JPH1051881A - Distortion removing device for speaker - Google Patents

Abstract

PROBLEM TO BE SOLVED: To decrease an arithmetic quantity and make the distortion removing device small-sized by dividing an A/D-converted audio signal to specific length and converting it into signals of frequency areas through fast Fourier transform and performing convolutional operation in the frequency areas by a multiplier and a multiplier adder. SOLUTION: The distortion removing device 10 converts the audio signal x(t) into a signal Xi(m) of a frequency area by an A/D converter 11, a frame dividing means 12, and a fast Fourier transform means 13, and processes the signal Xi(m) by a multiplier 14, a 1st storage means 15, a multiplier adder 16, and a 2nd storage means 17 to obtain a signal Xi(m). At this time, coefficients G1 and G2 stored in the 1st storage means 15 and 2nd storage means 17 are determined so as to remove the higher harmonic distortion and cross-modulation distortion of the speaker 22. Consequently, the arithmetic quantity is greatly reduced and the scale of the distortion removing device is reducible.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a distortion removing apparatus for removing harmonic distortion and cross-modulation distortion generated in a speaker for reproducing an audio signal and realizing high-fidelity reproduction of an input signal. is there.

[0002]

2. Description of the Related Art In a conventional loudspeaker, when an audio signal input from a CD player or the like to an loudspeaker through an amplifier is reproduced as audio, the audio signal faithfully reproduces the audio signal together with harmonic distortion generally called non-linear distortion. The sound of the intermodulation distortion was generated.

It is very uncomfortable to add these non-linear distortions to a music signal reproduced from a speaker, so reducing the distortion is a major problem in designing a speaker. In particular, second-order harmonic distortion and cross-modulation distortion (hereinafter collectively referred to as second-order distortion) are uncomfortable. Therefore, an object of the present invention is to reduce second harmonic distortion and cross modulation distortion.

Conventionally, there are two methods for removing these non-linear distortions when roughly classified. One is a technique that has been used for many years, and is a method in which the shape, dimensions, and material of a speaker are devised so that the distortion is minimized. The other is a method that has recently been proposed, in which an audio electric signal output from a sound source such as a CD player is corrected by a distortion removing device, input to a speaker, and finally output from the speaker. In this method, the emitted sound is converted into a sound with less distortion. The inventor studied a method belonging to the latter, and completed the invention of one system (not disclosed) shown in FIG. 7 prior to the present invention.

FIG. 7 is a block diagram showing a model of the distortion removing device 705 and the speaker 701 in the system of the above-mentioned invention completed by the inventor. The speaker 701 has a system 702 of a transfer function H1 of a speaker without distortion and a second-order transfer function H representing a second-order distortion of the speaker.
And the output signals of the two systems 703 and the adder 7
04 can be represented as a system for adding.

A distortion removing device 705 includes an A / D converter 709 for converting an input signal x (t), which is an audio electric signal from a CD player or the like, into a digital signal, and a transfer function G
A first filter 706 that performs a one-dimensional convolution operation with 1
And a second filter 707 for performing a two-dimensional convolution operation with a two-dimensional transfer function G2. Further, it has an adder 708 for adding the output signals of the first filter 706 and the second filter 707, and a D / A converter 709 for converting the output signal of the adder into an analog signal.

In such a general speaker distortion removing apparatus, the relationship between the input signal x (t) and the output signal y (t) is expressed by the following (Equation 7) when displayed in the frequency domain. You.

[0008]

(Equation 7)

The first term on the right side of (Equation 7) is the input signal x
(T) is the transfer function H of the first filter 706 and the speaker.
1 shows components passing through the system 702. The second of (Equation 7)
The second term in the curly braces of the term indicates a component of the input signal x (t) passing through the first filter 706 and the system 703 of the transfer function H2 representing the distortion of the speaker.

However, the second term in (Equation 7) is the sum of all combinations of m = m1 + m2 or | m1-m2 | with respect to m representing the address of the digital signal converted into the frequency domain. Note that the input signal x
The component of (t) passing through the second filter 707 and the system 703 of the transfer function H2 representing the distortion of the speaker is small compared to the other terms, and is ignored.

In order to remove the distortion component of the speaker 701, the two terms in the curly braces of the second term in (Equation 7) need only be canceled and their values become zero.

Next, the transfer function G of the first filter 706
1 and a method for determining the transfer function G2 of the second filter 707 will be described.

First, the transfer function G of the first filter 706
1 may be determined so that the component represented by the first term of (Equation 7) becomes equal to a desired output signal output from the speaker. For example, to make the output signal Y (m) equal to the audio electric signal X (m), (Equation 9) is derived from (Equation 8).

[0014]

(Equation 8)

[0015]

(Equation 9)

If it is desired that the output electric signal Y (m) be a signal affected by the transfer function H1 of the loudspeaker, the first signal is obtained from (Equation 10).
The transfer function G1 of the filter 706 may be as shown in (Equation 11).

[0017]

(Equation 10)

[0018]

[Equation 11]

Next, the transfer function G of the second filter 707
The method of determining 2 will be described. Since the two terms in the curly brackets of the second term in (Equation 7) need only be canceled, solving G2 assuming that the value in the parentheses is equal to 0 gives (Equation 12).

[0020]

(Equation 12)

To determine the transfer function G2, (Equation 12)
The transfer function H1 of the primary system 702 of the speaker 701
, The transfer function H2 of the second-order distortion of the second-order system 703 of the speaker 701, and the first filter 70 determined by the above method.
6 may be substituted for the transfer function G1.

[0022]

However, in the conventional distortion removing device 705, the input signal x (t) is C
If the audio signal is a very long time audio signal input from a D player or the like, the first filter 706 having M taps and the second filter 707 having M × M taps have ( It is necessary to perform a convolution operation as shown in Expression 13) in real time. In this real-time processing, the two-dimensional convolution operation by the second filter 707 represented by the second term has an enormous amount of operation. There was a problem of becoming.

[0023]

(Equation 13)

(Equation 13) is the input signal x in the time domain.
And the output signal w. Further, n indicates the order of the discretized signal, and corresponds to time. From (Equation 7), for one sample signal of the signal x (n) obtained by converting the analog input signal x (t) into a digital signal, the first filter 706 requires M times of multiplication and M-1 times of addition. Times. Similarly, from (Equation 13), the number of multiplications required in the second filter 707 is M × M × 2, M × M × 2
M-1 times.

For example, if M is 128 (M = 128), the number of multiplications required in the second filter 706 is 3
2,768 times, and the number of additions is 16,368. That is, a very large amount of calculation is required for one sample of the audio signal x (n), and the device becomes very large to realize this device.

The present invention has been made in view of the above-mentioned conventional problems. Even when a general audio signal having an infinite length is input to a general speaker, harmonic distortion or cross modulation distortion occurs. It is an object of the present invention to realize a distortion removing device capable of removing a signal and a calculation method thereof.

[0027]

In the distortion removing apparatus according to the present invention, an audio signal is divided into time-domain signals having a length of N by a frame dividing means.
After performing the operation shown in 4), the signal is converted again into a signal in the time domain, and a part thereof is sequentially connected by the frame rigidity means and output. As a result, the amount of calculation can be significantly reduced, and the size of the distortion removing device can be extremely reduced as compared with the related art. The detailed means will be described below.

First, an analog audio signal is converted into an A / D signal.
The signal is input to the conversion means and converted into a digital signal. next,
The digital signal output from the A / D conversion means is partially overlapped by the frame dividing means so as to have a length N.
And divide it in. The divided digital signal is
Fast Fourier transform is performed by the FT operation means to convert a signal in the time domain into a signal in the frequency domain.

The first storage means stores N first coefficients in advance. Here, the first multiplier is FFT
It is connected to the output side of the arithmetic means and the first storage means. Therefore, the first multiplier multiplies the N first coefficients by the signal of length N, which has been converted into a signal in the frequency domain by the FFT operation means. This operation is expressed as (Equation 1)
This is shown in the first item on the right side of 4).

On the other hand, the second storage means stores N ×
N two-dimensional second coefficients are stored. Here, the multiplier / adder is connected to the output side of the FFT operation means and the second storage means. Therefore, the multiplier-adder performs multiplication and addition as shown in the second term on the right side of (Equation 14) using the second coefficient and the signal converted into the frequency domain by the FFT operation means.

The output signal of the first multiplier and the output signal of the multiplier / adder are input to the adder and added.
This addition is the addition of the first and second terms on the right side of (Equation 14).

The added signal is the left-hand side of (Equation 14), which is subjected to inverse Fourier transform by an IFFT operation unit to convert a signal in the frequency domain into a signal in the time domain.

The IFFT output signal converted into the time domain signal is output by the frame synthesizing means, with a portion thereof being sequentially connected. Finally, the D / A conversion means converts the digital audio signal connected by the frame synthesis means into an analog signal and outputs it.

[0034]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A speaker distortion removing apparatus according to the present invention has an A / D converter to which an audio signal is input and an output signal of the A / D converter having a length N while partially overlapping. Frame dividing means for dividing and taking in; FFT calculating means for transforming the time domain signal divided by the frame dividing means into a frequency domain signal by performing a fast Fourier transform; Using the first storage means storing the coefficient of 1 and the output signal of the first coefficient and the FFT operation means,
A first multiplier for multiplying the first term on the right side of

[0035]

[Equation 14]

(Where W (m) is the m component of the distortion eliminator in the frequency domain, G1 (m) is the first coefficient, X
(M) is an m component, m, m1 of a signal obtained by transforming an input signal into a frequency domain by fast Fourier transform after discretization.
And m2 are integer values representing the number of points on the discretized frequency axis, G2 (m1, m2) is a second coefficient, and X (m
1) is an m1 component of a signal obtained by transforming an input signal into a frequency domain by Fourier transform, and X (m2) is an m2 component of a signal obtained by converting the input signal into a frequency domain by fast Fourier transform after discretization. ) In a frequency domain, using a second storage unit storing two-dimensional N × N second coefficients, and using the second coefficient and an output signal of the FFT operation unit, A multiplier / adder that performs multiplication and addition of the second term on the right side of (Equation 14), an adder that adds an output signal of the first multiplier and an output signal of the multiplier / adder, and an output of the adder IFFT operation means for performing an inverse fast Fourier transform on the signal to convert the signal into a time domain signal;
Frame synthesizing means for sequentially connecting and outputting a part of the output signal of the IFFT calculating means, and D / A for converting the output signal of the frame synthesizing means into an analog audio signal
Conversion means.

Further, the distortion removal device for the speaker receives the A / D conversion means to which the audio signal is input and the output signal of the A / D conversion means by dividing the output signal by a length N while partially overlapping. A group of frame division means, an FFT operation means for transforming the time domain signal divided by the frame division means into a frequency domain signal by fast Fourier transform, and a linear primary impulse response of a system from which distortion is to be removed First storage means for storing a frequency-domain coefficient obtained by Fourier-transforming an impulse response of an N-tap delay device having a delay amount substantially equal to the delay amount, and a coefficient of the first storage means And a first multiplier for multiplying the read coefficient by the output signal of the fast Fourier transform means, and a second multiplier for storing N first coefficients in the frequency domain. Storage means, said first coefficient,
A second multiplier that performs multiplication of the first term on the right side of (Equation 15) using the output signal of the first multiplier;

[0038]

(Equation 15)

(Where W (m) is the m component of the output signal of the distortion eliminator in the frequency domain, G1 (m) is the first coefficient, and D1 (m) is the m of the coefficient stored in the first storage means.
The component, X (m), is the m component of the signal obtained by discretizing the input signal and then converting the input signal to the frequency domain by the fast Fourier transform,
m, m1, and m2 are integer values representing the number of points on the discretized frequency axis, G2 (m1, m2) is a second coefficient,
D1 (m1) is the coefficient m of the coefficient read from the first storage means.
One component, X (m1), is m1 of a signal obtained by transforming an input signal into a frequency domain by fast Fourier transform after discretization.
A component, D1 (m2) is an m2 component of a coefficient read from the first storage means, and X (m2) is an m2 component of a signal obtained by discretizing an input signal and converting the input signal to a frequency domain by a fast Fourier transform. Using a third storage unit that stores N × N two-dimensional second coefficients in the frequency domain, the second coefficient, and an output signal of the first multiplier, 15) a multiply-adder that performs multiplication and addition of the second term on the right side, an adder that adds an output signal of the second multiplier and an output signal of the multiply-adder, and an output signal of the adder. IFFT operation means for performing an inverse fast Fourier transform to convert the signal into a time domain signal, frame synthesis means for sequentially connecting and outputting a part of the output signal of the IFFT operation means, and analog output signal of the frame synthesis means. D / A to convert to audio signal Conversion means.

The second storage device stores, among the two-dimensional second coefficients expressed in the frequency domain, an area defined by (Equation 16) and (Equation 17) and an area defined by (Equation 18) Memorize the coefficient of the area to be

[0041]

(Equation 16)

[0042]

[Equation 17]

[0043]

(Equation 18)

The multiplier / adder uses the second coefficient of only the defined area and the output signal of the FFT operation means to obtain (Equation 19)

[0045]

[Equation 19]

The following operation is performed.

[0047]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS. << Embodiment 1 >> Embodiment 1 of the present invention will be described with reference to FIG. The present embodiment provides a distortion removing apparatus that can greatly reduce the amount of calculation as compared with the related art. The distortion removing apparatus 10 includes an A / D conversion unit 11 to which an audio signal is input, and a digital output signal of the A / D conversion unit (N−N1), where N> N1 representing a numerical value is N> N1. N1 + 1) frame dividing means 12 for dividing and taking in the length N for each (N1 + 1) pieces. FFT calculation means 1 for fast Fourier transforming the time domain signal divided by the frame dividing means into a frequency domain signal
3, first storage means 15 for storing N first coefficients in the frequency domain, and the first coefficient and the FF
And a multiplier 14 for multiplying the first term of (Equation 20) using the output signal of the T operation means 13.

[0048]

(Equation 20)

Further, in the frequency domain, two-dimensional N ×
A second storage unit 17 for storing N second coefficients, and a multiplier for multiplying and adding the second term of (Equation 20) using the second coefficients and the output signal of the FFT operation unit. Adder 16
And the output signal of the first multiplier 14 and the multiply-adder 1
And an adder 18 for adding the output signals of the first and second output signals.
An IFFT operation unit 19 for performing an inverse fast Fourier transform on the output signal of the adder; a frame synthesizing unit 20 for sequentially connecting and outputting N1 to Nth data of the output signal of the IFFT operation unit; D / A for converting an output signal of the frame synthesizing means into an analog audio signal
Conversion means 21.

In such a configuration, when an analog audio signal x (t) is input from a CD player or the like to the distortion removing device 10, the A / D converter 11 converts the audio signal x (t) into a digital signal x (t). (N). Each time the (N−N1 + 1) output signals x (n) of the A / D conversion unit are input, the frame division unit 12
The output signal x (n) is divided into the previously input signal xi (n) of length N and output. However, here, N>
N1, where N1 is the conventional distortion removal device 705 of FIG.
Corresponds to the tap number M of the first filter 706. The divided signal of length N is input to the fast Fourier transform means 13 and converted into a frequency domain signal Xi (m).
“M” is an integer value representing the number of discrete points on the frequency axis.

The signal Xi (m) is supplied to a first multiplier 14, a first storage means 15, a multiply-adder 16, and a second storage means 17.
And the adder 18 performs the operation shown in (Equation 21) to obtain a signal Wi (m).

[0052]

(Equation 21)

The operation of (Equation 21) will be described in detail.
The first multiplier 14 multiplies the N first coefficients G1 (m) stored in the first storage device 15 by the output signal Xi (m) of the fast Fourier transform means 13 for each m. Match. This is the operation of the first term in (Equation 21). On the other hand, the multiplying / adding unit 16 calculates N × N second coefficients G2 (m1, m2) stored in the second storage device 17.
And the signals Xi (m1) and Xi (m2) to perform the calculation of the second term of (Equation 21). “M1” and “m2” each represent a specific value of the integer value m. Adder 18
Is the sum of the output signal of the multiplier 14 and the output signal of the multiplication / adder 16 for each m, and the signal Wi
(M) is output.

The output signal Wi (m) of the adder 18 is input to IFFT operation means 19 for performing an inverse fast Fourier transform.
It is converted into a signal wi (n) in the time domain. The frame synthesizing unit 20 outputs the signal wi (N1 +
The signal wi (N) is cut out from 1), and w1 (N1 +
1),... W1 (N), w2 (N1 + 1),.
2 (N), w3 (N1 + 1),... Here, “i” is the number of the frame divided by the frame dividing means 12. D / A conversion means 21
Converts the output signal of the frame combining means 20 into an analog signal w
(T) and output. w (t) is the speaker 22
Is input to

The loudspeaker 22, which generates harmonic distortion and cross-modulation distortion together with the input audio signal, has a system 23 of a linear transfer function H1 of the loudspeaker and a system 24 of a non-linear second-order transfer function H2 for generating the distortion. And an adder 25. The relationship between the input signal of the speaker 22 and the output sound pressure can be obtained by displaying in the frequency domain (Equation 22).
It is represented as

[0056]

(Equation 22)

Here, the relationship between the audio signal x (t) from a CD player or the like and the output sound pressure y (t) from the speaker is expressed in the frequency domain.
By substituting (m) for (Equation 22) and deleting, (Equation 7) is obtained.

The first term of (Expression 7) is the audio signal x
(T) shows a component passing through the multiplier 14 and the system 23 of the transfer function H1 of the speaker 22. The first term in the curly braces of the second term on the right side of (Equation 7) indicates a component of the audio signal x (t) passing through the multiplier / adder 16 and the system 23 of the transfer function H1 of the speaker 22. The second term in the curly braces of the second term on the right side of (Equation 7) indicates a component of the audio signal x (t) passing through the multiplier 14 and the system 24 of the transfer function H2 representing the distortion of the speaker 22.

Where m in the second term on the right side of (Equation 7) is m
= M1 + m2 or | m1−m2 |. Note that the audio signal x
The component of (t) passing through the multiplier / adder 16 and the system 24 of the transfer function H2 representing the distortion of the speaker 22 is small compared to the other terms and is ignored.

In order to remove the distortion component of the speaker 22, the two terms in the parentheses of the second term in (Equation 7) need only be canceled and their values become zero.

Next, the coefficient G1 of the first storage means 15
(M) and the coefficient G2 (m1,
The method for determining m2) will be described.

First, the coefficient G1 of the first storage means 15
(M) may be determined so that the component represented by the first term of (Equation 7) becomes equal to a desired output sound pressure output from the speaker. For example, to make the output signal Y (m) equal to the audio signal X (m), (Equation 9) is derived from (Equation 8).

Alternatively, when the output signal Y (m) is desired to be a signal affected by the transfer characteristic H1 of the linear system 23 of the speaker 22, the first signal is obtained from (Equation 10). The transfer function G1 of the filter may be set as shown in (Equation 11).

Next, the coefficient G2 (m
1, m2) will be described. Since the two terms in the curly braces in the second term on the right side of (Equation 7) need only be offset,
Assuming that the inside of the curly brackets is equal to 0, solving for G2 (m1, m2) yields (Equation 12).

In order to determine G2 (m1, m2), the transfer function H1 of the first-order system 23 of the loudspeaker 22 is calculated by (Equation 12).
And a second-order system 24 representing harmonic distortion and intermodulation distortion.
And the coefficient G1 (m) of the first storage means 15 determined by the above method may be substituted.

As described above, the coefficient G1 (m) of the first storage means 15 and the coefficient G2 (m) of the second storage means 17
If 1, m2) is determined, the harmonic distortion and the cross modulation distortion of the speaker 22 can be removed by the distortion removing device 10.

The distortion removing device 1 according to the first embodiment of the present invention.
0 in the first multiplier 14 and the multiplier / adder 16
The number of times of multiplication and the number of additions will be described below.

The number of multiplications required in the first multiplier 14 is the number shown in (Equation 23) for each digitally converted audio signal x (n).

[0069]

(Equation 23)

The number of times of multiplication required in the multiplying / adding unit 16 is determined by the digitally converted audio signal x (n).
The number of times is as shown in (Equation 24) for each of, and the number of additions is as shown in (Equation 25).

[0071]

(Equation 24)

[0072]

(Equation 25)

The required number of multiplications and additions in a system including the first multiplier 14, the multiplication / addition unit 16 and the adder 18 are summarized as follows. The number of times is shown in (Equation 27).

[0074]

(Equation 26)

[0075]

[Equation 27]

On the other hand, in the conventional method using the convolution operation in the time domain, the number of multiplications required in the convolution operation is the number shown in (Equation 28), and the number of additions is the number shown in (Equation 29).

[0077]

[Equation 28]

[0078]

(Equation 29)

For example, the above (Equation 26), (Equation 27),
N = 256 and N1 = 128 in (Equation 28) and (Equation 29)
, The method according to the present invention allows the number of multiplications to be about 1
018 times, and the number of additions is 508 times. On the other hand, in the conventional method using the convolution operation in the time domain, the number of times of multiplication is 32768 times and the number of times of addition is 16511 times. As described above, according to the present invention, the number of times of multiplication and the number of additions are about 1/30 of the conventional method, and the effectiveness of the present invention can be understood.

As described above, in the distortion removing apparatus 10 of the present invention, the amount of calculation is greatly reduced, and the scale of the apparatus can be reduced.

Example 2 Regarding Example 2 of the present invention,
This will be described with reference to FIGS.

The configuration of the second embodiment differs from that described in the first embodiment in the following points. The second storage unit in FIG. 1 stores the N × N second coefficients G2 (m1, m2).
It is only necessary to store only the coefficients of the area 201 and the area 202 indicated by oblique lines in FIG. When the area 201 shown in FIG. 2 is represented by a mathematical expression, it is an area defined by (Equation 30) and (Equation 31). When the area 202 is expressed by a mathematical formula,
This is the area defined in 2).

[0083]

[Equation 30]

[0084]

(Equation 31)

[0085]

(Equation 32)

In the distortion removing apparatus 10 according to the first embodiment,
(Equation 2) in the multiplier-adder 16 inside the distortion removing device 10
Although the number of multiplications and additions of the operation of the second term on the right side of 1) occupies a large amount of the operation amount of the entire device 10, in the second embodiment, the amount of operation in the multiplication / addition unit 16 is reduced, and Reduction can be realized.

The features of the second coefficient G2 (m1, m2) will be described. In the second coefficient G2 (m1, m2), the coefficients of the shaded area 301 in FIG. 3 and the coefficients of the other areas 302 have a line symmetry with m1 = m2 as the axis of symmetry. Further, the coefficient of the area 401 in the hatched area and the coefficient of the area 402 not in the hatched area in FIG. 4 have a conjugate relationship with (N / 2, N / 2) being the center of point symmetry. Further, the coefficient of the area 403 in the hatched area and the coefficient of the area 404 not in the hatched area in FIG. 4 have a conjugate relationship with (N / 2, N / 2) being the center of point symmetry. In addition, the coefficient of the region 405 in the hatched portion and the coefficient of the region 406 not in the hatched portion in FIG. 4 have a conjugate relationship with (N / 2, N / 2) being the center of point symmetry. According to the sampling theorem, in the product-sum operation of the second term of (Equation 21), the result of the product-sum operation is conjugate in the hatched areas 501 and 502 and the other areas 503 and 504 in FIG. In the actual calculation, only the shaded area needs to be considered. Therefore, taking the intersection of FIGS. 3, 4 and 5, FIG.
Are the areas 201 and 202 indicated by hatched portions, and (Equation 21)
When the second term of the right side is calculated, the multiply-adder 16 may perform a product-sum operation on the hatched areas 201 and 202 in FIG.

[0088]

[Equation 33]

As a result, the multiplier / adder 16 required for one sample of the digitally converted audio signal is obtained.
Is the number of times shown in (Equation 34), and the number of additions is the number of times shown in (Equation 35).

[0090]

(Equation 34)

[0091]

(Equation 35)

For example, when N = 256 and N1 = 128 are substituted into (Equation 34) and (Equation 35), the number of multiplications becomes about 161
The number of times of addition and the number of additions are about 79, and the amount of calculation in the multiplying / adding unit 16 is greatly reduced.

Third Embodiment A third embodiment of the present invention will be described with reference to FIG.

In the configuration of the third embodiment shown in FIG.
The second embodiment is different from the first embodiment in that a second multiplier 46 is provided at the subsequent stage of the fast Fourier transform means 33 to store the coefficient D1 (m).
Is provided in the second multiplier 46.

The impulse response h1 (t) of the speaker 42
Generally has a characteristic including a group delay with respect to the input signal x (t). The transfer function H1 of the primary system 43 of the speaker 42 is obtained by Fourier-transforming the impulse response h1. However, if the first coefficient G1 (m) is determined according to (Equation 9) when the impulse response h1 has a group delay, a general audio signal that changes randomly with time is input to the distortion removing device 30. In some cases, the causality of the input / output characteristics of the entire system including the distortion removing device 30 and the speaker 42 is violated, and the distortion is not removed as desired.

In the present invention, in order to solve this problem, the second multiplier 46 and the first storage means 47 are provided as shown in FIG.

In the above configuration, first, the first storage means 47 stores a coefficient D1 obtained by Fourier-transforming the impulse response characteristic of an N-tap delay device having a delay action substantially equivalent to the group delay amount τ of the impulse response h1. (M) is stored.

The second multiplier 46 applies a coefficient D1 (m) read from the first storage means 47 to the signal X (m).
Are multiplied by m and output. As a result, the component passing through the first multiplier 34 is given a delay corresponding to the group delay amount of the transfer characteristic of the primary system 43 of the speaker 42. As for the component passing through the multiplier / adder 36, the coefficient D 1 (m 1) read from the first storage means 47 in the second multiplier 46 is converted into the signal X
(M1) are multiplied, and the signal X (m2) is multiplied by the coefficient D1 (m
2) are multiplied and output. As a result, the output is delayed with a delay corresponding to the group delay amount of the harmonic distortion and the cross modulation distortion of the speaker 42. Thereby, the first multiplier 34, the second multiplier 46, the multiply-adder 36, the adder 3
The relationship between the signal X (m) and the signal W (m) according to Eq.
6).

[0099]

[Equation 36]

With the above operation, the distortion removing device 30
Even when a general audio signal having no periodicity is input, the distortion of the speaker can be removed.

In order to obtain the same effect as in this embodiment,
The following configuration is also possible. First, the first storage means 47 and the second multiplier 46 are not provided, and the configuration is the same as that of the first or second embodiment shown in FIG. The second storage unit 35 stores the coefficient D1 (m) stored in the first storage unit 47 and the second storage unit 3 in the third embodiment.
5, data G1 (m) D1 (m) as a result of multiplying the coefficient G1 (m) stored in 5 by m.
Is stored in advance. A coefficient D1 (m1) stored in the first storage means 47 is stored in the third storage means 37.
And data G2 (m1, m2) D1 obtained by multiplying the coefficient G2 (m1, m2) stored in the third storage means 37 in the third embodiment with respect to m1 and m2, respectively.
(M1) D1 (m2) is stored in advance.

With the above configuration, the first multiplier 34
Data G1 (m) D read from the second storage means 35
The signal X (m) is multiplied by 1 (m) for each m and output. As a result, the component passing through the first multiplier 34 is given a delay corresponding to the group delay amount of the impulse response h1 of the primary system 43 of the speaker. The multiplication / addition unit 36 outputs the data G2 (m1, m2) D1 (m1) D1 read from the third storage unit 37.
(M2) with respect to the signal X (m),
Multiply and output for m2. But where
The relationship between m, m1, and m2 is m = m1 + m2 or m = | m1-m2, as in the first embodiment or the second embodiment.
|

As a result, the relationship between the signal X (m) and the signal W (m) by the first multiplier 34, the multiplier / adder 36, and the adder 38 is as shown in (Expression 36).

With the above operation, the distortion removing device 42
Even when a general audio signal having no periodicity is input, the distortion of the speaker can be removed.

[0105]

According to the present invention, an A / D-converted audio signal is divided into a predetermined length and subjected to a fast Fourier transform to be converted into a signal in a frequency domain. By performing the one-dimensional and two-dimensional convolution operations in the distortion elimination device in the frequency domain with a multiplier and a multiply-adder, the arithmetic processing can be performed with a small number of multiplications and additions. A distortion removing device can be realized.

In addition to the above effects, a multiplier is further provided at a stage preceding the multiplier and the multiplier / adder, and the impulse response of the delay unit having a delay action substantially equivalent to the group delay of the impulse response of the speaker is provided. By performing multiplication by the coefficient D1 (m) obtained by Fourier-transforming the characteristic, a desired distortion removing effect can be obtained even when the transfer characteristic of the speaker includes the amount of group delay.

Further, in the frequency domain, the number of calculations can be further reduced by performing multiplication and addition using the coefficients of the two-dimensional predetermined area and the fast Fourier transform output.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a speaker distortion removing device and a speaker according to a first embodiment of the present invention.

FIG. 2 is a diagram showing areas of coefficients to be stored in a second storage device according to a second embodiment of the present invention;

FIG. 3 is a diagram showing the symmetry of coefficients to be stored in a second storage device according to the second embodiment of the present invention.

FIG. 4 is a diagram showing conjugate properties of coefficients to be stored in a second storage device according to the second embodiment of the present invention.

FIG. 5 is a diagram showing an effective range of coefficients to be stored in a second storage device, which can be considered by the sampling theorem, according to the second embodiment of the present invention;

FIG. 6 is a block diagram illustrating a configuration of a speaker distortion removing device and a speaker according to a third embodiment of the present invention.

FIG. 7 is a block diagram showing a configuration of a conventional speaker distortion removing device and a speaker.

[Explanation of symbols]

10, 30 Distortion removing device 11, 31 A / D conversion means 12, 32 Frame division means 13, 33 Fast Fourier transformation means 14, 34 First multiplier 15, 47 First storage means 16, 36 Multiplying adder 17 , 35 second storage means 18, 38 adder 19, 39 inverse fast Fourier transform means 20, 40 frame synthesizer 21, 41 D / A conversion means 22, 42 speakers 23, 43 linear transfer function H1 system 24, 44 Nonlinear distortion second-order transfer function H2 system 25, 45 Adder 46 First delay device 37 Third storage means 201 In Example 2, area 1 202 of coefficients to be stored in second storage means In Example 2, area 2 of the coefficient to be stored in the second storage means

Claims (3)

[Claims] An A / D conversion unit to which an audio signal is input; a frame division unit which divides an output signal of the A / D conversion unit by a length N while partially overlapping the frame; FFT for transforming the time domain signal divided by the dividing means into a frequency domain signal by performing a fast Fourier transform
Calculating means; first storing means for storing N first coefficients in a frequency domain; and using the first coefficient and the output signal of the FFT calculating means, A first multiplier for multiplying the first term on the right side; (W (m) is the m component of the output signal of the distortion eliminator in the frequency domain, G1 (m) is the first coefficient, and X (m) is the input signal that has been discretized into the frequency domain by fast Fourier transform. M component of the converted signal, m, m1, and m2 are integer values representing the number of points on the discretized frequency axis, G2 (m1, m2) is a second coefficient, X (m1) is an input signal, X1 (m2) represents an m1 component of a signal converted into a frequency domain by a fast Fourier transform after being discretized, and X (m2) represents an m2 component of a signal converted into a frequency domain by a fast Fourier transform after being discretized. A second storage means for storing two-dimensional (N × N) second coefficients; and using the second coefficient and the output signal of the FFT operation means, Multiply and add terms An adder, an adder for adding an output signal of the first multiplier and an output signal of the multiplier adder, and an IFFT for performing an inverse fast Fourier transform of the output signal of the adder to convert the output signal into a time domain signal. An arithmetic unit, a frame synthesizing unit for sequentially connecting and outputting a part of the output signal of the IFFT arithmetic unit, and a D / A conversion unit for converting the output signal of the frame synthesizing unit into an analog audio signal. A device for removing distortion of a speaker, comprising: 2. An A / D conversion unit to which an audio signal is input, a frame division unit for dividing and taking in an output signal of the A / D conversion unit by a length N while partially overlapping, FFT for transforming the time domain signal divided by the dividing means into a frequency domain signal by performing a fast Fourier transform
Calculating means, and a frequency-domain coefficient obtained by performing a Fourier transform on an impulse response of an N-tap delay device having a delay amount substantially equivalent to a group delay amount of a linear first-order impulse response of a system from which distortion is to be removed. A first storage means for storing the coefficient, a first multiplier for reading out the coefficient of the first storage means, and multiplying the coefficient by an output signal of the fast Fourier transform means; The second storing the first coefficient
A second multiplier that performs multiplication of the first term on the right side of (Equation 2) using the first coefficient and an output signal of the first multiplier; (W (m) is the m component of the output signal of the distortion eliminator in the frequency domain, G1 (m) is the first coefficient, D1 (m) is the m component of the coefficient stored in the first storage means, X (M) is an m component of a signal obtained by converting an input signal into a frequency domain by a fast Fourier transform after being discretized, m, m1, and m2 are integer values representing the number of points on the discretized frequency axis, G2 ( m1 and m2) are the second coefficients, and D1 (m1) is m of the coefficients read from the first storage means.
One component, X (m1) is an m1 component of a signal obtained by discretizing the input signal and then converting it to the frequency domain by the fast Fourier transform, and D1 (m2) is an m of a coefficient read from the first storage means.
X (m2) represents an m2 component of a signal obtained by discretizing an input signal and then transforming the input signal into a frequency domain by a fast Fourier transform. In the frequency domain, two-dimensional N × N second coefficients are expressed by: Third storage means for storing; a multiplier / adder for performing multiplication and addition of the second term on the right side of (Equation 2) using the second coefficient and an output signal of the first multiplier; An adder for adding the output signal of the second multiplier and the output signal of the multiply-adder; IFFT operation means for performing an inverse fast Fourier transform on the output signal of the adder to convert the output signal into a time-domain signal; A frame synthesizing unit for sequentially connecting and outputting a part of the output signal of the IFFT calculating unit; and a D / A converting unit for converting the output signal of the frame synthesizing unit into an analog audio signal. Speaker distortion Leaving device. 3. The second storage means stores, in the second coefficient of the two-dimensional digital filter represented in the frequency domain, an area defined by (Equation 3) and (Equation 4) and (Equation 5) ) Is stored, and the coefficient of the area defined by (Equation 4) (Equation 5) The multiplier / adder uses the second coefficient of only the defined area and the output signal of the FFT operation means, The speaker distortion removing apparatus according to claim 1 or 2, wherein the arithmetic operation described in (1) is performed.