JP2006217210A - Audio device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an "audio device" capable of eliminating a feeling of physical disorder by maintaining an output of a surround sound irrelevantly to contents of an input sound. <P>SOLUTION: The audio device 100 includes a correlation value calculation section 40 which calculates the degree of a correlation between an L signal and an R signal being stereo signals of two channels, an SL signal generation section 20 which generates a surround L signal by extracting a component in the R signal having a high correlation with the L signal by updating the filter coefficient of an adaptive filter by using adaptive algorithm and subtracting it from the L signal, an SR signal generation section 30 which similarly generates a surround R signal, and a μ setting section 42 which variably sets the value of a step-size parameter μ used to update the filter coefficient by the adaptive algorithm of the SL signal generation section 20 and SR signal generation section 30 according to whether the correlation value calculated by the correlation value calculation section 40 is large or small. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an audio apparatus that generates two or more sets of surround signals from two-channel stereo signals.

2. Description of the Related Art Conventionally, an audio device that generates a surround signal from a two-channel stereo signal is known (see, for example, Patent Document 1). In this audio apparatus, the surround signals SL and SR are generated by passing the input stereo signals INL and INR through an adaptive decorrelator. For example, this adaptive decorrelator is realized by adaptive signal processing using an FIR filter.
Japanese Unexamined Patent Publication No. 2003-333698 (page 3-6, FIG. 1-13)

  By the way, in the audio device disclosed in Patent Document 1, for example, the generation of the surround signal SL is performed by subtracting a highly correlated component from the stereo signal INL included in the stereo signal INR from the stereo signal INL. Is realized by an FIR filter whose filter characteristics are updated by an LMS (Least Mean Square) algorithm. The same applies to the surround signal SR. For this reason, when stereo signals INL and INR having a very high correlation like a monaural signal are input, the components of the surround signals SL and SR are almost lost, and almost no sound is output from the surround speakers. As a result, there is a problem that the listener of the audio device is uncomfortable.

  The present invention was created in view of the above points, and an object of the present invention is to provide an audio device that can maintain the output of a surround sound and prevent a sense of incongruity regardless of the content of the input sound. There is to do.

  In order to solve the above-described problem, an audio apparatus according to the present invention includes a correlation value calculation unit that calculates the degree of correlation between an L signal that is a two-channel stereo signal and an R signal, and a correlation between the L signal in the R signal. A first surround signal generating means for extracting a high-frequency component by updating a filter coefficient of an adaptive filter using an adaptive algorithm, and subtracting the component from the L signal; A second surround signal generating means for generating a second surround signal by extracting a component having a high correlation with the R signal by updating the filter coefficient of the adaptive filter using an adaptive algorithm and subtracting the component from the R signal; The adaptive algorithm by the first and second surround signal generating means according to the magnitude of the correlation value calculated by the correlation value calculating means And a μ setting means variably sets the value of the step size parameter μ for use in updating the filter coefficients in. Adjust the time until the surround signal that changes according to the degree of correlation between the L signal and the R signal disappears by setting the value of the step size parameter μ variably according to the degree of correlation between the L signal and the R signal. This makes it possible to maintain the output of the surround signal regardless of the content of the input sound and prevent the occurrence of a sense of incongruity.

  The μ setting means described above preferably sets the step size parameter μ to a small value when the correlation value is large, and conversely sets the step size parameter μ to a large value when the correlation value is small. As a result, it is possible to prevent the surround signal from being rapidly attenuated when the correlation between the L signal and the R signal is high, and to reduce the uncomfortable feeling felt by the listener of the surround sound.

  Further, the first surround signal generation means described above includes an error signal by subtracting a delay means for delaying and outputting the L signal and a signal after passing the R signal through the adaptive filter from a signal after passing through the delay means. And an LMS algorithm processing means for updating the filter coefficient of the adaptive filter using the LMS algorithm so that the power of the error signal is minimized. The second surround signal generating means Delay means for delaying and outputting, addition means for generating an error signal by subtracting the signal after passing through the adaptive filter from the L signal from the signal after passing through the delay means, and the power of the error signal to be minimized And LMS algorithm processing means for updating the filter coefficient of the adaptive filter using the LMS algorithm, and the μ setting means includes first and second μ setting means. It is desirable to set the value of the step size parameter μ for use in updating the filter coefficients by the LMS algorithm processing unit included in the surround signal generation means. This makes it possible to vary the degree of convergence of the filter coefficient update when an adaptive filter is used to extract a component having a high correlation with the L signal in the R signal or a component having a high correlation with the R signal in the L signal. The surround sound can be adjusted according to the degree of correlation between the L signal and the R signal.

  Further, the correlation value calculating means described above determines whether or not the stereo signal is a monaural signal, and the μ setting means determines the step size when the correlation value calculating means determines that the signal is a monaural signal. It is desirable to set the step size parameter μ to a large value when it is determined that the parameter μ is a small value and the signal is not a monaural signal. Accordingly, it is possible to reduce the attenuation of the surround signal when a monaural sound having the highest correlation between the L signal and the R signal is input, and it is possible to prevent a sense of incongruity from occurring. In addition, since it can be determined relatively easily whether or not the sound is monaural, the processing can be simplified.

  An audio apparatus according to an embodiment to which the present invention is applied will be described below with reference to the drawings. FIG. 1 is a diagram illustrating a configuration of an audio apparatus according to an embodiment. An audio apparatus 100 shown in FIG. 1 is mounted on a vehicle and includes adders 10 and 12, an LPF (low-pass filter) 14, an SL signal generator 20, an SR signal generator 30, a correlation value calculator 40, and a μ setting unit. 42 is provided. To this audio apparatus 100, six (5.1ch) speakers 110, 112, 120, 122, 130, 132 are connected.

  One adding unit 10 adds input stereo signals (L signal, R signal). The added signal is output from a speaker 110 serving as a center speaker installed in front of the listener. The other adding unit 12 adds the input stereo signals in the same manner as the adding unit 10. The added signal is output from a speaker 112 as a subwoofer installed behind the listener after extracting a low-frequency component through the LPF 14. In this embodiment, the stereo signals are simply added and output from the speaker 110. However, the method for generating the signal output from the speaker 110 is not limited to this, and other methods may be used.

  The SL signal generation unit 20 is a first surround signal generation unit that generates a surround L signal based on the input L signal and R signal, and from the speaker 130 installed on the left or rear left of the listener. Output. The SR signal generation unit 30 is a second surround signal generation unit that generates a surround R signal based on the input L signal and R signal, and from a speaker 132 installed on the right or right rear of the listener. Output. The generation of the surround L signal and the surround R signal described above is performed by updating the filter coefficient value of the adaptive filter using the LMS algorithm.

  The correlation value calculation unit 40 is a correlation value calculation unit, and calculates a correlation value between the input L signal and R signal. The closer the input stereo signal is to monaural sound, the greater the correlation value between the L signal and the R signal. The μ setting unit 42 is μ setting means, and updates the value of the filter coefficient of the adaptive filter by the SL signal generation unit 20 or the SR signal generation unit 30 based on the correlation value calculated by the correlation value calculation unit 40. The value of the step size parameter μ used at the time is set. Specifically, the step size parameter μ is set to a smaller value as the correlation value is larger (closer to monaural sound).

  Note that the L signal of the input stereo signal is directly output from the speaker 120 installed in the left front of the listener. In addition, the R signal in the input stereo signal is directly output from the speaker 122 installed on the right front side of the listener.

  FIG. 2 is a diagram illustrating a detailed configuration of the SL signal generation unit 20 and the SR signal generation unit 30. As shown in FIG. 2, the SL signal generation unit 20 includes an FIR filter 21, an adaptive filter (ADF) 22, an addition unit 23, and an LMS algorithm processing unit 24. The FIR filter 21 is used as a delay circuit (delay means), and outputs an input L signal with a delay corresponding to the number of taps (for example, 32 taps). The adaptive filter 22 has the same configuration as the FIR filter, and multiplies the input R signal by a predetermined tap coefficient W and outputs the result. The adding unit 23 is an adding unit, and subtracts the signal output from the adaptive filter 22 from the L signal output from the FIR filter 21 to output an error signal e. The LMS algorithm processing unit 24 is LMS algorithm processing means, and varies the filter coefficient of the adaptive filter 22 using the LMS algorithm so that the power of the error signal e output from the adding unit 23 is minimized. Further, the error signal e output from the adding unit 23 is extracted as it is as a surround L signal (SL signal) and output from the speaker 130.

  FIG. 3 is a diagram showing a detailed configuration of the adaptive filter 22. As shown in FIG. 3, the adaptive filter 22 includes a plurality of delay elements 221, a multiplying unit 222 that multiplies a signal held in each delay element 221 by a variable filter coefficient, and each of the multiplying units 222. And an adder 223 for adding outputs. The values of the filter coefficients (multipliers) of the plurality of multipliers 222 are updated by the LMS algorithm processor 24.

  The LMS algorithm processing unit 24 updates the value of the filter coefficient of the adaptive filter 22 so that the power of the error signal e output from the adding unit 23 is minimized, and the adaptive filter 22 receives the input R signal component. The value of the filter coefficient is updated so as to extract a component having a high correlation with the L signal. That is, the LMS algorithm processing unit 24 receives the R signal and the error signal e output from the adding unit 23, and the RMS algorithm is processed by the RMS algorithm by processing the R signal and the error signal e. A filter coefficient update command is output from the processing unit 24 to each multiplication unit 222 in the adaptive filter 22, and the value of the filter coefficient superimposed on the signal held in each delay element 221 is changed.

  In this way, a component having a high correlation with the L signal in the R signal is extracted by the adaptive filter 22, and this component is subtracted from the L signal by the adding unit 23. Therefore, the error signal e output from the adder 23 includes only a component that is not highly correlated with the R signal in the L signal, and this is used as the surround L signal.

  By the way, the LMS algorithm is an algorithm using the instantaneous square error as an evaluation amount, and the LMS algorithm processing unit 24 updates the value of the filter coefficient W according to the following expression.

W (n + 1) = W (n) +2 μ · e (n) · R (n) (1)
Here, μ is a step size parameter. When this value is set larger, the convergence of the filter coefficient W becomes faster. Conversely, when this value is set smaller, the convergence of the filter coefficient W becomes slower.

  As described above, since the surround L signal output from the SL signal generation unit 20 includes only components that are not highly correlated with the R signal in the L signal, the L signal and the R signal (monaural audio) having a high correlation are included. Alternatively, when the L signal and the R signal close to this are input, the surround L signal component is almost lost. In this embodiment, when the correlation value calculation unit 40 calculates the correlation value between the L signal and the R signal and a high correlation value is obtained, the step size parameter μ shown in the equation (1) by the μ setting unit 42 is obtained. The value of is set small. For example, when the correlation value is greater than or equal to a predetermined value, the value of the step size parameter μ is set to 0.0001, and otherwise, the value of the step size parameter μ is set to 0.001. Thereby, since the degree of correlation removal performed by updating the value of the filter coefficient of the adaptive filter 22 can be weakened, it can be avoided that the component of the surround L signal is almost lost.

  The SR signal generation unit 30 includes an FIR filter 31, an adaptive filter (ADF) 32, an addition unit 33, and an LMS algorithm processing unit 34. The FIR filter 31 is used as a delay circuit, and delays an input R signal by a time corresponding to the number of taps (for example, 32 taps) and outputs the delayed signal. The adaptive filter 32 has the same configuration as the FIR filter, and multiplies the input L signal by a predetermined tap coefficient W and outputs the result. The adder 33 subtracts the signal output from the adaptive filter 32 from the R signal output from the FIR filter 31, and outputs an error signal e. The LMS algorithm processing unit 34 varies the filter coefficient of the adaptive filter 32 using the LMS algorithm so that the power of the error signal e output from the adding unit 33 is minimized. Further, the error signal e output from the adding unit 33 is extracted as it is as a surround R signal (SR signal) and output from the speaker 132.

  The LMS algorithm processing unit 34 updates the value of the filter coefficient of the adaptive filter 32 so that the power of the error signal e output from the adding unit 33 is minimized. In the adaptive filter 32, the component of the input L signal The value of the filter coefficient is updated so as to extract a component having a high correlation with the R signal. In other words, the LMS algorithm processing unit 34 is supplied with the L signal and the error signal e output from the adding unit 33, and the LMS algorithm is processed by the LMS algorithm. A filter coefficient update command is output from the processing unit 34 to each multiplication unit in the adaptive filter 32, and the value of the filter coefficient superimposed on the signal held in each delay element is changed.

  Thus, the adaptive filter 32 extracts a component having a high correlation with the R signal in the L signal, and this component is subtracted from the L signal by the adding unit 33. Therefore, the error signal e output from the adder 33 includes only a component that is not highly correlated with the L signal in the R signal, and this is used as the surround R signal.

  By the way, the LMS algorithm is an algorithm using the instantaneous square error as an evaluation amount, and the LMS algorithm processing unit 34 updates the value of the filter coefficient W according to the following equation.

W (n + 1) = W (n) +2 μ · e (n) · L (n) (2)
Here, μ is a step size parameter. When this value is set larger, the convergence of the filter coefficient W becomes faster. Conversely, when this value is set smaller, the convergence of the filter coefficient W becomes slower.

  As described above, since the surround R signal output from the SR signal generation unit 30 includes only components that are not highly correlated with the L signal in the R signal, the L signal and the R signal (monaural audio) having a high correlation are included. Alternatively, when the L signal and the R signal close to this are input, the surround R signal component is almost lost. In this embodiment, the correlation value calculation unit 40 calculates the correlation value between the L signal and the R signal, and when a high correlation value is obtained, the μ setting unit 42 sets the step size parameter μ shown in the equation (2). The value of is set small. As a result, the value of the filter coefficient of the adaptive filter 32 is updated, and it takes a long time until a component having a high correlation with the R signal is extracted from the L signal, so that the component of the surround R signal is almost eliminated. You can avoid that.

  Thus, by setting the value of the step size parameter μ variably according to the degree of correlation between the L signal and the R signal, until the surround signal that changes according to the degree of correlation between the L signal and the R signal disappears. It is possible to adjust the duration of the signal, and it is possible to maintain the output of the surround signal regardless of the content of the input sound and prevent the occurrence of a sense of incongruity.

  In addition, this invention is not limited to the said embodiment, A various deformation | transformation implementation is possible within the range of the summary of this invention. In the above-described embodiment, the correlation value is calculated by the correlation value calculation unit 40 based on the input L signal and R signal. However, when the input signal format includes information on monaural sound or stereo sound. Alternatively, the correlation value calculation unit 40 may simply detect this information and determine whether the signal is a stereo signal or a monaural signal. The μ setting unit 42 sets the step size parameter μ to a small value in the case of monaural sound, and sets the step size parameter μ to a large value in the case of stereo sound. Accordingly, it is possible to reduce the attenuation of the surround signal when a monaural sound having the highest correlation between the L signal and the R signal is input, and it is possible to prevent a sense of incongruity from occurring. In addition, since it can be determined relatively easily whether or not the sound is monaural, the processing can be simplified.

  FIG. 4 is a flowchart showing a modification example regarding the variable setting of the step size parameter μ. For example, it is assumed that the configuration of the audio device 100 shown in FIG. 1 is built in or externally attached to a CD player.

  When track playback for a CD is started by the CD player (step 100), the correlation value calculation unit 40 determines whether a monaural signal is detected as a playback signal (step 101). For example, assuming that a reproduction signal is a monaural signal or a stereo signal by a decoder (not shown) that generates the reproduction signal, and a predetermined flag is set in the case of a monaural signal, a correlation value calculation unit 40 performs the determination of step 101 by checking whether or not this flag is set. When the reproduction signal is a monaural signal, an affirmative determination is made in the determination of step 101, and then the μ setting unit 42 sets μ to a small value (μ1) for the monaural signal (step 102). On the other hand, when the reproduction signal is not a monaural signal (in the case of a stereo signal), a negative determination is made in the determination of step 101, and then the correlation value calculation unit 40 has a constant correlation value between L / R of the stereo signal. It is determined whether or not a certain value (or range) has been exceeded (step 103). Among stereo signals, there are music having a high correlation between L / R and music having a low correlation. For music having a high correlation, an affirmative determination is made in the determination of step 103, and then the μ setting unit 42 sets μ to a small value (μ2) for a stereo signal having a high correlation between L / R ( Step 104). Further, for music having a low correlation, a negative determination is made in the determination in step 103, and then the μ setting unit 42 sets μ to a large value (μ3) for a stereo signal having a low correlation between L / R. (Step 105).

It is a figure which shows the structure of the audio apparatus of one Embodiment. It is a figure which shows the detailed structure of SL signal generation part and SR signal generation part. It is a figure which shows the detailed structure of an adaptive filter. 10 is a flowchart showing a modification regarding variable setting of the step size parameter μ.

Explanation of symbols

10, 12, 23, 33 Adder 14 LPF (low pass filter)
20 SL signal generator 21, 31 FIR filter 22, 32 Adaptive filter (ADF)
24, 34 LMS algorithm processing unit 30 SR signal generation unit 40 Correlation value calculation unit 42 μ setting unit 100 Audio device 110, 112, 120, 122, 130, 132 Speaker

Claims (4)

Correlation value calculating means for calculating the degree of correlation between the L signal and the R signal which are two-channel stereo signals;
A component having a high correlation with the L signal in the R signal is extracted by updating a filter coefficient of an adaptive filter using an adaptive algorithm, and is subtracted from the L signal to generate a first surround signal. 1 surround signal generating means;
A component having a high correlation with the R signal in the L signal is extracted by updating a filter coefficient of an adaptive filter using an adaptive algorithm, and subtracted from the R signal to generate a second surround signal. Two surround signal generating means;
In accordance with the magnitude of the correlation value calculated by the correlation value calculation means, the value of the step size parameter μ used when updating the filter coefficient in the adaptive algorithm by the first and second surround signal generation means Μ setting means for variably setting;
An audio apparatus comprising: In claim 1,
The μ setting means sets the step size parameter μ to a small value when the correlation value is large, and conversely sets the step size parameter μ to a large value when the correlation value is small. apparatus. In claim 1 or 2,
The first surround signal generation means includes delay means for delaying and outputting the L signal, and subtracting the signal after passing the R signal through the adaptive filter from the signal after passing through the delay means. Adding means for generating a signal; and LMS algorithm processing means for updating a filter coefficient of the adaptive filter using an LMS algorithm so that the power of the error signal is minimized.
The second surround signal generating means includes delay means for delaying and outputting the R signal, and subtracting the signal after passing the L signal through the adaptive filter from the signal after passing through the delay means. Adding means for generating a signal; and LMS algorithm processing means for updating a filter coefficient of the adaptive filter using an LMS algorithm so that the power of the error signal is minimized.
The μ setting unit sets a value of the step size parameter μ used when the filter coefficient is updated by the LMS algorithm processing unit included in the first and second surround signal generation units. Audio device to play. In any one of Claims 1-3,
The correlation value calculating means determines whether or not the stereo signal is a monaural signal,
The μ setting means sets the step size parameter μ to a small value when the correlation value calculating means determines that the signal is a monaural signal, and the step size parameter μ when the determination that the signal is not a monaural signal is made. Is set to a large value.

Images