JP2001326991A - Audio processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce memory capacity and the number of times of calculation of a circuit for processing a low frequency region, in order to amplify sound by a plurality of amplifiers provided for every individual frequency band to output the sound to a plurality of speakers. SOLUTION: Both band separating filters 2041-204k at a side of the low frequency region and speaker property correction filters 2071-207k are composed of warp filters, and both band separating filters 204k+1-204n at a side of the high frequency region and speaker property correction filters 207k+1-207n are composed of FIR filters.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio processing apparatus for dividing an audio input signal into a plurality of frequency bands so that the signals are amplified by a plurality of amplifiers provided for each frequency band and output to a plurality of speakers. .

[0002]

2. Description of the Related Art In a multi-channel speaker system in which an audio input signal is divided into a plurality of frequency bands and a dedicated speaker is driven by each dedicated amplifier for each frequency band, a filter for dividing the signal into a plurality of frequency bands. By using a digital filter instead of an analog filter, a speaker response with a linear phase and a flat frequency amplitude characteristic can be realized.

FIG. 18 shows a conventional example as disclosed in Japanese Patent Laid-Open No. 3-143.
FIG. 1 is a block diagram illustrating a multi-amplifier type speaker system disclosed in Japanese Patent Application Publication No. 195/195. 18, the audio input signal of the analog is of the n A / D converter 6 ₁ to 6 _n
Each is converted into a digital signal, then is bandlimited to each frequency band by the A / D converter ₆₁ each digital signal converted by to 6 _n each band division circuit 4 ₁ to 4 _n by which the n-number Frequency band. Band dividing circuit 4 ₁ to 4 _n are both formed of a FIR filter. Then the band division circuit 4 ₁ to 4 _n output signals of the is applied to the inverse filter 7 ₁ to _7-n, respectively, a speaker to cancel the amplitude-phase characteristic of the speaker unit 3 ₁ to 3 _n which are driven in each band frequency characteristic for an amplitude characteristic of the unit 3 ₁ to 3 _n to the straight flat and phase characteristic is added. Inverse filter 7 ₁ to _7-n is constituted by the band division circuit 4 ₁ to 4 _n similarly to the FIR filter.

Next, an inverse filter 7₁~ 7_nEach output signal
Delay correction circuit 8₁~ 8_nIs applied to
Circuit 4₁~ 4_nAnd inverse filter 7₁~ 7_nTime that occurred in
Delayed to absorb the difference. Next, the delay correction circuit 8
₁~ 8_nAre output from the D / A converter 5 respectively.₁~ 5_nTo
Is converted to an analog signal, and then the D / A converter 5 ₁
~ 5_nThe analog signals converted by
2₁~ 2_nAmplified by the speaker unit 3₁~ 3_nTo
Applied. In this system, the band dividing circuit 4₁Inverse to ~ 4n
Filter 7₁~ 7_nIs constituted by an FIR filter.
The amplitude characteristics of each frequency band are flattened and phased with high accuracy.
The characteristics can be linear.

[0005]

However, in the above-mentioned conventional system, the band division circuits 4 ₁ to 4 _n and the inverse filter 7 _{1 are used.}
Since to _7-n is constituted by a FIR filter, there is a problem that requires a delay memory with a number of product-sum operation. FIG. 3 shows an FIR filter connected to a digital signal processor (DS).
P) shows the configuration in the case of realization. In FIG. 3, the input signal S11 is a delay memory (z ^-1 ) 11 ₁ to 1
Each sample is delayed by 1 _k-1 for each sample is stored, the input signal S11 and the delay memory 11 ₁ to 11 _k-1 the samples each coefficient multiplier 12 which is stored in the _0-1
Each coefficient stored in the DSP is multiplied by 2 _{k -1} . Next, each multiplication result is added by the adder 13,
This addition result S13 is the output of the FIR filter. The number of product-sum operations is called the tap length of the FIR filter, and is k taps in FIG.

Here, in order to filter relatively low frequency components, a long signal needs to be stored, so that the filter coefficient length and the number of product-sum operations increase. For this reason, it is disadvantageous for processing that handles a wide frequency range such as hearing. Considering that the lower limit of the audible frequency band is 20 Hz, when an FIR filter having a frequency resolution of 20 Hz is realized, the tap length k is 2400 (assuming the sampling frequency fs to be 48 kHz). This processing requires a DSP having an operation speed of 105.2 MIPS, and it is difficult to realize the present general-purpose one-chip DSP.

As a conventional example of reducing the calculation load in the low frequency band, for example, Japanese Patent Application Laid-Open No. 7-59186 proposes a method using a different sampling frequency fs for each band, and the configuration is shown in FIG. However, in this method, decimation processing (first and third decimators in the drawing) for reducing the sampling frequency fs of the band-divided signal and interpolation processing for returning the sampling frequency fs of the signal whose characteristic has been corrected to the original state (1st and 3rd interpolators in the figure) and filtering processing (2nd and 4th LPFs in the figure) to prevent aliasing are required.

SUMMARY OF THE INVENTION The present invention has been made in consideration of the above-described problems of the related art, and has as its object to provide an audio processing device capable of reducing the memory capacity and the number of operations of a circuit for processing low frequencies. In particular, an object of the present invention is to provide an audio processing device capable of realizing band division processing and speaker characteristic correction processing with simple hardware.

[0009]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention is designed to divide the low frequency band by a warp filter having a high resolution in the low frequency range even with a small amount of operation and to correct the speaker characteristics. . That is, according to the present invention, one or more warp filters that divide the low-frequency side of the audio input signal into one or more frequency bands, and one that divides the high-frequency side of the audio input signal into one or more frequency bands. An audio processing device including the above FIR filter is provided.

Further, according to the present invention, one or more warp filters for correcting the low-frequency characteristic of each of the audio signals divided into a plurality of frequency bands for each band according to the characteristics of the loudspeaker, There is provided an audio processing device provided with one or more FIR filters for correcting, for each band, the characteristics of a high frequency band of the audio signal divided into frequency bands in accordance with the characteristics of a speaker.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a first embodiment of the audio processing device according to the present invention.

In FIG. 1, an analog audio input signal is converted into a digital signal by an A / D converter 206, and the digital signals converted by the A / D converter 206 are then converted into band division filters 204 _{1 to} 204 _k , respectively.
Bands are limited to individual frequency bands by 204 _{k + 1 to} 204 _n and divided into n frequency bands. Next, the output signals of the band division filters 204 _{1 to} 204 _k and 204 _{k + 1 to} 204 _n are respectively applied to the speaker characteristic correction filters 207 ₁ to 207 ₁ .
207 _k , 207 _{k + 1 to} 207 _n are applied to the speaker units 203 _{1 to} 203 _k and 203 driven for each band.
_k + 1 _{~203 n} frequency characteristic for cancel the amplitude phase characteristic to the linear flat and phase characteristic of the amplitude characteristic of the speaker unit _{203 1 ~203 k, 203 k +} 1 ~203 n of is added.

The low-frequency band division filters 204 _{1 to} 20 4
4 _k and the speaker characteristic correction filters 207 _{1 to} 207 _k are each configured by a warp filter as shown in detail in FIG. 2, and the band dividing filters 204 _{k + 1 to} 204 _n on the high frequency side.
Each of the speaker characteristic correction filters 207 _{k + 1 to} 207 _n is constituted by an FIR filter as shown in detail in FIG. Note that the band splitting filter 20 on the lowest band side
4 ₁ may even low-pass filter (LPF), also the band division filter 204 _n the highest frequency side may be a high pass filter (HPF).

The speaker characteristic correction filter 207₁~ 20
7_k, 207_{k + 1}~ 207_nThe output signals of
Positive circuit 208₁~ 208_k, 208_{k + 1}~ 208_nApplied to
And the band division filter 204₁~ 204_k, 204_{k + 1}~
204_nAnd speaker characteristic correction filter 207₁~ 207_k,
207_{k + 1}~ 207_nAbsorb the time difference that occurred in
Is delayed. Next, the delay correction circuit 208₁~ 20
8_k, 208_{k + 1}~ 208 _nOutput signals are D / A
Converter 205₁~ 205_k, 205_{k + 1}~ 205_nBy
Converted to a analog signal, and then to a D / A converter 205₁~
205_k, 205_{k + 1}~ 205_nAnalog converted by
The signal is the amplifier 202₁~ 202_k, 202_{k + 1}~ 2
02_nAmplified by the speaker unit 203₁~ 20
3_k, 203_{k +1}~ 203_nIs applied to

Here, in general, the FIR filter has a uniform frequency resolution characteristic over all frequencies, but when the audio signal processing is performed for each band as in the present invention, the frequency resolution is uniform over the entire band. When the band is limited or after the band is limited, it is only necessary that the signal in each of the limited bands can be processed with an optimal frequency resolution from an aural point of view for the band. That is, the frequency resolution in the band is fine, and the frequency resolution is coarse in other bands (high band), and the purpose can be sufficiently achieved, and the warp filter satisfies this purpose in the low band. For the principle of operation of the warp filter, see "Comparison of Lou" by Matti Karjalainen et al.
dspeaker Equalization Methods Based on DSP Techniq
ues ", J. Audio Eng. Soc., Vol. 47, No. 1/2, (1999
Jan./Feb.) Pp. 14-31, but I will briefly explain the basic operation here.

The configuration of the warp filter will be described with reference to FIG. The input signal S15 is delayed by delay elements (D ₁ ) 15 ₁ to 15 _l−1 constituted by an all-pass filter, and the input signal S15 and the signal delayed by the delay elements 15 ₁ to 15 _l−1 are each a coefficient multiplier. It is multiplied by each coefficient by 16 ₀ ~16 _l-1. Next, each multiplication result is added to the adder 17.
And the addition result S17 is the output of the warp filter.

Here, the delay element is configured to simply delay one sample in the FIR filter shown in FIG. 3, but is characterized in that it is configured by an all-pass filter in the warp filter. The all-pass filter is a filter that can change the phase characteristic while maintaining the frequency-amplitude characteristic. As an example, FIG. 4 shows the configuration of a primary all-pass filter. The transfer function D ₁ (z) can be expressed by the following equation (1).

[0018]

(Equation 1)

In the equation (1), λ is called a warping parameter and is a parameter for determining the resolution in the frequency band. When the frequency characteristic of the all-pass filter is calculated from Expression (1), the amplitude characteristic is always 1, while
The group delay characteristic depends on the frequency as shown in FIG. The group delay characteristic can be changed by appropriately setting the warping coefficient λ. By setting λ> 0, a delay element having a longer delay time in a lower frequency band can be configured. This all-pass filter is shown in FIG.
One sample delay memory 11 ₁ to 1 in R filter
By replacing with 1 _k-1 , a filter having a fine frequency resolution in a low band and a coarse frequency resolution in a high band can be formed.

Band splitting filter 2 by warp filter
An example of 04 _{1 to} 204 _k will be described. FIG. 6 shows the frequency amplitude response of the band division filters 204 _{1 to} 204 _k using the 256 tap warp filter (λ = 0.9), and FIG. 7 shows the band division filter 204 _{k +} using the 4096 tap FIR filter. ₁ shows a frequency amplitude response of _{1 to} 204 _n . The cut-off frequency f _kL on the low frequency side is 100 Hz, the cut-off frequency f _kH on the high frequency side is 200 Hz, and the sampling frequency f
s is 44.1 kHz. FIGS. 6 and 7 show that almost the same frequency amplitude response can be obtained in the pass band.

FIG. 8 shows the time response of the band division filters 204 _{1 to} 204 _k using the same 256 tap warp filter, and FIG. 9 shows the band division filter 204 _{k + 1} using the same 4096 tap FIR filter. 4 shows the time response of ２０４204 _n . As shown in FIG. 8 and FIG. 9, the two time responses almost coincide with each other, which indicates that the warp filter can control not only the frequency amplitude response but also the phase characteristic.

Next, an example of the speaker characteristic correction filters 207 _{1 to} 207 _k using the warp filter will be described. FIG.
Indicates the frequency amplitude response characteristics of the speaker units 203 _{1 to} 203 _n. To correct these characteristics, the speaker characteristic correction filters 207 _{1 to} 207 _k using the same 256-tap warp filter and the same 4096-tap FIR filter are used. Used speaker characteristic correction filter 207 _{k + 1} to 2
07 _n respectively 11 each frequency amplitude response of FIG. 1
It is shown in FIG. 13 and 14 show frequency amplitude response characteristics corrected by the respective filters.
From the figure, it can be seen that the FIR filter is superior in the correction in the high frequency range, but the correction accuracy in the low frequency range is almost the same as that of the FIR filter even if a warp filter is used. Therefore, by using the warped filter, it is possible to realize a band division filter 204 ₁ to 204 _k and the speaker characteristic correction filter 207 ₁ to 207 _k in the low-pass tap number fewer taps than FIR filters.

Next, FIG. 15 shows a comparison result between the operation cost of the filter and the amount of used memory in the embodiment and the conventional example.
Shown in The number of operation instructions per tap is four times that of the warp filter compared to the FIR filter.
In the low frequency range, the tap length beyond the increase in the number of instructions can be reduced, so that the total number of instructions can be reduced. Further, since the tap length can be reduced, the memory can also be saved.

Here, a description will be given of a range from the low band to the high band in which the warp filter is to be formed, by taking as an example a case where the band is divided by an octave band width. f
If s = 48 kHz and up to 24 kHz are divided into 10 bands and the lowest band (band = 1) processing is realized by an 8192 tap FIR filter, the frequency resolution in this band is about 5.9 Hz. The ratio of the upper limit frequency f _1H in band = 1 to the frequency resolution f _1R in this band is f _1H / f _1R = about 46.9 / about 5.9 = 8. FIG. 16 shows the number of taps of the FIR filter required in the processing of each band under the condition that the ratio = 8 is maintained in consideration of the property that the auditory sensitivity is logarithmic. As shown in FIG. 16, the required number of taps of the FIR filter can be reduced as the frequency becomes higher.

On the other hand, in the case of a warp filter, the warping coefficient λ is selected according to each band, so that the number of taps required in any band is substantially constant. As described above, a 4096 tap FIR filter can be replaced by a 256 tap warp filter.
Since 1024 instructions are required per fs, the critical frequency for determining whether to use the FIR filter or the warp filter (k, k + 1 shown in FIG. 1) is 375 Hz. Therefore, the vicinity of this frequency may be used as a criterion for determining whether to use the FIR filter or the warp filter.

The point that the characteristics of the speaker are corrected using the warp filter is described in the above Matti Karjalai.
"Comparison of Loudspeaker Equalizatio" by nen et al.
n Methods Based on DSP Techniques ", which describes a method for collectively compensating for speaker characteristics for all characteristics. In contrast, in the present invention, a warp filter has a small amount of computation in a low frequency range. However, the point that attention is paid to the property of high resolution, and the point where the warp filter is used is limited to a low frequency band is greatly different.

Next, a second embodiment will be described with reference to FIG. FIG. 17 is basically the same as the first embodiment (FIG. 1), but a plurality of speaker units 203 ₁
Of ~203 _n, only the bandwidth = 1 to m required characteristic correction speaker characteristic correction filter 207 ₁ to 207 _m is provided. That is, after the processing up to the band division is performed in the same manner as in the first embodiment, the speaker unit 2 that needs characteristic correction is required.
03 ₁ ~203 _m is only bandwidth = 1 to m and characteristic correction belonging, signals of other bands = m + 1 to n is applied as it is passed to the next stage of the delay correction circuit 208 ₁ ~208 _n. According to the second embodiment, the loudspeaker characteristic correction filters 207 _{1 to} 207 _m are provided only in the band = _{1 to m to} which the loudspeaker units 203 _{1 to} 203 _m requiring characteristic correction belong, so that the calculation load is further reduced. be able to.

[0028]

As described above, according to the present invention, the low frequency band is divided and the speaker characteristics are corrected by the warp filter having a high resolution even with a small amount of calculation.
It is possible to reduce the memory capacity and the number of operations of the circuit for processing low frequencies.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of an audio processing device according to the present invention.

FIG. 2 is a block diagram illustrating a configuration of a warp filter of FIG. 1;

FIG. 3 is a block diagram illustrating a configuration of the FIR filter of FIG. 1;

FIG. 4 is a block diagram illustrating a configuration of an all-pass filter as the delay element in FIG. 2;

FIG. 5 is a graph showing group delay characteristics of the all-pass filter of FIG.

FIG. 6 is a graph showing a frequency amplitude response characteristic of a warp filter included in the low band side band division filter of FIG. 1;

FIG. 7 is a graph showing a frequency amplitude response characteristic in a case where the low band side division filter of FIG. 1 is configured by an FIR filter.

FIG. 8 is a graph showing a time response of a warp filter included in the low-frequency band division filter of FIG. 1;

FIG. 9 is a graph showing a time response when the low-frequency band division filter of FIG. 1 is configured by an FIR filter.

FIG. 10 is a graph showing a frequency amplitude response characteristic of the speaker unit.

FIG. 11 is a graph showing a frequency amplitude response characteristic when the speaker characteristic correction filter is constituted by a warp filter.

FIG. 12 is a graph showing a frequency amplitude response characteristic when the speaker characteristic correction filter is constituted by an FIR filter.

FIG. 13 is a graph showing a corrected frequency amplitude response characteristic when the speaker characteristic correction filter is constituted by a warp filter.

FIG. 14 is a graph showing a corrected frequency amplitude response characteristic when the speaker characteristic correction filter is constituted by an FIR filter.

FIG. 15 is an explanatory diagram showing a comparison result between the calculation cost of the filter and the amount of used memory in the embodiment and the conventional example.

FIG. 16 is an explanatory diagram showing the number of taps of an FIR filter required in each band.

FIG. 17 is a block diagram illustrating an audio processing device according to a second embodiment.

FIG. 18 is a block diagram showing a conventional audio processing device.

FIG. 19 is a block diagram showing another conventional audio processing device.

[Explanation of symbols]

204 _{1 to} 204 _k band division filter (warp filter) 204 _{k + 1 to} 204 _n band division filter (FIR filter) 207 _{1 to} 207 _k speaker characteristic correction filter (warp filter) 207 _{k + 1 to} 207 _n speaker characteristic correction filter (FI
R filter)

Claims (2)

[Claims] 1. One or more warp filters each of which divides a low-frequency side of an audio input signal into one or more frequency bands, and one or more which respectively divides a high-frequency side of the audio input signal into one or more frequency bands. An audio processing device comprising: an FIR filter. 2. An audio signal divided into a plurality of frequency bands, one or more warp filters for respectively correcting low-frequency characteristics for each band according to speaker characteristics, and divided into a plurality of frequency bands. An audio processing apparatus comprising: one or more FIR filters for correcting, for each band, characteristics of a high-frequency band of the audio signal according to characteristics of a speaker.