JPH1084239A - Digital graphic equalizer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a desired frequency characteristic by allowing even a band pass filter with a narrow pass frequency band to conduct filtering processing without degrading arithmetic precision in the fixed decimal point arithmetic operation. SOLUTION: Down-sampling sections 142 -14n reducing a sampling frequency of an input digital audio signal are provided in front of band pass filters 112 -11n for a prescribed frequency band other than a highest frequency band. The down- sampling sections 142 -14n reduce more the sampling frequency of audio data received in the band pass filters as the pass band width of the band pass filters is made narrower. Thus, a difference between a maximum value and a minimum value of filter coefficients of the band pass filters consisting of IIR filters is decreased, the filtering processing is conducted without degrading the arithmetic precision in the fixed decimal point arithmetic operation and a desired frequency characteristic is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital graphic equalizer having a plurality of bandpass filters having a digital structure, and more particularly, to a filtering process which does not reduce fixed-point arithmetic precision even in a bandpass filter having a narrow pass frequency band. With regard to digital graphic equalizer.

[0002]

BACKGROUND OF THE INVENTION Digital graphic equalizer, as shown in FIG. 3, a plurality of digital band-pass filter (BPF1~BPFn) ₁ 1 ~1n digital audio signal AS of the predetermined sampling frequency Fs is input, the bandpass It includes a gain control unit 2 ₁ to 2n for controlling the gain of the signal output from the filter, the combining unit 3 to a respective signal gain control synthesized and output.

[0003] Digital bandpass filter 1 ₁ 1n is constituted by a digital IIR filter, so as to set the frequency characteristic K ₁ Kn having peaks at the center frequency of the band corresponding respectively as shown in FIG. 4 . As shown in FIG. 5, the digital IIR filter converts the input digital audio data AS having the sampling frequency Fs into one.
Delay units 4 ₁ to ₄ 4 for delaying by a sampling time; input digital audio data AS;
~b5 multiplying unit 5 ₁ to 5 ₅ for multiplying the delay unit ₆₁ through ₄ for 1 sampling time increments delaying the output digital audio data AS ', multiplied by a coefficient a1~a5 to the adder output and the delayed signal will be described later multiplying section 7 _{1-7 5,} the multiplying unit 5 ₁
It is constituted by an adder 8 for adding 5 _5, 7 the output of _{2-7 5.} By appropriately determining the multiplication coefficients b1 to b5 and a1 to a5, desired frequency characteristics can be set.

[0004] The gain control unit 2 ₁ to 2n is for controlling the gain according to the set value set by the gain setting unit illustrated which is provided for each band, for example, setting values 6dB
In this case, the corresponding band pass filter output is doubled and output. That is, the gain control unit 2 ₁ to 2n and outputs the multiplying a value corresponding to the set gain bandpass filter output. According to the digital graphic equalizer, the gain controller 2 ₁ the output level of the band-pass filter 1 ₁ 1n the corresponding band in accordance with the set gain for each band
The frequency characteristics of the reproduced sound can be changed by adjusting and outputting at ~ 2n.

[0005]

A digital IIR filter constituting each band pass filter in a digital graphic equalizer is realized by a DSP (digital signal processor). This DSP is
For example, each operation (multiplication, addition) of the IIR filter is executed by a 16-bit fixed-point operation. A numerical value represented by a fixed point has a smaller number of significant figures as the value is smaller, and includes an error.

[0006] Each band-pass filter 1 ₁ 1n digital graphic equalizer has a passband of approximately octave band width, respectively, the pass band width becomes narrower toward the low frequency range. When the pass band becomes narrow, some of the coefficients b1 to b5 and a1 to a5 of the IIR filter constituting the bandpass filter have a small value, and the coefficients include errors as described above because of the fixed-point representation. , The coefficient cannot be set accurately. Also, the difference between the maximum value and the minimum value of the coefficient increases. For this reason, IIR
When a fixed point calculation is performed in the filter processing, a calculation error increases, and the gain cannot be sharply reduced as intended in a cutoff frequency band of the filter, so that desired characteristics cannot be obtained.

For example, a sampling frequency of 44,100 Hz
In the above, the multiplication coefficients b1 to b5 and a1 to a5 of the IIR filter for giving desired frequency characteristics to the bandpass filter having a pass band of 10,000 Hz to 20,000 Hz are as follows. b1 = 1.00000000000000 b2 = -1.55891754415659 b3 = 1.66908529348425 b4 = -0.990951035139676 b5 = 0.36781068945875 a1 = 0.08315986992995 a2 = 0.00000000000000 a3 = -0.16631973985990 a4 = 0.00000000000000 a5 = the same band as the octave band. The multiplication coefficients b1 to b5 and a1 to a5 of the IIR filter for giving similar frequency characteristics to a narrow bandpass filter, for example, a bandpass filter having a pass bandwidth of 100 Hz to 200 Hz are as follows.

B1 = 1.00000000000000 b2 = -3.98972303850606 b3 = 5.96942311860601 b4 = -3.996967608379560 b5 = 0.98997601394542 a1 = 0.00001262346512 a2 = 0.00000000000000 a3 = -0.00002524693024 a4 = 0.00000000000000 a5 = 0.00001262320500 Hz ~ 200
Hz band-pass filter multiplication coefficients b1-b5, a1-
When comparing a5, the pass bandwidth is 10,000Hz to 20,000Hz
Passband is 100Hz better than the bandpass filter of
Multiplication coefficient a1 in a band-pass filter of ~ 200 Hz,
The values of a3 and a5 are very small, and the values of the multiplication coefficients b2 and b5 are large. For this reason, the difference between the maximum value and the minimum value of the coefficient becomes large, and the presence of a coefficient having a small value causes the pass bandwidth to be 100 Hz to 200 Hz.
The fixed-point arithmetic precision of the Hz band-pass filter is reduced, and the gain cannot be sharply reduced as intended in the cut-off frequency band of the filter, so that desired characteristics cannot be obtained.

From the above, it is an object of the present invention to provide a digital graphic equalizer capable of performing a filtering process in a fixed-point arithmetic operation without lowering the arithmetic accuracy even with a band-pass filter having a narrow pass frequency band, and obtaining an intended frequency characteristic. It is to provide.

[0010]

According to the present invention, a plurality of digital band-pass filters to which a digital audio signal of a predetermined sampling frequency is inputted and a gain of a signal outputted from each band-pass filter are controlled. A gain control unit, a synthesis unit that synthesizes and outputs each of the gain-controlled signals, and a down-sampling unit that is provided before a band-pass filter of a predetermined frequency band other than the highest frequency band and reduces a sampling frequency of an input digital audio signal. This is achieved by a digital graphic equalizer with Further, according to the present invention, the above-mentioned object is achieved in such a manner that the ratio f _B / f _S of the pass bandwidth f _B of each band-pass filter to the sampling frequency f _S of the digital audio signal input to the band-pass filter is constant. ,
This is achieved by providing a down-sampling unit for reducing the sampling frequency of the input digital audio signal at a stage preceding each band-pass filter.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION

(A) Outline of the present invention A sampling frequency of 44,100 Hz and a pass band of 10,000 Hz
The multiplication coefficients b1 to b5 and a1 to a5 of the IIR filter forming the bandpass filter of up to 20,000 Hz are as follows. b1 = 1.00000000000000 b2 = -1.55891754415659 b3 = 1.66908529348425 b4 = -0.990951035139676 b5 = 0.36781068945875 a1 = 0.08315986992995 a2 = 0.00000000000000 a3 = -0.166319739859990 a4 = 0.00000000000000 a5 = 0.08315996995
, Each of the multiplication coefficients b1 to b of the IIR filter constituting the bandpass filter having a pass band of 5,000 Hz to 10,000 Hz
5, a1 to a5 are as follows.

B1 = 1.00000000000000 b2 = -1.55891754415659 b3 = 1.66908529348425 b4 = -0.90951035139676 b5 = 0.36781068945875 a1 = 0.08315986992995 a2 = 0.00000000000000 a3 = -0.16631973985990 A4 = 0.00000000000000 a4 = 0.00000000000000 a4 = 0.00000000000000 a5 = 0.0983159 If the ratio f _B / f _S of the pass band width f _B of each band pass filter for f _S is constant _{(f B / f S = 10,000} / 44,100 in the above example), IIR filter constituting the band pass filter of each band Multiplication coefficients b1 to b5 and a1 to a5 have exactly the same value. This means that, even if the pass band is narrow, if the sampling frequency is lowered, the difference between the maximum value and the minimum value of the coefficient does not increase, and it is possible to prevent the calculation error from increasing.

By the way, sampling frequency 44,100Hz
And the multiplication coefficients b1 to b5 of the IIR filter constituting the bandpass filter having a pass band of 5,000 Hz to 10,000 Hz.
a1 to a5 are as follows. And the calculation error increases.

On the other hand, the sampling frequency 689.0625 Hz (=
44,100Hz / 64), and each multiplication coefficient b1 of an IIR filter constituting a bandpass filter having a pass band of 100 Hz to 200 Hz.
B1 = 1.00000000000000 b2 = -2.68104871219106 b3 = 3.20207253870144 b4 = -1.92538798848290 b5 = 0.52516590999064 a1 = 0.03894696653981 a2 = 0.00000000000000 a3 = -0.077893933010000 a4 = And the minimum value is smaller than in the prior art, the calculation error is reduced, and desired characteristics can be obtained. In this case, f _B / f _S is 6,400 / 44,100, and the value of f _B / f _S in the pass band width 10,000Hz~20,000Hz (= 10,0
00 / 44,100). This does not necessarily mean that f _B
If the sampling frequency is lowered in accordance with the bandwidth, the difference between the maximum value and the minimum value of the coefficient becomes small, and the desired characteristic can be obtained by reducing the calculation error, even if / f _{S is} not fixed. Means you can do it.

(B) First Embodiment FIG. 1 shows the configuration of a digital graphic equalizer according to the present invention.
FIG. In the figure, 11 ₁~ 11n is the predetermined sampling frequency
Number of digital audio signals
Filters (BPF1 to BPFn) with smaller subscripts.
It means a high frequency band. Band pass
Ruta 11₁11n are digital IIR filters (see FIG. 5).
). 12₁~ 12n is each band pass
Gain control to control the gain of the signal output from the filter
The control unit 13 synthesizes and outputs each signal whose gain is controlled.
Combining section (adding section), 14_Two~ 14n is below the highest frequency band
Bandpass filter 11 according to outside frequency band_Two~ 1
1n, and is provided before the digital audio signal AS.
Downsampling to reduce sampling frequency Fs
Part, 15_Two~ 15n is folded by lowering the sampling frequency
Low-pass filter that limits the band so that repeated distortion does not occur
It is a filter. In the figure, each down sampling unit 14_Two
~ 14n are connected in a cascade, but digitally
The audio signal AS is applied to each low-pass filter 15_Two~ 15n
Directly to the corresponding
Can be configured to input to the sampling section.
You.

[0016] Fs sampling frequency of the input digital audio data, the highest frequency bandpass filter 11 ₁ pass bandwidth for band F _B1, the pass bandwidth of the other band-pass filter 11 ₂ ~11n F _Bi (i = 2 to n), the downsampling unit 14 _i (i = 2 to n)
In order to obtain a sampling frequency f _i that satisfies the following equation: f _i = (F _Bi / F _B ) · Fs (i = 2 to n)
Reduce the sampling frequency of input digital audio data. By performing down-sampling, the apparent pass band of the band-pass filter with respect to the sampling frequency is widened, so that the filter coefficient of the band-pass filter can be prevented from being reduced, and the calculation error in the fixed-point calculation in the IIR filter processing is reduced. be able to.

As explained in the outline of (A), it is not always necessary to reduce the sampling frequency so as to satisfy the above equation. In short, the band-pass filter 11 _2-1
If reducing the sampling frequency of the audio data inputted thereto band-pass filter 11 ₂ ~11n as the frequency bandwidth of 1n is narrowed, a predetermined effect can be obtained. The low-pass filter 15 ₂ ~15n, if the sampling frequency of the audio data output from the subsequent stage of the down-sampling unit 14 ₂ 14n and f _i, the frequency fi /
Two or more frequency components are restricted to prevent aliasing. Band-pass filter 11 ₁ is subjected to the IIR filter processing to the input digital audio data AS of the sampling frequency Fs, the other band-pass filter 11 ₂ ～11N
It is subjected to IIR filtering on the digital audio data of a predetermined sampling frequency f _{i (i} = 2~n) output from the downsampling unit 14 ₂ 14n, the gain control unit 12 ₁ ~12n separately to each band pass filter outputs The gain is multiplied by the set gain, and the adder 13 adds and outputs the output signals of the respective gain controllers.

(C) Second Embodiment FIG. 2 is a block diagram of another digital graphic equalizer according to the present invention, and the same reference numerals are assigned to the same parts as in the first embodiment of FIG. The difference from the first embodiment is that a down-sampling unit and a low-pass filter are not provided in the preceding stage of all band-pass filters. The downsampling (reduction of the sampling frequency) may be performed only in a stage preceding the bandpass filter where an operation error occurs. Thus, the difference between the maximum value and the minimum value of the filter coefficient is large, is obtained in advance a band-pass filter whose value is contains a small filter coefficients (Fig. 2, the band-pass filter 11 _3, 1
₁₅ ), down-sampling units 14 ₃ , 14 ₅ and low-pass filters 15 ₃ , 15 ₅ are provided only before the band-pass filters 11 ₃ , 11 ₅ . As described above, the present invention has been described with reference to the embodiments. However, the present invention can be variously modified in accordance with the gist of the present invention described in the claims, and the present invention does not exclude these.

[0019]

As described above, according to the present invention, a downsampling unit for reducing the sampling frequency of an input digital audio signal is provided before a bandpass filter in a predetermined frequency band other than the highest frequency band. The generation of numerical values can be prevented, and the difference between the maximum value and the minimum value of the coefficient can be reduced. As a result, even in a band-pass filter having a narrow pass frequency band, the filtering process can be performed without lowering the calculation accuracy of the fixed-point calculation. And a desired frequency characteristic can be obtained.

Further, according to the present invention, the ratio f _B / f _S of the pass band width f _B of each band-pass filter to the sampling frequency f _S of the digital audio signal input to the band-pass filter is made constant. Since the down-sampling unit for reducing the sampling frequency of the input digital audio signal is provided at the preceding stage of each band-pass filter, the filter coefficient values of the band-pass filter composed of the IIR filters can be made the same, and the above-mentioned effect can be further improved. Can be significant.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a digital graphic equalizer of the present invention.

FIG. 2 is a configuration diagram of another digital graphic equalizer of the present invention.

FIG. 3 is a configuration diagram of a conventional digital graphic equalizer.

FIG. 4 is an explanatory diagram of frequency characteristics of each bandpass filter.

FIG. 5 is a configuration diagram of an IIR filter.

[Explanation of symbols]

11 _{1 to} 11 n ··· band pass filter 12 _{1 to} 12 n ··· gain control unit 13 ··· synthesis unit (addition unit) 14 _{2 to} 14 n ··· downsampling unit 15 _{2 to} 15 n ··· low pass filter

Claims (2)

[Claims] 1. A plurality of digital bandpass filters to which a digital audio signal of a predetermined sampling frequency is input, a gain control unit for controlling a gain of a signal output from each bandpass filter, and synthesizing each of the gain-controlled signals. In a digital graphic equalizer having a synthesizer for outputting a digital signal, a downsampling unit for reducing a sampling frequency of an input digital audio signal is provided before a bandpass filter of a predetermined frequency band other than the highest frequency band. Digital graphic equalizer. 2. A way rate f _B / f _S of the pass band width f _B of each band-pass filter for the sampling frequency f _S of the digital audio signal input to the band-pass filter becomes a constant value, for each band-pass filter 2. The digital graphic equalizer according to claim 1, wherein a downsampling unit for reducing a sampling frequency of the input digital audio signal is provided at a preceding stage.