close

Вход

Забыли?

вход по аккаунту

?

JPH0759186

код для вставкиСкачать
Patent Translate
Powered by EPO and Google
Notice
This translation is machine-generated. It cannot be guaranteed that it is intelligible, accurate,
complete, reliable or fit for specific purposes. Critical decisions, such as commercially relevant or
financial decisions, should not be based on machine-translation output.
DESCRIPTION JPH0759186
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
method and apparatus for linear distortion compensation of an acoustic signal, and more
particularly to a method and apparatus for compensating linear distortion of an acoustic signal
generated in a speaker system in view of human visual characteristics. It is. In the present
invention and Patent Application No. 92-15114 filed on August 22, 1992, the audio frequency
band is band-split using a bulk scale described later. However, although the above application
uses a band pass filter for frequency band division, the present invention uses a low pass filter to
divide the audio frequency band into multiple sub-bands and perform linear distortion
compensation for each sub-band.
[0002]
2. Description of the Related Art In general, linear distortion in a speaker system affects the
timbre of a sound. In the case of an ideal loudspeaker system, the sound pressure / frequency
characteristics become flat. However, if distortion occurs in the sound signal due to various
causes, the sound pressure / frequency characteristics are distorted and the timbre of the sound
changes.
[0003]
08-05-2019
1
As prior art for compensating for distortion generated in a speaker system to obtain flat sound
pressure / frequency characteristics, U.S. Pat. No. 4,888,811 such as Takashi, which was
patented on Dec. 19, 1989. There is. The Takashi et al. Patent discloses a technique for
compensating for distortion found in amplitude / frequency characteristics and phase /
frequency characteristics. The input acoustic signal is band-divided into high frequency,
intermediate frequency and low frequency. Then, a flat sound pressure / frequency characteristic
and a linear phase / frequency characteristic can be obtained by the impact response coefficient
determined by the sound pressure / frequency characteristic information and the phase /
frequency characteristic information.
[0004]
Other prior art for distortion compensation of speaker systems is U.S. Pat. No. 4,888,808 to
Ishikawa et al., Which was patented on December 19, 1989. The Ishikawa et al. Patent corrects
the phase and amplitude of the digital acoustic signal passing through the FIR filter based on the
independently established amplitude / frequency and phase / frequency characteristics. In
particular, the patent of Ishikawa et al. Discloses a technique for filtering an acoustic signal using
filtering coefficients calculated for a large number of band pass filters.
[0005]
However, the prior art described above provides constant frequency resolution for the frequency
domain to be compensated. However, if a constant resolution is given to the whole area to be
compensated, if the number of taps of the filter is increased for linear distortion compensation in
the low frequency area, the filter for linear distortion compensation in the high frequency area is
The number of taps also increases.
[0006]
The prior art for performing signal equalization only in the low frequency domain or the
intermediate frequency domain is disclosed in International Publication No. WO 90/00, 851,
such as Nelson, published on January 25, 1990. According to this specification, acoustic signals
in the low or intermediate acoustic frequency band are signal equalized and added to the signal
in the entire acoustic frequency band. Then, it is equalized by the difference between the signal
reproduced by the speaker only from the signal passing through the low pass filter and the signal
08-05-2019
2
passing through the low pass filter. Thus, not only can the number of filter taps for signal
distortion compensation be reduced, but optimal crosstalk suppression and equalization to the
listening location is obtained.
[0007]
SUMMARY OF THE INVENTION Therefore, the object of the present invention is to set the
frequency resolution of the filter for the frequency range sensitive to human beings by the
auditory characteristics higher than the frequency resolution of the filter for the relatively low
frequency range. Accordingly, it is an object of the present invention to provide a linear
distortion compensation method capable of compensating for distortion of an acoustic signal by
adapting to human auditory characteristics while reducing the number of filter taps.
[0008]
Another object of the present invention is to provide a linear distortion compensation device
embodying the above method.
[0009]
SUMMARY OF THE INVENTION The object of the present invention described above is a linear
distortion compensation method for compensating distortion of an acoustic signal generated
from a speaker system, which comprises a bulk scale and a frequency scale according to human
auditory resolution. A resolution setting step of setting frequency resolution differently for an
audio frequency band and at least one sub-band whose maximum frequency is lower than the
audio frequency band using a relationship; and a first corresponding to the audio frequency band
A first filtering step for compensating signal distortion generated in the audio signal in the audio
frequency band based on the frequency resolution of the first sub-band, and a second frequency
resolution corresponding to the first sub-band A second filtering step to compensate for signal
distortion generated in the acoustic signal, and the acoustic filtered by the first and second
filtering steps It is achieved by linear distortion compensation method which comprises an
addition step of adding the items.
[0010]
Here, each of the filtering steps comprises: dividing the corresponding frequency band into equal
intervals based on the corresponding frequency resolution; calculating an FIR filter coefficient
corresponding to each divided position; and the calculated filter coefficient And equalizing the
acoustic signal of the corresponding frequency band based on
08-05-2019
3
The second filtering step equalizes the acoustic signal of the first sub-band based on a second
frequency resolution higher in frequency resolution than the first frequency resolution.
Also, the second filtering step applies a first low pass filtering step of applying an acoustic signal
of an audio frequency band to extract an acoustic signal of a first sub band, and the low pass
filtered first subband. A first decimation step for decimating the acoustic signal at a first
decimation ratio to lower the sampling frequency of the signal, and the acoustic signal whose
sampling frequency has become lower to the second frequency resolution and the first
decimation ratio And a first interpolation step of interpolating and outputting an output signal of
the first FIR filter at an inverse ratio of the first decimation ratio.
[0011]
Furthermore, a third filtering step is performed to compensate for signal distortion generated in
an acoustic signal of a second subband whose maximum frequency is lower than that of the first
subband.
Here, the third filtering step equalizes the acoustic signal of the second sub-band based on a third
frequency resolution higher in frequency resolution than the second frequency resolution. Also,
the third filtering step applies an acoustic signal in an audio frequency band to extract an
acoustic signal in a second sub-band, and the second low-pass filtered second sub-band. The
second decimation step to lower the sampling frequency of the signal by decimating the acoustic
signal of the second decimation ratio larger than the first decimation ratio, and the third
frequency resolution of the acoustic signal whose sampling frequency has become lower A
second FIR filtering step equalized by FIR filter coefficients calculated based on the second
decimation ratio and the second decimation ratio, and a second interpolation step of
interpolating and outputting the output signal of the second FIR filtering step with the second
decimation ratio And.
[0012]
And another object of the present invention is a linear distortion compensation device for
compensating distortion of an acoustic signal generated from a speaker system, wherein an
acoustic signal in an audio frequency band is applied to obtain a predetermined first frequency
resolution. First filter means for compensating for signal distortion, and an acoustic signal in the
08-05-2019
4
audio frequency band is applied to extract an acoustic signal in a first sub-band having a
maximum frequency lower than the audio frequency band, and the first frequency resolution A
second filter means for performing distortion compensation of an acoustic signal in a first subband based on a second frequency resolution with higher resolution; and an adder for adding
output signals of the first and second filter means The present invention is achieved by a linear
distortion compensator characterized by including.
[0013]
Here, the first and second filter means define the frequency value corresponding to the value
divided into equal intervals on the bulk scale as the maximum frequency of the corresponding
band, and set the filter coefficient according to the frequency resolution for each band.
Also, the first filter means comprises an FIR filter. In addition, the second filter means includes a
first low pass filter for extracting an acoustic signal in the first sub-band, and an output signal of
the first low pass filter at a first decimation ratio in a frequency domain. A first decimator for
decimating, and a first FIR filter for filtering an acoustic signal whose sampling frequency is
lowered by the first decimator, by a filter coefficient calculated based on the second frequency
resolution and the first decimation ratio; And a first interpolator which receives the output signal
of the first FIR filter, interpolates at an inverse ratio of the first decimation ratio, and outputs the
interpolated signal to the adder. The second filter means further comprises a delay for
compensating signal delay between the output signal of the first filter means and the output
signal of the first interpolator. The second filter means further comprises a second low pass filter
for removing noise from the output signal of the first interpolator.
[0014]
Furthermore, an acoustic signal of a second sub-band whose maximum frequency is lower than
that of the first sub-band is applied, and the signal distortion is compensated based on a third
frequency resolution higher than the second frequency resolution, thereby adding the adder And
third filter means for outputting Here, the third filter means applies a third low-pass filter for
extracting an acoustic signal of the second sub-band and an output signal of the third low-pass
filter, and A second decimator decimating with a second decimation ratio larger than 1
decimation ratio, and an acoustic signal whose sampling frequency is lower than that of the
second decimator are filters calculated by the third frequency resolution and the second
decimation ratio A second FIR filter for filtering by a coefficient, and a second interpolator which
08-05-2019
5
receives the output signal of the second FIR filter, interpolates at an inverse ratio of the second
decimation ratio, and outputs the interpolated signal to the adder. The third filter means further
comprises a fourth low pass filter for removing noise from the output signal of the second
interpolator. The third filter means further comprises a delay for compensating a signal delay
generated between the output signal of the first filter means and the output signal of the second
interpolator.
[0015]
In the linear distortion compensation apparatus embodied by the present invention, the
resolution of the low frequency region in the audio frequency band is enhanced, and the
resolution is relatively lowered in the high frequency region. In order to increase the resolution
for the low frequency region, the low pass filter performs band decomposition on the entire
audio frequency band. The acoustic signal in the entire audio frequency band that does not pass
through the low pass filter is subjected to distortion compensation by the first filter unit. The
audio signal of the decomposition band obtained by passing through the low-pass filter is filtered
to be able to remove the distortion portion which could not be compensated for the signal
distortion of the whole audio frequency band by the second filter part, Is added to the distortioncompensated acoustic signal. In other words, the distortion-compensated acoustic signal of the
entire audio frequency band is subtracted by the signal obtained from the second filter unit.
Therefore, in the acoustic signal distortion-compensated by the first filter unit, the signal
distortion portion left in accordance with the signal output from the second filter unit is removed.
The characteristics or the number of filters that play the role of the second filter unit are adjusted
appropriately by the degree of signal distortion compensation of the acoustic signal output from
the first filter unit and the human auditory ability.
[0016]
An embodiment of the present invention will be described in detail below with reference to the
attached drawings. 1A and 1B are characteristic diagrams showing distortion generated by the
speaker system. FIG. 1A shows a sound pressure / frequency characteristic in which distortion
occurs, and FIG. 1B shows an impulse response characteristic in which distortion occurs.
[0017]
08-05-2019
6
Most acoustic devices convert electrical signals into mechanical vibrational acoustic signals. In
this process, the signal is distorted by the transfer function of the system. The present
embodiment discloses a multi-resolution linear distortion compensation method and apparatus
for compensating for such distortion. In general, the cochlea, which strikes the human inner ear,
analyzes the frequency of the sound. Since the above-mentioned coiled tube differs in resonance
frequency depending on the position, each position in the coiled tube corresponds to a specific
frequency. And, the frequency scale of the spiral tube is not linear but has an exponential
characteristic. That is, assuming that a human sound analysis system is a filter bank having the
same frequency resolution, the frequency band of each filter constituting the filter bank has a
constant -Q (the bandwidth increases as the center frequency increases. It has the form of
Constant-Q). Therefore, it is easy for a person to detect the frequency change of the acoustic
signal as the frequency is lower, and it is more difficult to detect the frequency change as the
frequency is higher. In this embodiment, distortion of an acoustic signal generated from the
speaker system is compensated by using such human auditory characteristics, with high
frequency resolution in the low frequency band and relatively low frequency resolution in the
high frequency band. Such a multi-resolution linear distortion compensation method of this
embodiment generates the frequency band which provides higher frequency resolution than the
frequency resolution of the audio frequency band, the bulk scale described in FIGS. 2A and 2B is
used. Use.
[0018]
2A and 2B show the relationship between the bulk scale and the frequency scale according to
this embodiment. In FIGS. 2A and 2B, the horizontal axis shows the frequency scale and the
vertical axis shows the bulk scale. Human auditory resolution is shown by the slope of the curve
in the audio frequency band of about 20 Hz to 20 KHz shown in FIG. 2A or 2B. A non-linear
frequency scale according to human auditory characteristics is shown on a linear frequency scale
by the following equation (1) showing the relationship between the bulk scale variable Z and the
frequency f.
[0019]
Z = 7 × 1 n [(f / 650) + {(f / 650) 2 +1} 1/2] (1) In the above equation (1), the bulk value (Z) is
linear with the frequency detection position in the spiral tube Have a corresponding relationship.
In order to define two frequency bands with different frequency resolutions, one embodiment of
the present invention divides the bulk scale at equal intervals as shown in FIG. 2A. Based on the
divided bulk scale, a frequency band corresponding to a bulk value of 29.5 or less is divided into
08-05-2019
7
a "first band", and a frequency band corresponding to a bulk value of 14.8 or less is divided into
a "second band". The first band is identical to the audio frequency band. FIG. 2A shows the
frequency for each of the two bands. The number of taps of the FIR filter used for signal
distortion correction of each band is determined by the following equation (2) based on the
characteristics of the constant -Q.
[0020]
29.5 / N1 = 14.8 / N2 (2) In the above equation (2), N1 is the number of taps of the FIR filter in
the first band, and N2 is the number of taps of the FIR filter in the second band. When the
number of filter taps for the entire audio frequency band is set to 200, the following equation (3)
holds between the total number of taps and the number of taps in each band.
[0021]
N1 + N2 = 200 (3) Accordingly, N1 = 133 and N2 = 67 when the above equations (2) and (3) are
used. When distortion compensation is performed on an acoustic signal using a finite impact
response filter, the frequency resolution to be compensated is a value obtained by dividing the
sampling frequency by the order of the FIR filter. When the sampling frequency fs is 44.1 KHz,
the resolution of the FIR filter is approximately 330 Hz in the first band of N1 = 133. For the
second band, the data sampled at the 44.1 KHz sampling frequency is decimated 7: 1 to a
sampling frequency of 6.3 KHz. The frequency resolution for the 6.3 KHz sampling frequency
and N2 = 67 is approximately 90 Hz.
[0022]
As described above, if the second band acoustic signal is decimated and then subjected to FIR
filtering, the 1: 7 interpolation improves the frequency resolution for the second band acoustic
signal. As a result, since the low frequency band of the audio frequency band can be filtered with
high frequency resolution, efficient signal distortion compensation is possible even with a small
number of filters.
[0023]
08-05-2019
8
Next, signal distortion compensation for each band when the audio frequency band is divided
into three bands having different frequency resolutions will be described. If the bulk scale is
divided at equal intervals to divide the audio frequency band into three bands, as shown in FIG.
2B, from 20 Hz to 22.05 KHz is the “first band” and from 20 Hz to 5.4 KHz The second band
"and from 20 Hz to 1.2 KHz are the" third band ". The frequency resolution of the FIR filter for
each band is calculated based on the following equations (4) and (5) according to the
characteristics of the constant -Q, when the number of filter taps for the total audio frequency is
200:
[0024]
29.5 / N1 = 19.7 / N2 = 9.8 / N3 (4) N1 + N2 + N3 = 200 (5) The number of taps and resolution
for the three bands are two for audio frequency bands having different frequency resolutions It is
calculated by the same method as in the case of dividing into bands, and the result is shown in
Table 1 below.
[0026]
Since the boundary value on the bulk scale used for band division does not limit the present
invention, other bulk values are used to set a region having a higher frequency resolution than
that of the audio frequency band, FIR filtering for each region is also possible within the scope of
the present invention.
An actual arrangement for compensating signal distortion for bands with different frequency
resolutions will be described with reference to FIG. 3 below.
[0027]
FIG. 3 is a block diagram showing a linear distortion compensation apparatus according to a
preferred embodiment of the present invention. In FIG. 3, between the acoustic signal input
terminal 20 and the adder 60, a first filter unit 30, a second filter unit 40, and a third filter unit
40 for distortion compensation of the acoustic signal in each band of FIG. 2B. The filter units 30
are connected in parallel. The adder 60 adds the signals supplied from the respective filter units.
The first filter unit 30 compensates for distortion contained in the first band acoustic signal. The
second filter unit 40 and the third filter unit 50 compensate for distortion of the acoustic signals
08-05-2019
9
in the second and third bands, respectively. Each of the filter units 30, 40 and 50 is provided
with an FIR filter having the number of taps shown in Table 1, respectively.
[0028]
The first filter unit 30 includes one FIR filter for equalizing an audio signal in an audio frequency
band. The second filter unit 40 includes a first low pass filter 41 and a second low pass filter 45
for passing an acoustic signal of 5.4 KHz or less. The output signal of the first low pass filter 41
is input to a first decimator 42 that performs 3: 1 decimation. The first FIR filter 43 receives and
equalizes the output signal of the first decimator 42, and outputs the equalized signal to the first
interpolator 44. The first interpolator 44 restores the 3: 1 decimated acoustic signal to the
original signal. The second low pass filter 45 removes noise contained in the output signal of the
first interpolator 44. The first delay unit 46 delays the output signal of the second low pass filter
45 to synchronize with the output signal of the first filter unit 30.
[0029]
Similar to the second filter unit 40, the third filter unit 50 includes the third low pass filter 51,
the second decimator 52, the second FIR filter 53, the second interpolator 54, the fourth low
pass filter 55, and the second delay. The device 56 is provided. The third low pass filter 51
outputs an acoustic signal having a frequency of 1.32 KHz or less. The second decimator 52 and
the second interpolator 54 are used for 15: 1 decimation and 1:15 interpolation.
[0030]
Before describing the operation of the apparatus of FIG. 3, a process of obtaining filter
coefficients for the FIR filter of the first filter unit 30 will be described. Assuming that the
impulse response of the speaker system is S (n), S (n) measured in the anechoic chamber is
expressed by the following equation (6). The frequency response S (ωk) of the speaker system is
obtained by DFT (Discrete Fourier Transform) of the impulse response S (n).
[0031]
The frequency response F 1 (ω k) for the inverse system of a loudspeaker system having the
frequency response shown in equation (7) is given by the following equation (8).
08-05-2019
10
[0032]
Here, the reason why the value less than 20 KHz is set to "0" is to prevent the speaker from being
damaged by an extreme boost in a constant frequency region.
[0033]
The first filter unit 30 is composed of an FIR filter having the frequency response characteristic
shown in the above equation (8), and compensates for distortion generated by the speaker
system for the entire band of audio frequencies.
In order to determine the filter coefficient of the first filter unit 30, values for N1 (= 98) taps
allocated to the first band in FIG. 2 are selected at equal intervals within the range of 0 to 44.1
KHz, and IDFT Be done.
As a result, N1 coefficients obtained are used as filter coefficients of the first filter unit 30.
Accordingly, the acoustic signal having the frequency of the first band is distortion-compensated
by the first filter unit 30 having the filter coefficient calculated in the above-described manner.
[0034]
However, in order to completely compensate the distortion of the acoustic signal using the first
filter unit 30, the first filter unit 30 needs an FIR filter of infinite order. That is, complete
distortion compensation is not performed by the first filter unit 30 having a limited filter
coefficient. Therefore, in order to compensate for signal distortion that is not compensated by the
first filter unit 30, the second filter unit 40 is used.
[0035]
The second filter unit 40 is configured to have a frequency response F2 (ω) determined by the
following equation (9) in order to compensate for distortion that could not be compensated by
the first filter unit 30. S (ω) F 2 (ω) = − E 1 (ω) (9) Frequency response F 2 (ω) of second filter
08-05-2019
11
unit 40 and frequency response F 2 ′ (ω) of first FIR filter 43 and low pass filter 41, The
relationship between the 45 transfer functions is shown in the following equation (10).
[0036]
F2 (.omega.) =-{(E1 (.omega.) / S (.omega.)) = F2 '(.omega.) L22 (.omega.) Where F2' (.omega.) Has
the effect of the low pass filter removed L2 (ω) means the transfer function of the low pass filter.
The frequency response {F 2 ′ (ω)} of the first FIR filter 43 is determined by the following
equation (11).
[0037]
In the present embodiment, the first FIR filter unit 43 of the second filter unit 40 is configured to
process a signal having a sampling frequency of 14.7 KHz. The first low pass filter 41 receives an
audio signal in the audio frequency band to prevent aliasing, and outputs only an audio signal
having a frequency band of 5.4 KHz or less to the first decimator 42. The first decimator 42
decimates the acoustic signal supplied from the first low pass filter 41 to 3: 1 and reduces the
sampling frequency of the acoustic signal to 14.7 KHz (= 44.1 KHz / 3). As a result, an acoustic
signal with a maximum frequency of 7.35 KHz is obtained. The acoustic signal obtained by
decimation is input to the first FIR filter 43.
[0038]
The first FIR filter 43 filters the acoustic signal with the filter coefficient calculated based on the
above equation (11). The N 2 filter coefficients of the first FIR filter 43 are obtained by equally
selecting frequency values between 0 and 14.7 KHz, and performing IDFT on the selected
frequency values. The filtered acoustic signal is 1: 3 interpolated by the first interpolator 44 and
noise is removed by the second low pass filter 45. The first delay unit 46 receives the output
signal from the second low pass filter 45, delays the output signal so as to be temporally
matched with the corresponding acoustic signal of the first filter unit 30, and outputs the delayed
signal.
[0039]
08-05-2019
12
In fact, the second filter unit 40 can not completely compensate for the error generated from the
first filter unit 30. Therefore, the third filter unit 50 is used to compensate for errors that can not
be compensated by the first filter unit 30 and the second filter unit 40. The number of such filter
units does not limit the present invention, and it is also possible to divide an audio frequency
band into two bands or other number of bands to realize distortion compensation for each band.
It is possible inside.
[0040]
Returning to the description of the third filter unit 50, assuming that the frequency response of
the third filter unit 50 is F3 (ω), the impulse response S (ω) and the frequency response F3 (ω)
in the frequency domain and the error E2 (ω) Have a relationship shown in the following
equation (12). S (ω) F 3 (ω) = − E 2 (ω) (12) Therefore, the frequency response F 3 (ω) of the
third filter unit 50 and the frequency response F 3 ′ (ω) of the second FIR filter 53 and the low
pass filter The transfer functions of 51 and 55 have the relationship shown in the following
equation (13).
[0041]
F3 (ω) =-{(E2 (ω) / S (ω)) = F3 '(ω) L32 (ω) (13) where F3' (ω) is the effect of the low pass
filter is removed L.sub.3 (.omega.) Is the transfer function of the low pass filter. Therefore, the
frequency response F3 '(. Omega.) Of the second FIR filter 53 for distortion compensation of the
acoustic signal belonging to the third band is determined by the following equation (14).
[0042]
The N 3 filter coefficients of the second FIR filter 53 are obtained by IDFTing frequency values
determined by dividing 0 to 2.94 KHz at equal intervals. If the second FIR filter 53 has filter
coefficients determined as described above, the third filter unit 50 processes the signal in the
same manner as the second filter unit 40.
[0043]
08-05-2019
13
First, the third low pass filter 51 outputs, to the second decimator 52, only an acoustic signal
having a frequency band of 1.2 KHz or less from the input acoustic signal. The second decimator
52 performs 15: 1 decimation on the input signal to generate an acoustic signal having a
sampling frequency of 2.94 KHz (= 44.11 KHz / 15). The second FIR filter 53 receives the signal
decimated at 15: 1, filters it using the filter coefficient already set, and outputs it to the second
interpolator 54. The second interpolator 54 complements the input signal to a ratio of 1:15 and
outputs it, and the fourth low pass filter 55 removes noise generated by interpolation and
outputs it. The second delay unit 56 receives the output signal from the fourth low pass filter 56,
delays the output signal so as to temporally coincide with the corresponding acoustic signal of
the first filter unit 30, and outputs the delayed signal. The adder 60 receives the output signals
from the filter units 30, 40, 50, adds them up and outputs them.
[0044]
FIGS. 4A to 4D are characteristic diagrams showing compensation results of distortion generated
from a speaker. FIGS. 4A to 4C show the results of multi-resolution linear distortion
compensation when the audio frequency band is divided into two bands and signal processing is
performed according to this embodiment. That is, a characteristic curve in the case where the
band division using the bulk scale and the filter coefficient determination method shown in the
description of the apparatus of FIG. 3 are applied to two bands is shown. On the other hand, FIG.
4D shows the result of distortion compensation using an FIR filter having 1024 taps according to
the conventional method, ie, a method having constant resolution over the entire band.
[0045]
FIG. 4A shows the result of acoustic signal distortion compensation using an FIR filter whose
number of taps has been determined using bulk scale for the entire audio frequency band. As can
be seen from FIG. 4A, ripples within ± 0.5 dB appear in the high frequency band of 3 KHz or
more. And in the frequency band from 1 KHz to 3 KHz, there is a ripple within about ± 1 dB. On
the other hand, at 1 KHz or less, there is a ripple of ± 1.5 dB or more, so additional distortion
compensation is required.
[0046]
08-05-2019
14
FIG. 4B shows the result of compensation for an acoustic signal falling in the 0-5.2 KHz
frequency band. In FIG. 4B, signal distortion compensation is not performed for a frequency band
of 5.2 KHz or more. On the other hand, in a frequency band of 100 Hz or more in a frequency
band of 5.2 KHz or less, good amplitude / frequency characteristics having a ripple of about ±
0.5 dB are exhibited. In fact, it is difficult to distinguish the original sound and the reproduced
sound at about ± 1 dB.
[0047]
FIG. 4C shows an impulse response characteristic obtained after the distortion generated from
the speaker is compensated by the multi-resolution linear distortion compensator of this
embodiment. When compared with the impulse response of FIG. 1B, it can be seen that it has an
improved response after distortion compensation. The degree of distortion compensation by the
multi-resolution distortion compensator of this embodiment is further clarified by comparing
FIGS. 4B and 4D. FIG. 4D is a result of distortion compensation by a conventional FIR filter having
1024 taps, and FIG. 4B is a distortion compensation result by a multi-resolution linear distortion
compensator having 272 taps.
[0048]
The sampling frequencies of the second filter unit and the third filter unit used in the
embodiment of the present invention described above are not limited to the present invention,
and embodiments using FIR filters having other sampling frequencies are used. Configuration is
also possible within the scope of the present invention.
[0049]
As described above, the linear distortion compensation method and apparatus of the present
invention can reduce the number of taps of the filter while reducing the number of taps of the
filter by dividing the band and determining the filter coefficient using bulk scale. The distortion
of the acoustic signal can be compensated to an extent that can not be distinguished.
08-05-2019
15
Документ
Категория
Без категории
Просмотров
0
Размер файла
28 Кб
Теги
jph0759186
1/--страниц
Пожаловаться на содержимое документа