples τ cand Regarding this, the sequence of phase difference spectra from the estimated value φ(0) input from the phase difference spectrum estimation unit 122 to φ(T-1) is inversely transformed as shown in equation (2-1) below, τ max from τ min Number of candidate samples up to τ cand Regarding the phase difference signal ψ(τ cand ) obtain.

number

[0102] In equation (2-1), w(k) is a weighting coefficient with respect to frequency k, and is a positive value. For example, w(k) may be a value greater than 0 and less than or equal to 1, with smaller values ​​as k approaches 0 or T-1, and larger values ​​as k is farther from 0 and T-1.

[0103] Equation (2-1) shows the phase difference signal ψ(τ cand When obtaining the phase difference spectrum, the phase difference signal ψ(τ) of the phase difference spectrum estimation accuracy by the phase difference spectrum estimation unit 122 is used. candThe impact on the phase difference spectrum is smaller for frequencies k where the weight coefficient w(k) is small. In other words, for frequencies k where the weight coefficient w(k) is small, the accuracy of the phase difference spectrum estimation unit 122 can be lower than for frequencies k where the weight coefficient w(k) is large. For example, when using the phase difference spectrum estimation unit 122 of the third example, for frequencies k where the weight coefficient w(k) is small, the number of divisions N in each quadrant can be smaller than for frequencies k where the weight coefficient w(k) is large. Also, for example, when using the phase difference spectrum estimation unit 122 of the fourth example, for frequencies k where the weight coefficient w(k) is small, the number of binary searches P can be smaller than for frequencies k where the weight coefficient w(k) is large. Also, for example, when using the phase difference spectrum estimation unit 122 of the fifth example, for frequencies k where the weight coefficient w(k) is small, the number of candidate phase difference spectra Q can be smaller than for frequencies k where the weight coefficient w(k) is large.

[0104] Furthermore, if the phase difference spectrum estimation unit 122 of the fourth example is used, the number of binary searches (comparison steps) s(k) for each frequency k can be determined within the total number of binary searches (comparison steps) S for the entire frequency domain in such a way that the sum of the estimation errors for the deflection angle in the entire frequency domain is minimized. First, the total number of comparison steps S for the entire frequency domain and the number of comparison steps s(k) for each frequency k are expressed by the following equation (2-2).

number

[0105] The estimation error for the deflection angle of each sample is 2 -2s(k) Since it is proportional to the coefficient of error, the sum of estimation errors D for the entire frequency domain is expressed by the following equation (2-3).

number

[0106] Therefore, in order to minimize the sum of the estimation errors D of the deflection angles across the entire frequency domain while keeping the total number of comparison steps S constant across the entire frequency domain, the number of comparison steps s(k) for each frequency k should be determined using the following equation (2-4).

number

[0107] Since the weight coefficient w(k) is predetermined, when using the phase difference spectrum estimation unit 122 of the fourth example, the number of binary searches P for each frequency k should be predetermined based on the number of comparison steps s(k) for each frequency k calculated by equation (2-4). In other words, the predetermined number of binary searches P for each frequency k should be smaller for frequencies k where the weight coefficient w(k) is small.

[0108] <Third Embodiment> In the first and second embodiments, the phase difference spectrum estimation process of the present invention was described in which it was applied to an audio signal downmixing device. However, the phase difference spectrum estimation process of the present invention may also be applied to an inter-channel relationship information estimation device that estimates information representing the relationship between a first channel input audio signal and a second channel input audio signal. This embodiment will be described as the third embodiment.

[0109] <<Channel-to-Channel Relationship Information Estimation Device 120>> The inter-channel relationship information estimation device 120 of the third embodiment includes a Fourier transform unit 121, a phase difference spectrum estimation unit 122, and an inter-channel relationship information acquisition unit 123, as shown in Figure 5. In other words, the inter-channel relationship information estimation device 120 includes the phase difference spectrum estimation device 200 of the fourth embodiment, which will be described later, as the phase difference spectrum estimation unit 122. The inter-channel relationship information estimation device 120 of the third embodiment obtains and outputs inter-channel relationship information, which is information representing the relationship between two input sound signals from an input 2-channel stereo sound signal, in frame units of a predetermined time length, for example, 20 ms. The 2-channel stereo time-domain sound signal input to the inter-channel relationship information estimation device 120 is, for example, a digital audio signal or sound signal obtained by AD conversion after sound such as speech or music is picked up by two microphones, and consists of a first channel input sound signal and a second channel input sound signal. The inter-channel relationship information output by the inter-channel relationship information estimation device 120 is input to a device that encodes or processes sound signals. The inter-channel relationship information estimation device 120 of the third embodiment performs the processing of steps S121, S122, and S123, as illustrated in Figure 6, for each frame. The inter-channel relationship information estimation device 120 of the third embodiment will be described below with appropriate reference to the descriptions of the first and second embodiments.

[0110] [Fourier transform section 121] The Fourier transform unit 121 is the same as the Fourier transform unit 121 of the first embodiment. The Fourier transform unit 121 performs a Fourier transform on the first channel input sound signals x1(1), x1(2), ..., x1(T) and the second channel input sound signals x2(1), x2(2), ..., x2(T), respectively, to obtain the frequency spectrum X1(k) of the first channel and the frequency spectrum X2(k) of the second channel at each frequency k from 0 to T-1 (step S121).

[0111] [Phase difference spectrum estimation unit 122] The phase difference spectrum estimation unit 122 is the same as the phase difference spectrum estimation unit 122 in the first embodiment. The phase difference spectrum estimation unit 122 includes a representative value storage unit 1221 in which representative values of a plurality of phase difference spectra, which are values on the circumference of the unit circle in the complex plane and have different arguments in the complex plane, are stored in advance. The phase difference spectrum estimation unit 122 selects one of the representative values of the plurality of phase difference spectra stored in the representative value storage unit 1221 based on the relationship between the real part u(k) value and the imaginary part v(k) value of the product Y(k) of the frequency spectrum X1(k) of the first channel and the complex conjugate \(\overline{X2(k)}\) of the frequency spectrum X2(k) of the second channel, and obtains it as the phase difference spectrum φ(k) (step S122). A specific example of the phase difference spectrum estimation unit 122 is as described in the first to fifth examples of the phase difference spectrum estimation unit 122 in the first embodiment, their modified examples, and the second embodiment.

[0112] [Inter-channel relationship information acquisition unit 123] The inter-channel relationship information acquisition unit 123 is the same as the phase difference spectrum estimation unit 122 in the first embodiment. However, as long as at least any one of the inter-channel correlation value γ, the preceding channel information, and the inter-channel time difference described later is output as the inter-channel relationship information by the inter-channel relationship information acquisition unit 123. That is, the inter-channel relationship information acquisition unit 123 first determines τ max to τ min For each candidate sample number τ cand , the series from the phase difference spectrum φ(0) to φ(T - 1) is subjected to inverse Fourier transform to obtain the phase difference signal ψ(τ max to τ min For each candidate sample number τ cand , and the phase difference signal ψ(τ cand ) is obtained, and the maximum value of the correlation value γ cand , which is the absolute value of the phase difference signal ψ(τ cand ), is obtained. Next, when the inter-channel relationship information acquisition unit 123 outputs the inter-channel correlation value γ, the correlation value γ cand , which is the absolute value of the phase difference signal ψ(τ cand

[0112] [チャネル間関係情報取得部123] チャネル間関係情報取得部123は、第1実施形態の位相差スペクトル推定部122と同様である。ただし、チャネル間関係情報取得部123がチャネル間関係情報として出力するのは、チャネル間相関値γ、先行チャネル情報、後述するチャネル間時間差、の少なくとも何れかであればよい。すなわち、チャネル間関係情報取得部123は、まず予め定めたτ max からτ min までの各候補サンプル数τ cand について、位相差スペクトルφ(0)からφ(T-1)による系列を逆フーリエ変換してτ max からτ <000,0087> までの各候補サンプル数τ cand について位相差信号ψ(τ cand )を得て、位相差信号ψ(τ cand )の絶対値である相関値γ cand の最大値を得る。次に、チャネル間関係情報取得部123は、チャネル間相関値γを出力する場合には、位相差信号ψ(τ cand )の絶対値である相関値γ cand <000,0564> <000,0565> <000,0566> [Inter-channel relationship information acquisition unit 123] <000,0567> The inter-channel relationship information acquisition unit 123 is the same as the phase difference spectrum estimation unit 122 in the first embodiment. However, as long as at least any one of the inter-channel correlation value γ, the preceding channel information, and the inter-channel time difference described later is output as the inter-channel relationship information by the inter-channel relationship information acquisition unit 123. That is, the inter-channel relationship information acquisition unit 123 first determines τ <000,0083> to τ <000,0084> For each candidate sample number τ <000,0085> , the series from the phase difference spectrum φ(0) to φ(T - 1) is subjected to inverse Fourier transform to obtain the phase difference signal ψ(τ <^000,0086> to τ <000,0087> For each candidate sample number τ <000,0088> , and the phase difference signal ψ(τ <000,0089> ) is obtained, and the maximum value of the correlation value γ <000,0090> , which is the absolute value of the phase difference signal ψ(τ <000,0091> ), is obtained. Next, when the inter-channel relationship information acquisition unit 123 outputs the inter-channel correlation value γ, the correlation value γ <000,0092> , which is the absolute value of the phase difference signal ψ(τ <000,0093>The maximum value is obtained and output as the inter-channel correlation value γ. Furthermore, when the inter-channel relationship information acquisition unit 123 outputs the inter-channel time difference, it outputs the value of τ when the correlation value is at its maximum. cand This is obtained and output as the inter-channel time difference. Furthermore, when the inter-channel relationship information acquisition unit 123 outputs preceding channel information, it outputs the τ at the maximum value of the correlation value. cand If the value is positive, information indicating that the first channel is leading is obtained as leading channel information, and the τ at which the correlation value is maximum is obtained. cand If the value is negative, information indicating that the second channel is preceding is obtained as leading channel information. (End of step S123)

[0113] <Fourth Embodiment> As can be seen from the first and second embodiments, which describe an application of the phase difference spectrum estimation process of the present invention to an audio signal downmixing device, and the third embodiment, which describes an application of the phase difference spectrum estimation process of the present invention to an inter-channel relationship information estimation device, in short, an independent device, the phase difference spectrum estimation device, may perform the phase difference spectrum estimation process of the present invention. This embodiment will be described as the fourth embodiment.

[0114] Phase difference spectrum estimation device 200 The phase difference spectrum estimation device 200 of the fourth embodiment includes a Fourier transform unit 121 and a phase difference spectrum estimation unit 122, as shown in Figure 7. The phase difference spectrum estimation device 200 of the fourth embodiment obtains and outputs estimated values ​​of the phase difference spectrum for each frequency in the frequency domain from the input signals of two channels, namely a first channel input signal and a second channel input signal. An example of the two channel signals input to the phase difference spectrum estimation device 200 is, for example, a time-domain sound signal of a 2-channel stereo frame unit with a predetermined time length of 20 ms. However, the two channel signals input to the phase difference spectrum estimation device 200 are not limited to sound signals; they may be image signals or any other type of signal. When the signals input to the phase difference spectrum estimation device 200 are time-domain sound signals of a 2-channel stereo system, these time-domain sound signals are, for example, digital audio signals or acoustic signals obtained by AD conversion of sound such as speech or music picked up by two microphones, and consist of a first channel input sound signal and a second channel input sound signal. The phase difference spectrum output by the phase difference spectrum estimation device 200 is input to devices that estimate inter-channel relationship information using the phase difference spectrum, devices that downmix signals, encoding devices, signal processing devices, etc.

[0115] The phase difference spectrum estimation device 200 of the fourth embodiment performs the processing shown in steps S121 and S122, illustrated in Figure 8, for each predetermined unit, for example, for each frame in the case of an audio signal. Hereinafter, the phase difference spectrum estimation device 200 of the fourth embodiment will be described with reference to the description of the first embodiment as appropriate, with the predetermined unit being T samples, the first channel input signals being x1(1), x1(2), ..., x1(T), and the second channel input signals being x2(1), x2(2), ..., x2(T).

[0116] [Fourier transform section 121] The Fourier transform unit 121 is the same as the Fourier transform unit 121 of the first embodiment. The Fourier transform unit 121 performs a Fourier transform on the first channel input signals x1(1), x1(2), ..., x1(T) and the second channel input signals x2(1), x2(2), ..., x2(T), respectively, to obtain the frequency spectrum X1(k) of the first channel and the frequency spectrum X2(k) of the second channel at each frequency k from 0 to T-1 (step S121).

[0117] [Phase difference spectrum estimation unit 122] The phase difference spectrum estimation unit 122 is the same as the phase difference spectrum estimation unit 122 of the first embodiment. The phase difference spectrum estimation unit 122 includes a representative value storage unit 1221 that stores in advance representative values ​​of a plurality of phase difference spectra, which are values ​​that lie on the circumference of the unit circle in the complex plane and have different argument angles in the complex plane. The phase difference spectrum estimation unit 122 selects one of the plurality of representative values ​​of phase difference spectra stored in the representative value storage unit 1221 based on the relationship between the real part u(k) and the imaginary part v(k) of the product Y(k) of the complex conjugate ￣X2(k) of the frequency spectrum X1(k) of the first channel and the frequency spectrum X2(k) of the second channel (step S122). Specific examples of the phase difference spectrum estimation unit 122 are as described in the first to fifth examples and their variations of the phase difference spectrum estimation unit 122 of the first embodiment, for example, as follows.

[0118] For example, the phase difference spectrum estimation unit 122 determines which quadrant Y(k) is in, using P as a predetermined integer greater than or equal to 0. If P=0, it obtains the representative value of the phase difference spectrum for the quadrant in which Y(k) is located from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit as the phase difference spectrum φ(k). If P≠0, it performs a binary search for the range of the argument angle for the quadrant in which Y(k) is located P times to identify the range of the argument angle in which Y(k) is located, and obtains the representative value of the phase difference spectrum for the identified range of the argument angle from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit as the phase difference spectrum φ(k).

[0119] More specifically, the phase difference spectrum estimation unit 122 obtains the phase difference spectrum φ(k) by the following first to sixth substeps, with P being a predetermined integer greater than or equal to 0. First substep: The phase difference spectrum estimation unit 122 sets p=0 and determines which quadrant of the complex plane Y(k) is in based on the sign of u(k) or whether u(k) is positive or negative, and the sign of v(k) or whether v(k) is positive or negative, and obtains a representative value of the argument within the range of the argument in the quadrant in which Y(k) exists. Second substep: Following the first substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in the first substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Third substep: Following the first substep, if p=P is not true, the phase difference spectrum estimation unit 122 sets 1 as the new p, obtains the range of the argument angles in the quadrant where Y(k) exists as the search range for the next substep (the fourth substep to be performed), and obtains the absolute value of the tangent of the representative value of the argument angles in that search range. Fourth substep: The phase difference spectrum estimation unit 122 determines that if the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in the previous substep (third substep or sixth substep) by |u(k)| is greater than |v(k)|, then Y(k) exists in the real axis range of the search range obtained in the previous substep, and obtains a representative value of the argument angle of the real axis range of the search range obtained in the previous substep. If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument angle of the search range obtained in the previous substep by |u(k)| is less than |v(k)|, then Y(k) exists in the imaginary axis range of the search range obtained in the previous substep, and obtains a representative value of the argument angle of the imaginary axis range of the search range obtained in the previous substep. Fifth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in the fourth substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Sixth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, if p=P is not true, adds 1 to p to obtain a new value of p, and uses the range of the argument angle within the range where Y(k) determined in the fourth substep exists as the search range for the next fourth substep. It also obtains the absolute value of the tangent of the representative value of the argument angle obtained in the fourth substep as the absolute value of the tangent of the representative value of the argument angle within the search range for the next fourth substep.

[0120] Alternatively, the phase difference spectrum estimation unit 122 obtains the phase difference spectrum φ(k) by the following first to sixth substeps, with P being a predetermined integer greater than or equal to 0. First substep: The phase difference spectrum estimation unit 122 sets p=0 and determines which quadrant of the complex plane Y(k) is in based on the sign of u(k) or whether u(k) is positive or negative, and the sign of v(k) or whether v(k) is positive or negative, and obtains the median of the range of argument angles in the quadrant in which Y(k) is located. Second substep: Following the first substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the median value obtained in the first substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Third substep: Following the first substep, if p=P, the phase difference spectrum estimation unit 122 sets 1 as the new p and obtains the range of the argument angles in the quadrant where Y(k) exists as the search range for the next substep (the fourth substep to be performed). Fourth substep: If performed after the third substep, the phase difference spectrum estimation unit 122 determines that if |u(k)| is greater than |v(k)|, Y(k) exists in the real axis half of the search range obtained in the third substep, and obtains a representative value of the argument in the real axis half of the search range obtained in the third substep. If |u(k)| is less than |v(k)|, Y(k) exists in the imaginary axis half of the search range obtained in the third substep, and obtains a representative value of the argument in the imaginary axis half of the search range obtained in the third substep. If performed after the sixth substep, the sixth substep If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument of the search range obtained in step 6 by |u(k)| is greater than |v(k)|, it is determined that Y(k) exists in the real axis range of the search range obtained in step 6, and the representative value of the argument of the real axis range of the search range obtained in step 6 is obtained. If the value obtained by multiplying the absolute value of the tangent of the representative value of the argument of the search range obtained in step 6 by |u(k)| is less than |v(k)|, it is determined that Y(k) exists in the imaginary axis range of the search range obtained in step 6, and the representative value of the argument of the imaginary axis range of the search range obtained in step 6 is obtained. Fifth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in the fourth substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Sixth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, if p=P is not true, adds 1 to p to obtain a new value of p, and uses the range of the argument angle within the range where Y(k) determined in the fourth substep exists as the search range for the next fourth substep. It also obtains the absolute value of the tangent of the representative value of the argument angle obtained in the fourth substep as the absolute value of the tangent of the representative value of the argument angle within the search range for the next fourth substep.

[0121] Alternatively, the phase difference spectrum estimation unit 122 obtains the phase difference spectrum φ(k) by the following first to sixth substeps, with P being a predetermined integer greater than or equal to 0. First substep: The phase difference spectrum estimation unit 122 sets p=0 and determines which quadrant of the complex plane Y(k) is in based on the sign of u(k) or whether u(k) is positive or negative, and the sign of v(k) or whether v(k) is positive or negative, and obtains a representative value of the argument within the range of the argument in the quadrant in which Y(k) exists. Second substep: Following the first substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in the first substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Third substep: Following the first substep, if p=P, the phase difference spectrum estimation unit 122 sets 1 as the new p, obtains the range of the argument angles in the quadrant where Y(k) exists as the search range for the next substep (the fourth substep to be performed), and obtains the absolute value of the cotangent of the representative value of the argument angle within that search range. Fourth substep: The phase difference spectrum estimation unit 122 determines that if |u(k)| is greater than the product of the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the previous substep (third substep or sixth substep) and |v(k)|, Y(k) exists in the real axis range of the search range obtained in the previous substep, and obtains the representative value of the argument angle of the real axis range of the search range obtained in the previous substep. If |u(k)| is less than the product of the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the previous substep and |v(k)|, it determines that Y(k) exists in the imaginary axis range of the search range obtained in the previous substep, and obtains the representative value of the argument angle of the imaginary axis range of the search range obtained in the previous substep. Fifth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in the fourth substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Sixth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, if p=P is not true, adds 1 to p to obtain a new value as p, and obtains the range of the argument angle in the range where Y(k) determined in the fourth substep exists as the search range for the next fourth substep. It also obtains the absolute value of the cotangent of the representative value of the argument angle obtained in the fourth substep as the absolute value of the cotangent of the representative value of the argument angle in the search range for the next fourth substep.

[0122] Alternatively, the phase difference spectrum estimation unit 122 obtains the phase difference spectrum φ(k) by the following first to sixth substeps, with P being a predetermined integer greater than or equal to 0. First substep: The phase difference spectrum estimation unit 122 sets p=0 and determines which quadrant of the complex plane Y(k) is in based on the sign of u(k) or whether u(k) is positive or negative, and the sign of v(k) or whether v(k) is positive or negative, and obtains the median of the range of argument angles in the quadrant in which Y(k) is located. Second substep: Following the first substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the median value obtained in the first substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Third substep: Following the first substep, if p=P, the phase difference spectrum estimation unit 122 sets 1 as the new p and obtains the range of the argument angles in the quadrant where Y(k) exists as the search range for the next substep (the fourth substep to be performed). Fourth substep: If performed after the third substep, the phase difference spectrum estimation unit 122 determines that if |u(k)| is greater than |v(k)|, Y(k) exists in the real axis half of the search range obtained in the third substep, and obtains a representative value of the argument in the real axis half of the search range obtained in the third substep. If |u(k)| is less than |v(k)|, Y(k) exists in the imaginary axis half of the search range obtained in the third substep, and obtains a representative value of the argument in the imaginary axis half of the search range obtained in the third substep. If performed after the sixth substep, |u(k)| If |u(k)| is greater than the product of the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the 6th substep and |v(k)|, it is determined that Y(k) exists in the real axis range of the search range obtained in the 6th substep, and the representative value of the argument angle of the real axis range of the search range obtained in the 6th substep is obtained. If |u(k)| is less than the product of the absolute value of the cotangent of the representative value of the argument angle of the search range obtained in the 6th substep and |v(k)|, it is determined that Y(k) exists in the imaginary axis range of the search range obtained in the 6th substep, and the representative value of the argument angle of the imaginary axis range of the search range obtained in the 6th substep is obtained. Fifth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, when p=P, obtains the complex value of a point on the circumference of the unit circle whose argument in the complex plane is the representative value of the argument obtained in the fourth substep, from among the representative values ​​of the phase difference spectrum stored in the representative value storage unit, as the phase difference spectrum φ(k). Sixth substep: Following the fourth substep, the phase difference spectrum estimation unit 122, if p=P is not true, adds 1 to p to obtain a new value as p, and obtains the range of the argument angle in the range where Y(k) determined in the fourth substep exists as the search range for the next fourth substep. It also obtains the absolute value of the cotangent of the representative value of the argument angle obtained in the fourth substep as the absolute value of the cotangent of the representative value of the argument angle in the search range for the next fourth substep.

[0123] For example, the phase difference spectrum estimation unit 122 sets N as an integer of 2 or more, sets n as each integer from 1 to N, sets θ as the argument of Y(k), and when (n - 1)π / 2N < θ < nπ / 2N, the complex number value of the point on the circumference of the unit circle whose argument on the complex plane is (2n - 1)π / 4N among the representative values of the phase difference spectrum stored in the representative value storage unit is obtained as the phase difference spectrum φ(k).

[0124] For example, the phase difference spectrum estimation unit 122 sets Q as an integer of 2 or more, sets q as each integer from 1 to Q, sets each representative value stored in the representative value storage unit as φ(q), sets the argument on the complex plane of φ(q) as θ(φ(q)), and obtains the representative value φ(q) corresponding to tanθ(φ(q)) for which |u(k)×tanθ(φ(q)) - v(k)| is the smallest value as the phase difference spectrum φ(k).

[0125] Note that signals of two channels in the frequency domain may be input to the phase difference spectrum estimation device 200. In this case, since it is not necessary for the phase difference spectrum estimation device 200 to perform step S121, the phase difference spectrum estimation device 200 may not include the Fourier transform unit 121. That is, the phase difference spectrum estimation device 200 includes only the phase difference spectrum estimation unit 122, sets the signal in the frequency domain of the first channel input to the phase difference spectrum estimation device 200 as X1(0), X1(2),..., x1(T - 1), sets the signal in the frequency domain of the second channel input to the phase difference spectrum estimation device 200 as X2(0), X2(2),..., x2(T - 1), and may obtain the phase difference spectrum φ(k) for each frequency k by performing the above-described step S122.

[0126] <Fifth Embodiment> A coding device that codes a signal using the phase difference spectrum obtained by the phase difference spectrum estimation device 200 of the fourth embodiment may be configured, and this embodiment will be described as the fifth embodiment.

[0127] ≪Signal Coding Device 300≫ The signal coding device 300 of the fifth embodiment includes at least a phase difference spectrum estimation unit 122 and an encoding unit 340, as shown in Figure 9. In other words, the signal coding device 300 includes the phase difference spectrum estimation device 200 of the fourth embodiment as the phase difference spectrum estimation unit 122. The signal coding device 300 obtains and outputs a signal code, which is a code representing the input signal, from the first channel input signal and the second channel input signal, which are the signals of the two input channels. The signal input to the signal coding device 300 is the same as the signal input to the phase difference spectrum estimation device 200 of the fourth embodiment. The signal code output by the signal coding device 300 is input to the signal decoding device. If the signal input to the signal coding device 300 is a frequency domain signal, the signal coding device 300 performs the processing of steps S122 and S340 illustrated in Figure 10 for each predetermined unit. When the signal input to the signal coding device 300 is a time-domain signal, the signal coding device 300 also includes a Fourier transform unit 121, as shown by the dashed line in Figure 9, and performs step S121, as shown by the dashed line in Figure 10. Step S121 performed by the Fourier transform unit 121 and step S122 performed by the phase difference spectrum estimation unit 122 are the same as in the fourth embodiment.

[0128] [Encoding unit 340] The encoding unit 340 encodes the first channel input signal and the second channel input signal input to the encoding device 300 using the phase difference spectrum obtained by the phase difference spectrum estimation unit 122 to obtain a signal code and output it (step S340). The encoding process performed by the encoding unit 340 can be any encoding process that uses the phase difference spectrum obtained by the phase difference spectrum estimation unit 122.

[0129] <Sixth Embodiment> A signal processing device may be configured to process a signal using the phase difference spectrum obtained by the phase difference spectrum estimation device 200 of the fourth embodiment, and this configuration will be described as the sixth embodiment.

[0130] ≪Signal Processing Device 400≫ The signal processing device 400 of the sixth embodiment includes at least a phase difference spectrum estimation unit 122 and a signal processing unit 450, as shown in Figure 11. In other words, the signal processing device 400 includes the phase difference spectrum estimation device 200 of the fourth embodiment as the phase difference spectrum estimation unit 122. The signal processing device 400 processes the first channel input signal and the second channel input signal, which are the signals of two input channels, to obtain and output the signal processing result. The signal input to the signal processing device 400 is the same as the signal input to the phase difference spectrum estimation device 200 of the fourth embodiment. If the signal input to the signal processing device 400 is a frequency domain signal, the signal processing device 400 performs the processing of steps S122 and S450 illustrated in Figure 12 for each predetermined unit. If the signal input to the signal processing device 400 is a time domain signal, the signal processing device 400 also includes a Fourier transform unit 121 as shown by the dashed line in Figure 11, and also performs step S121 as shown by the dashed line in Figure 12. Step S121 performed by the Fourier transform unit 121 and step S122 performed by the phase difference spectrum estimation unit 122 are the same as in the fourth embodiment.

[0131] [Signal Processing Unit 450] The signal processing unit 450 processes the first channel input signal and the second channel input signal input to the signal processing device 400 using the phase difference spectrum obtained by the phase difference spectrum estimation unit 122, obtains the signal processing result and outputs it (step S450). The signal processing performed by the signal processing unit 450 can be any signal processing that uses the phase difference spectrum obtained by the phase difference spectrum estimation unit 122.

[0132] <Note> The processing of each part of the above-mentioned devices may be implemented by a computer. In this case, the processing content of the functions that each device should have is described by a program. This program is then loaded into the memory unit 1020 of the computer 1000 shown in Figure 13, and the arithmetic processing unit 1010, input unit 1030, output unit 1040, etc. are made to operate, thereby realizing the various processing functions of each of the above-mentioned devices on the computer.

[0133] The device of the present invention, for example, as a single hardware entity, has an input unit that can receive signals from outside the hardware entity, an output unit that can output signals to outside the hardware entity, a communication unit to which a communication device (e.g., a communication cable) can be connected that can communicate with outside the hardware entity, a CPU (Central Processing Unit, which may include cache memory and registers), RAM or ROM as memory, an external storage device such as a hard disk, and a bus that connects these input unit, output unit, communication unit, CPU, RAM, ROM, and external storage device to enable data exchange between them. Furthermore, if necessary, the hardware entity may be provided with a device (drive) that can read and write recording media such as CD-ROMs. An example of a physical entity equipped with such hardware resources is a general-purpose computer.

[0134] The external storage device of the hardware entity stores the programs necessary to realize the above-mentioned functions, as well as the data required for processing these programs (this is not limited to external storage; for example, programs may be stored in ROM, a read-only storage device). Furthermore, data obtained through the processing of these programs is appropriately stored in RAM or other external storage devices.

[0135] In hardware entities, each program stored in an external storage device (or ROM, etc.) and the data necessary for processing each program are loaded into memory as needed, and interpreted, executed, and processed by the CPU as appropriate. As a result, the CPU realizes predetermined functions (each component represented as ...part, ...means, etc. above). In other words, each component in the embodiment of the present invention may be composed of a processing circuit.

[0136] As described above, when the processing functions of the hardware entity (device of the present invention) described in the above embodiment are implemented by a computer, the processing content of the functions that the hardware entity should have is described by a program. Then, by executing this program on the computer, the processing functions of the hardware entity are implemented on the computer.

[0137] The program describing this process can be recorded on a computer-readable recording medium. Computer-readable recording media are, for example, non-temporary recording media, specifically magnetic recording devices, optical discs, etc.

[0138] Furthermore, this program may be distributed, for example, by selling, transferring, or lending portable recording media such as DVDs or CD-ROMs on which the program is recorded. Alternatively, the program may be stored in the storage device of a server computer and distributed by transferring the program from the server computer to other computers via a network.

[0139] A computer executing such a program first stores the program recorded on a portable recording medium or transferred from a server computer in its own non-temporary storage device, the auxiliary recording unit 1050. Then, when processing is to be executed, the computer reads the program stored in the auxiliary recording unit 1050 into the storage unit 1020 and executes the processing according to the loaded program. Alternatively, the computer may directly read the program from the portable recording medium into the storage unit 1020 and execute the processing according to that program. Furthermore, each time a program is transferred to this computer from a server computer, it may sequentially execute the processing according to the received program. Alternatively, the above processing may be executed by a so-called ASP (Application Service Provider) type service, where the server computer does not transfer programs to this computer, but the processing function is realized only by execution instructions and result acquisition. In this embodiment, the program includes information used for processing by an electronic computer that is equivalent to a program (data that is not a direct instruction to the computer but has the property of defining the processing of the computer).

[0140] Furthermore, in this configuration, the device is configured by executing a predetermined program on a computer, but at least a part of these processes may be implemented in hardware.

[0141] The present invention is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention.

Claims

1. For frequency k, the frequency spectrum X of the input signal of the first channel. 1 (k) and the frequency spectrum X of the input signal of the second channel 2 A phase difference spectrum estimation step to estimate the phase difference spectrum φ(k) of (k), A channel relationship information acquisition step that uses the estimated frequency k phase difference spectrum φ(k) to obtain a value representing the correlation between the input sound signal of the first channel and the input sound signal of the second channel, and leading channel information which is information indicating which channel, the first channel or the second channel, is leading, A downmix step that uses the correlation value and the preceding channel information to obtain a downmix signal from the input sound signal of the first channel and the input sound signal of the second channel, A method for downmixing audio signals, including, u(k) and v(k) are the frequency spectra X, respectively. 1 (k) and frequency spectrum X 2 The complex conjugate of (k) -X 2 Let Y(k) be the real and imaginary parts of the product of (k), The phase difference spectrum estimation step is, One of several representative phase difference spectra, which are values ​​that lie on the circumference of the unit circle in the complex plane and whose arguments in the complex plane are distinct from each other, is selected as the phase difference spectrum φ(k) based on the combination of signs indicating whether u(k) is positive or negative and the signs indicating whether v(k) is positive or negative. Audio signal downmixing method.

2. A method for downmixing an audio signal according to claim 1, The phase difference spectrum estimation step is, If the sign indicating whether u(k) is positive or negative and the sign indicating whether v(k) is positive or negative both indicate a positive value, then the representative value of the phase difference spectrum in the first quadrant is obtained as the phase difference spectrum φ(k). If the sign indicating whether u(k) is positive or negative is a negative sign, and the sign indicating whether v(k) is positive or negative is a positive sign, then the representative value of the phase difference spectrum in the second quadrant is obtained as the phase difference spectrum φ(k). If the sign indicating whether u(k) is positive or negative and the sign indicating whether v(k) is positive or negative both indicate negative values, then the representative value of the phase difference spectrum in the third quadrant is obtained as the phase difference spectrum φ(k). If the sign indicating whether u(k) is positive or negative is a sign indicating a positive value, and the sign indicating whether v(k) is positive or negative is a sign indicating a negative value, then the representative value of the phase difference spectrum in the fourth quadrant is obtained as the phase difference spectrum φ(k). Audio signal downmixing method.

3. For frequency k, the frequency spectrum X of the input signal of the first channel. 1 (k) and the frequency spectrum X of the input signal of the second channel 2 A phase difference spectrum estimation unit that estimates the phase difference spectrum φ(k) of (k), A channel relationship information acquisition unit obtains, using the phase difference spectrum φ(k) for the estimated frequency k, a value representing the correlation between the input sound signal of the first channel and the input sound signal of the second channel, and leading channel information which indicates which channel, the first channel or the second channel, is preceding. A downmixing unit that uses the correlation value and the preceding channel information to obtain a downmix signal from the input sound signal of the first channel and the input sound signal of the second channel, An audio signal downmixing device including, Let \(u(k)\) and \(v(k)\) be the real part and the imaginary part of the product \(Y(k)\) of the frequency spectrum \(X\) 1 1 (k) and the complex conjugate \(\overline{X}\) 2 2 (k) of the frequency spectrum \(X\) 2 2 (k), respectively. The phase difference spectrum estimation unit is, One of several representative phase difference spectra, which are values ​​that lie on the circumference of the unit circle in the complex plane and whose arguments in the complex plane are distinct from each other, is selected as the phase difference spectrum φ(k) based on the combination of signs indicating whether u(k) is positive or negative and the signs indicating whether v(k) is positive or negative. Audio signal downmixing device.

4. A program for causing a computer to execute the audio signal downmixing method described in claim 1.

Images