JPH05315890A - Noise reduction device - Google Patents

Abstract

PURPOSE:To reduce effectively noise without use of an expensive high performance A/D converter. CONSTITUTION:A coefficient calculation circuit 22 discriminates an input level from a 2nd-order integration low pass filter 21, and when the level is less than a 1st setting value, filter coefficients a, b, c given to variable low pass filters 13, 18 are varied to change it cut-off frequency. Thus, a high frequency including much of noise is cut off effectively. Furthermore, when the input level is decreased more than a 2nd setting value smaller than the 1st setting value, a gain coefficient (d) given to the variable low pass filters 13, 18 is changed in response to the input level to decrease the gain itself of the filter. Thus, noises (hiss noise for example) is more effectively eliminated at an area where the input level is very low.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise reduction device for improving sound quality in audio equipment and the like.

[0002]

2. Description of the Related Art Audio equipment has a problem of noise, which must be reduced as much as possible in order to improve sound quality.

On the other hand, in recent years, digitization of audio equipment has progressed, and so-called DSP (Digital Signal Processor) has been used to digitally process audio signals. In order to perform digital signal processing by this DSP, an A / D (analog / digital) converter is required as a prerequisite, but in order to improve the quality of signal processing, a high performance A / D converter is required. Had to use. That is, since noise is particularly audible to the ear when it is a small sound, in order to deal with it, A / N having a wide dynamic range S / N characteristic is used.
A D converter is required.

[0004]

However, such a high-performance A / D converter is expensive, and there is a problem that the entire apparatus is expensive.

The present invention has been made to solve the above problems, and an object thereof is to obtain a noise reduction device capable of effectively reducing noise without using an expensive high-performance A / D converter. And

[0006]

According to a first aspect of the present invention, there is provided a noise reducing device comprising: (i) a digital filter for cutting a part of a frequency band of an input digital audio signal; and (ii) a digital audio signal. And (iii) when the input level is smaller than the first set value as a result of the determination by the level determining means, the cutoff frequency of the digital filter is changed according to the input level. And a cutoff frequency varying means.

In the noise reducing device according to the second aspect of the present invention, when the input level is smaller than the second set value set smaller than the first set value as a result of the determination by the level setting means, And a gain control means for decreasing the gain of the digital filter according to the input level.

According to a third aspect of the present invention, there is provided the noise reducing apparatus according to the first aspect, wherein the level of the digital audio signal is equal to or higher than the first set value as a result of the determination by the level determining means. , The frequency characteristic of the digital filter is made flat.

According to a fourth aspect of the present invention, in the noise reduction apparatus according to the first aspect, the level determination means obtains absolute values of the digital audio signals of both the left and right channels and then integrates them. The level determination is performed based on these added values.

According to a fifth aspect of the present invention, there is provided the noise reduction device according to the fourth aspect, wherein the level determination means determines the level based on a value obtained by logarithmically converting the added value. To do.

According to a sixth aspect of the present invention, in the noise reduction apparatus according to the fifth aspect, the cutoff frequency varying means determines a filter coefficient based on a preset function and digitally determines the filter coefficient. It is characterized by being set in a filter and changing its cutoff frequency.

According to a seventh aspect of the present invention, there is provided the noise reducing apparatus according to the sixth aspect, wherein the cutoff frequency varying means has a plurality of functions used for determining the filter coefficient, and the source type of the digital audio signal. It is characterized in that the function to be used is selected according to.

According to an eighth aspect of the noise reducing apparatus of the present invention, in the noise reducing apparatus according to the first aspect, the level determining means determines the input level only for the middle and high frequency components of the digital audio signal. It is a feature.

[0014]

In the noise reduction device according to the first aspect of the invention,
When the input level is less than the first set value, the cutoff frequency of the digital filter changes according to that value,
The frequency band in which noise is often heard in the sense of hearing is effectively cut.

In the noise reducing apparatus according to the second aspect of the present invention, when the cutoff frequency changes and the input level becomes smaller than the second set value, the digital signal is changed according to the input level. The gain itself of the filter is reduced, and noise removal is more effectively performed in the low input level region.

In the noise reducing device according to the third aspect of the present invention, when the level of the digital audio signal is equal to or higher than the first set value, the frequency characteristic of the digital filter becomes flat, so that in the high input level region. , You can get a natural hearing.

In the noise reducing apparatus according to the fourth aspect of the present invention, since the level determination is performed based on the value obtained by integrating and adding the absolute values of the digital audio signals of the left and right channels, the input levels of the left and right channels are determined. There is no offset, and the level can be determined reliably.

In the noise reducing apparatus according to the fifth aspect of the present invention, since the level determination is performed based on the value obtained by logarithmically converting the added value, it is possible to perform the level determination closer to human hearing.

In the noise reducing apparatus according to the sixth aspect of the present invention, the coefficient is determined based on a preset function, and the cutoff frequency is changed by setting the coefficient in the digital filter, so that the coefficient is determined. Simplified.

In the noise reducing apparatus according to the seventh aspect of the present invention, since the function to be used can be selected according to the source type of the digital audio signal, the optimum coefficient can be determined according to the signal source and the quality of the input signal can be determined. It is possible to obtain the cut-off frequency characteristic according to.

In the noise reducing apparatus according to the eighth aspect of the present invention, the low frequency components of the digital audio signal are excluded from the target, and the input level determination is performed only on the middle and high frequency components containing a lot of noise components in the sense of hearing. Since this is performed, it is possible to remove noise more closely to the sense of hearing.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the drawings.

FIG. 1 shows a noise reduction device according to an embodiment of the present invention. In this device, a digital audio signal input from the input terminal Ri of the right channel is branched into two via a high-pass filter (hereinafter referred to as a DC cut filter) 11 for cutting a DC component, and an absolute value circuit 12 And variable low-pass filter 1
Input to 3. The DC cut filter 11 is composed of coefficient multipliers 101 to 103, delay circuits 104 and 105, an adder 106 and the like, and operates so as to remove low frequency components including direct current which does not affect the auditory sense of about 0 to 15 Hz. To do. The output of the DC cut filter 11 is input to the absolute value circuit 12, where the absolute value of the signal is taken.

The output of the absolute value circuit 12 is input to the low-pass filter 14 for primary integration and integrated, and the adder 1
6 is input.

The low-pass filter for first-order integration 14 includes coefficient multipliers 108 to 110, a delay circuit 111, and an adder 1.
It is composed of 12, 113 and operates so as to cut the input high frequency component.

Similarly, the digital audio signal input from the input terminal Li of the left channel is input to the adder 16 via the DC cut filter 31, the absolute value circuit 32 and the low-pass filter 34 for primary integration. Also in this case, as in the case of the right channel, the DC cut filter 31
Output is branched into two, and one of them is input to the variable low-pass filter 18. The configurations of the DC cut filter 31 for the left channel and the low-pass filter 34 for primary integration are the same as those for the right channel.

The signals thus obtained by integrating and adding the absolute values of the left and right channels are input to the logarithmic conversion circuit 19 where they are converted into decibel values.

The output of the logarithmic conversion circuit 19 is input to the low-pass filter 21 for quadratic integration. The second-order integration low-pass filter 21 includes coefficient multipliers 121 to 125, delay circuits 126 to 128, and adders 129 and 130, and selectively sets the attack and recovery time constants for a fast-moving ripple signal. Filtering. That is, a small time constant is set for the rising edge (attack) of the signal, and a large time constant is set for the falling edge (recovery) of the signal. As a result, it is possible to prevent the level from changing greatly with respect to a small ripple.

The output of the low-pass filter for quadratic integration 21 is input to the coefficient calculation circuit 22. As will be described later, the coefficient calculation circuit 22 determines the input level and changes the variable low-pass filter 13 according to the input level.
And filter coefficients a to d to be set to 18 are determined.
The variable low-pass filters 13 and 18 include multipliers 131 to 131.
136, delay circuits 137 and 138, and an adder circuit 139,
The input signal is filtered to have frequency characteristics and gain according to the set filter coefficient. These outputs are output from the right channel output terminal Ro and the left channel output terminal Lo, respectively.

Next, the operation contents of the coefficient calculation circuit 22 will be described. The characteristic of the operation of the coefficient calculation circuit 22 is that the input level is the first set value L as shown in FIG.
If it is less than 1, the variable low-pass filters 13 and 18
The control is performed such that the cutoff frequency is changed to a low value and the gain itself of the filter is decreased when the cutoff frequency is less than the second set value smaller than the first set value L1. This operation will be described in detail below with reference to FIG.

First, the coefficient calculation circuit 22 checks the input level from the low-pass filter for quadratic integration 21 (step S101 in FIG. 2). If this input level is smaller than the first set value L1 (step S102; N) and larger than the second set value L2 (step S103; Y), a straight line as shown in FIG. From the formula
The filter coefficients a, b, and c are obtained (steps S104 and S105), and these are set to the variable low-pass filter 13
And 18 in the multipliers 132 to 134 (step S106).

Here, how to obtain the filter coefficients a, b and c will be described. For example, if the cutoff frequency at an input level of -30 dB is set to 10 KHz and the cutoff frequency at -60 dB is desired to be set to 1 KHz, a straight line connecting these two points is obtained by the following equation (1), Is set so that the coefficient a can be read out by setting its address using, for example, a ROM table.

A = k1 × L + k2 (1) where k1 and k2 are linear coefficients, and L is the input level (d
B) is shown.

In this case, the cutoff frequency is 10 KH.
The value of the filter coefficient a to be z is 0.4342, and the value of the filter coefficient a when the cutoff frequency is 1 KHz is 0.0615. Therefore, the straight line represented by the equation (1) is
It becomes like FIG. As can be seen from this figure, for example, when the input level is -40 dB, the value of the coefficient a is about 0.
32, and when the input level is -50 dB,
The value of the coefficient a is about 0.19.

The cutoff frequencies corresponding to these values are 6.4 KHz and 3.5 KHz, respectively.

On the other hand, the variable low-pass filters 13 and 18
In the digital filter having such a configuration, the coefficients a, b, and c have a relationship as shown in the following expressions (2) and (3).

A = b (2) c = 1-2a (3) Therefore, the filter coefficients b and c are calculated from these equations.

On the other hand, in step S103, the input level is smaller than the second set value L2 (step S10
3; N) and when this input level is larger than the third set value L3 (step S107; Y), the coefficient d for determining the gain of the filter is obtained by the linear equation shown in FIG. (Step S108), this is set in the multiplier 135 of the variable low-pass filters 13 and 18 (step S109), and the process proceeds to step S104.

Here, the gain coefficient d is obtained as follows. As shown in FIG. 6, for example, the input level is −60.
The gain coefficient for dB is set to 1, and the gain coefficient for -70 dB is set to 0.1. Then, a straight line connecting these two points is used to obtain the gain coefficient d1 according to the input level x in the same manner as in the case of the coefficients a, b and c.

As described above, as shown in FIG. 3, in the area A where the input level is L1 or more, the input level is output as it is without any change. In the area B between the input levels L2 and L1, the cutoff frequency fc changes according to the input level and filtering is performed. And the input level is L2 and L
In the region C between 3 and 4, the cutoff frequency does not change and only the gain changes.

In this way, only the frequency band containing a large amount of audible noise components is cut off, but since the cutoff frequency changes in the region B according to the input level, the audible noise is effective. It feels removed. Further, when the input level is extremely low as in C, the gain itself is narrowed down, so that, for example, so-called background noise (for example, hiss noise at the time of tape deck reproduction) is erased even at a level close to silence, which is good for hearing. The result can be obtained.

The variable low-pass filters 13 and 18
In the above, the reason why the bypass multiplier 136 is provided is as follows. That is, when the filtering is simply performed without providing the bypass multiplier, the cut in the high frequency range in FIG. 3 is too strong, which causes a slight discomfort in the sense of hearing. Therefore, by providing a bypass with a constant offset value, it is possible to limit how it falls in a wide area. However, in this case, it is necessary to provide the multiplier 131 in addition to the bypass multiplier 136 and set the coefficient value so that the sum of these multiplication coefficients becomes 1. Then, the output of the multiplier 136 and the output of the variable low-pass filter are added by the adder 140 and output through the multiplier 135.

In the present embodiment, the equation for obtaining the filter coefficient a is decided based on the straight line shown in the equation (1), but the present invention is not limited to this, and for example, a function of another form is used. May be substituted by. Furthermore, a plurality of types of functions may be prepared according to the type of signal source (eg, analog tape, CD, etc.), and these functions may be selectively used. Such a selection may be performed, for example, by a manual input 23 from the outside, but it is also possible to detect a signal indicating the type of the source and automatically select the function based on the detected signal. Further, the setting condition for obtaining the gain coefficient d in the region C may be made variable by manual input from the outside or an automatic method. In addition, the multiplier 13 before the variable low-pass filter
The level of the offset portion 24 in the wide range in FIG. 3 can be made variable by making the value of the coefficient given to 1 and the bypass multiplier 136 variable.

Further, in the present embodiment, the DC cut filters 31 and 11 from which the direct current component is removed are directly input to the absolute value circuits 32 and 12, respectively. A pass filter may be provided so that only the middle and high frequency components are subject to level determination.

Further, it is also possible to adopt the construction shown in the second embodiment below. That is, as shown in FIG. 4, the digital audio signals inputted from the input terminals Ri and Li of each channel are respectively branched into two, and one of them is cut by a DC cut filter (HPF) 35 and 15 to remove a DC component, and then a variable low-pass filter is used. It is inputted to the filters 18 and 13, and the other one is inputted to the absolute value circuits 32 and 12 after the low-pass is cut by the medium and high pass filters (MPF) 37 and 17. The other structure is similar to that of the first embodiment.

With such a configuration, the target of level determination in the coefficient calculation circuit 22 is only the middle and high frequency bands that have a great influence on the auditory sense as noise, and more appropriate filter coefficients a to d are determined. The Rukoto.

[0047]

As described above, according to the first aspect of the invention, when the input level is less than the first set value, the cutoff frequency of the digital filter is changed according to the input level. It is possible to effectively cut the frequency band only in the low level region where the presence of is noticeable. On the other hand, in a region where the signal level is large in which noise does not pose a problem, the original frequency characteristic is maintained, and therefore, there is an effect that a sense of discomfort in hearing is not generated.

According to the second aspect of the invention, when the input level becomes smaller than the second set value, the gain itself of the digital filter is lowered according to the input level. It is possible to more effectively remove noise in a region where the input level is extremely low.

Furthermore, by determining the input level with a decibel value and performing a function for obtaining the filter coefficient in a decibel display, or by limiting the input level determination target to the mid-high frequency band, It is possible to change the cutoff frequency so as to match the sensation of hearing.

Since the noise reducing apparatus as described above can be constructed without using an expensive high-performance A / D converter, there is an effect that the cost can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing a noise reduction device according to an embodiment of the present invention.

FIG. 2 is a flowchart for explaining the operation of the main part of this noise reduction device.

FIG. 3 is a characteristic diagram showing frequency characteristics of this noise reduction device.

FIG. 4 is a block diagram showing a noise reduction device according to a second exemplary embodiment of the present invention.

FIG. 5 is an explanatory diagram showing a straight line used for determining a filter coefficient for varying the cutoff frequency.

FIG. 6 is an explanatory diagram showing straight lines used to determine a gain coefficient.

[Explanation of symbols]

11, 15, 31, 35 DC cut filter (HP
F) 12,32 Absolute value circuit 13,18 Variable low-pass filter 14,34 Low-pass filter for first-order integration 17,37 Middle high-pass filter (MPF) 19 Logarithmic conversion circuit 21 Low-pass filter for second-order integration 22 Coefficient calculation circuit

Front page continuation (72) Inventor Takeshi Sato 1006 Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Industrial Co., Ltd. (72) Inventor Shigeru Kitagawa 1006 Kadoma, Kadoma City, Osaka Matsushita Electric Industrial Co., Ltd.

Claims (8)

[Claims] 1. A digital filter for cutting a part of a frequency band of an input digital audio signal, a level judging means for judging an input level of the digital audio signal, and an input level as a result of judgment by the level judging means. Is smaller than the first set value, the noise reduction device is provided with a cutoff frequency varying means for changing the cutoff frequency of the digital filter according to the input level. 2. The result of the judgment by the level setting means,
When the input level is smaller than the second set value set smaller than the first set value, the gain control means for lowering the gain of the digital filter according to the input level is provided. The noise reduction device according to claim 1. 3. The result of the judgment by the level judgment means,
2. The noise reduction device according to claim 1, wherein when the level of the digital audio signal is equal to or higher than the first set value, the frequency characteristic of the digital filter is made flat. 4. The level determining means obtains absolute values of digital audio signals of both the left and right channels, integrates the absolute values, and determines the level based on the added value. The described noise reduction device. 5. The noise reduction device according to claim 4, wherein the level determination means performs level determination based on a value obtained by logarithmically converting the added value. 6. The cut-off frequency varying means determines a coefficient based on a preset function, sets the coefficient in the digital filter, and changes the cut-off frequency. The described noise reduction device. 7. The noise reduction according to claim 6, wherein the cutoff frequency varying means has a plurality of functions used for determining the coefficient and selects the function according to a source type of the digital audio signal. apparatus. 8. The noise reduction apparatus according to claim 1, wherein the level determination means determines the input level only for the middle and high frequency components of the digital audio signal.