close

Вход

Забыли?

вход по аккаунту

?

DESCRIPTION JP2007067578

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 JP2007067578
PROBLEM TO BE SOLVED: To avoid or reduce an excessive clip of an input audio signal boosted
by an equalizer or an effector instantly. SOLUTION: A variable phase shifter 113 to which an
audio signal from a filter 112 such as an equalizer or an effector is input changes a phase angle
of the input signal according to a control signal. The output from variable phase shifter 113 is
temporarily stored in output buffer 114. The control unit 115 determines whether or not there is
a clip in the output temporarily stored in the output buffer 114, outputs a control signal
according to the determination result, and controls the phase angle of the variable phase shifter
113. The control unit 115 variably controls the phase angle of the variable phase shifter 113
until the output temporarily accumulated in the output buffer 114 is not clipped. [Selected
figure] Figure 1
Audio signal processing apparatus and method
[0001]
The present invention relates to an audio signal processing apparatus and method for avoiding or
reducing clipping of an input audio signal.
[0002]
In general, when a digital audio signal is boosted by using a filter such as an equalizer to
reproduce a specific band, depending on the signal waveform of the source, if clipping is
performed beyond the corresponding bit number of the D / A converter There is.
10-04-2019
1
Conventionally, with respect to this clip, limiter processing is performed on the clip location
according to the value of the maximum bit, or dynamics control (compressor) is used as an input
of a D / A converter, etc. Although known, all have been accompanied by significant sound
quality deterioration.
[0003]
As a prior art, Patent Document 1 discloses a technique for suppressing the deterioration of S / N
so that a signal is not clipped at the time of characteristic change.
[0004]
JP, 2002-345075, A
[0005]
The present invention has been proposed in view of such conventional circumstances, and an
audio signal processing apparatus and method capable of instantaneously avoiding / reducing an
excessive clip of an audio signal boosted by an equalizer, an effector or the like. Intended to
provide.
[0006]
In order to solve the above-described problems, the present invention determines whether there
is a clip in the variable phase shift unit that changes the phase angle of the input audio signal
according to the control signal, and the output from the variable phase shift unit. And a control
unit for outputting the control signal for controlling the phase angle of the variable phase shift
unit according to the determination result.
[0007]
The variable phase shift section comprises a phase shift filter for changing the phase angle of the
input signal by 90 °, delay means for delaying the input signal by a time corresponding to the
delay amount of the phase shift filter, and these phase shift filters. And multiplying means for
multiplying the output from the delay means and the output from the delay means by a
weighting gain, and variably controlling the weighting gain by a control signal from the control
unit.
10-04-2019
2
It is preferable to use a shift Hilbert transform filter as the phase shift filter.
[0008]
According to the present invention, even when clipping occurs in the input audio signal, by
changing the phase angle, the amplitude of the entire waveform can be suppressed, and
excessive clipping of the input signal can be avoided or reduced instantaneously. .
[0009]
Hereinafter, specific embodiments to which the present invention is applied will be described in
detail with reference to the drawings.
[0010]
FIG. 1 is a block diagram showing a schematic configuration of an audio signal processing
apparatus according to an embodiment of the present invention.
In FIG. 1, an audio signal from a source (such as a sound source) 111 is sent to a filter 112 such
as an equalizer or an effector to be filtered.
During this filtering process, for example, the waveform may exceed the allowable number of bits
(maximum number of bits) of the D / A converter 116 in the subsequent stage.
This is also referred to as a so-called waveform clip. In the prior art, this clip was subjected to
limiter or compressor processing (dynamics control) according to the maximum number of bits,
but these sound quality degradation It was supposed to be accompanied.
[0011]
In the embodiment of the present invention, the output audio signal from the filter circuit 112 is
sent to the variable phase shifter 113 whose phase shift amount (rotational phase angle) changes
according to the control signal. Is temporarily stored in the output buffer 114, the control unit
10-04-2019
3
115 determines whether the clip is generated or not, and the phase shift amount of the variable
phase shifter 113 is controlled according to the determination result to generate the clip. To
avoid.
This is in consideration of the fact that clipping is avoided due to the dispersion of the peaks of
each frequency component in the signal when the phase of the audio signal in which the clip is
generated is shifted, and the audibility characteristics of the audio signal. .
That is, although the human ear is sensitive to the phase difference between the left and right
ears, it is noted that the ear is not sensitive to the “phase change of the audio signal itself”
without the “phase difference”, and the phase control is mainly performed. It is intended to
avoid the clip.
[0012]
The output from the output buffer 114 is converted to an analog audio signal by the D / A
converter 116 and converted from an electrical signal to an acoustic signal by the speaker 117.
In the example of FIG. 1, a signal processing apparatus for processing a digital audio signal is
assumed, the source 111 is a digital sound source, the filter 112 and the variable phase shift unit
113 perform digital processing, and the output buffer 114 Although a digital memory or the like
is used, it goes without saying that all or part of these may be replaced with an analog circuit.
[0013]
The variable phase shifter 113 changes only the phase without changing the level of each
frequency component, and is a kind of so-called all-pass filter. This variable phase shifter 113
creates a waveform (phase shifted) having an arbitrary phase delay or phase lead, for example,
over all frequencies or within the required frequency band. It is a phase conversion means, and is
further controlled to change the amount of phase shift (the angle of the phase) in accordance
with a control signal supplied to the control terminal. A specific example of the variable phase
shifter 113 will be described later.
[0014]
10-04-2019
4
The control unit 115 detects the amplitude or the peak value of the audio signal stored in the
output buffer 114 to determine whether the clip is generated or not, and the variable phase
shifter 113 is controlled so as to eliminate the generation of the clip. A control signal for
changing the phase shift amount is output and sent to the control terminal of the variable phase
shifter 113. Specifically, for example, when the control unit 115 determines that there is a clip, a
control signal for controlling the phase shift amount of a predetermined amount is sent to the
variable phase shifter 113, and the variable phase shifter 113 controls the control unit 115. And
shift the phase of the input signal (the output audio signal from the filter circuit 112) by the
predetermined amount, and send the phase-shifted output signal to the output buffer. The
control unit 115 detects whether or not a clip is generated by detecting the amplitude or peak
value of the phase-shifted output signal stored in the output buffer 114, and when it is
determined that a clip is present, A control signal for further changing the phase shift amount
from the predetermined amount is output, and this is repeated until it is determined that there is
no clip. A specific example of the operation in the control unit 115 will be described later.
[0015]
Here, in order to prevent the clipped waveform from being sent to the D / A converter 116 at all,
a buffer is provided on the input side of the variable phase shifter 113, and the audio signal
stored in the output buffer 114 is While the clip is occurring, the audio signal from the buffer on
the input side of the variable phase shifter 113 is repeatedly used to rewrite (overwrite) the
phase-shifted audio signal to the output buffer 114. When there are no clips in the audio signal
of (1), an audio signal may be sent from the output buffer 114 to the D / A converter 116.
[0016]
Next, FIG. 2 shows another example of the audio signal processing apparatus according to the
embodiment of the present invention, and in this example of FIG. 2, a variable phase shifter 120
corresponding to the variable phase shifter 113 of FIG. As a phase shift filter that rotates 90 °
(90 °) with respect to the input signal, a signal subjected to shift Hilbert transform filter
processing and a signal subjected to delay processing equivalent to the delay of the filter
processing A configuration is used in which phase conversion processing is performed by
outputting an appropriate combination of component ratios.
[0017]
In FIG. 2, an audio signal from a source (such as a sound source) 111 is sent to a variable phase
shifter 113 via a filter 112 such as an equalizer, and an output from the variable phase shifter
113 is sent via an output buffer 114 It is converted into an analog audio signal by the D / A
converter 116 and reproduced as an acoustic signal by the speaker 117.
10-04-2019
5
[0018]
The variable phase shifter 120 is also referred to as a shift Hilbert transform filter (hereinafter,
Hilbert filter) to which an input signal (output signal from the filter 112) is supplied.
121, a buffer 122 for temporarily storing the output from the Hilbert filter 121, a multiplier 123
for weighting (gain control) the output from the buffer 122, and the input signal supplied, which
corresponds to the signal delay in the Hilbert filter 121 Delay circuit 124, a buffer 125 for
temporarily accumulating the output from delay circuit 124, and a multiplier 126 for weighting
the output from buffer 125, and the output from each multiplier 123, 126 is an output. The
phase shift at an angle determined by the ratio of the weight of each component by each
multiplier 123, 126 is realized by being sent to the buffer 114 and being added.
[0019]
The shift-type Hilbert filter (Hilbert transform filter) 121 used in the variable phase shifter 120
will be described below.
This filter is intended for an FIR (finite impulse response) filter that operates with a 90 ° phase
delay.
Specifically, the impulse response of the Hilbert transform is used as the source.
[0020]
In general, the transfer function H (ω) of the Hilbert transform is represented by the following
equation (1), and has a characteristic of 90 ° (90 °) delay regardless of the frequency in the
positive frequency domain .
[0021]
10-04-2019
6
[0022]
Also, the impulse response h (t) of this transfer function H (ω) can be represented by h (t) = 1 /
πt (a hyperbolic curve).
[0023]
The example which calculated | required this impulse response in the digital domain is shown to
FIG. 3 (A)-(C).
FIG. 3 (A) shows the impulse response of a Hilbert transform filter consisting of 1024 samples,
and FIGS. 3 (B) and 3 (C) show the vicinity of the beginning and end of FIG. 3 (A) in an enlarged
manner, respectively. It is
Strictly speaking, the amplitude characteristics at the low and high end sides within the sampling
frequency of the Hilbert transform filter will fall because it is made with a finite impulse
response, but this is a source (sound source) There is no problem at all when operating with, for
example, Fs = 48 kHz, Fs = 44.1 kHz, etc., in consideration of being a music signal targeting an
audible band of 20 Hz to 20 kHz.
[0024]
Next, since the Hilbert filter of the impulse response shown in FIG. 3 is a filter in a circulatory
system and does not satisfy the causality, a filter with a cyclic shift in the sample is regarded as a
“shift Hilbert filter”. The coefficients are FIR (Finite Impulse Response) filters.
The waveforms in this case are shown in FIG. 4 and FIG. 4 and 5 show examples shifted by 512
samples of 1024, and FIG. 4 shows normal impulses shifted by the same amount according to
this cyclic shift, and FIG. 5 shows impulses. It shows a response. From the above, basically, the
filter processing results of the FIR filter (the delay circuit 124 and the Hilbert filter 121 in FIG. 2)
using the coefficients in FIGS. 4 and 5 have the same time delay, and the coefficients in FIG. In
the Hilbert filter 121 using the above, the waveform is 90 ° delayed in phase. The number of
samples of the cyclic shift is not limited to the above-mentioned 512 samples, and can be set
arbitrarily. For example, as shown in FIG. 6, the shift may be performed.
10-04-2019
7
[0025]
Next, with respect to the FIR filter (the delay circuit 124 and the Hilbert filter 121 of FIG. 2)
having the coefficients of FIGS. 4 and 5 as coefficients, the phase lag and phase lead of any angle
can be obtained by weighting and combining both processing signals. It is possible to create a
waveform. This is because, as shown in FIG. 7, the sum of vector a (the output vector on the delay
circuit 124 side) and b (the output vector on the Hilbert filter 121 side) of orthogonal systems
whose phases are mutually different by 90 ° in complex space It is also apparent from the fact
that phases other than 90 ° can be expressed. In addition, even if it is rotation phase angle
(except for 0 degree) except 90 degrees, it is also possible to obtain arbitrary rotation phase
angles by carrying out weighted addition in combination with 0 degree of simple delay.
[0026]
Here, referring to the variable phase shifter 120 of the audio signal processing device of FIG. 2,
the weighting gain of the multiplier 126 to which the output from the delay circuit 124 which is
the delay system FIR filter of the coefficient of FIG. The weighting gain of the multiplier 123 to
which the output from the shift-type Hilbert filter 121, which is the FIR filter of FIG. Assuming
that D ° (D °), the following equation is established between them.
[0027]
Ga = cos (2πD / 360) Gb = -sin (2πD / 360) Here, the negative sign (-) added before sin on the
Gb side is 90 ° (90 °) of the basic phase of the Hilbert transform. ) (Due to the downward
direction in FIG. 7).
[0028]
As described above, it has been described that the input signal waveform can be manipulated at
an arbitrary angle phase.
Next, a specific example of the control operation for determining the phase angle in the control
unit 115 of the audio signal processing device of FIG. 2 will be described with reference to FIG.
[0029]
10-04-2019
8
The control operation for determining the phase angle in the control unit 115 is to determine the
phase angle through the processing procedure as shown in the flowchart of FIG. 8 and repeat
this.
In order to simplify the explanation, it is assumed that the internal processing here is performed
by floating point processing, and if the data peak value of the audio signal is larger than 1.0,
clipping occurs in the D / A converter 116. Assume that Also, basically, the signal processing
operation in this specific example is performed in block units, and each buffer stores samples of
at least one block. The number of samples in one block also affects the delay of the entire system.
Here, for the sake of simplicity, the block processing unit is set to 1024 samples.
[0030]
In the first step S101 of FIG. 8, the control unit 115 of FIG. 2 sets the rotational phase angle (the
amount of phase shift) D in the variable phase shifter 120 to 0 ° (D = 0). This corresponds to
setting the weighting gains Ga and Gb of the multipliers 126 and 123 in the variable phase
shifter 120 to 1.0 and 0.0, respectively. In the next step S102, all the output buffers 14 are filled
at this set angle. That is, the output from the variable phase shifter 120 (sum of the outputs from
the multipliers 126 and 123) at the set angle is stored in the output buffer 114 (one block). In
the next step S103, it is determined in the output buffer 114 whether or not a clip has occurred
for the accumulated (the one block worth) output. That is, it is determined whether or not the
maximum peak value of accumulated data of one block is larger than 1.0. When it is determined
as YES (with clip) in step S103, the process proceeds to step S104, and when it is determined
with NO (no clip), the process proceeds to step S119 after a routine R110 described later is
performed as necessary.
[0031]
In step S104 and subsequent steps, the rotational phase angle (phase shift amount) D of the
variable phase shifter 120 is sequentially changed to determine the presence or absence of a clip
for the output from the variable phase shifter 120 at the changed setting angle, Repeat this until
there are no clips. That is, in step S104, as the rotational phase angle D, for example, a plurality
of setting angles serving as options predetermined as 15 °, 30 °, 45 °,. And the output (one
block) from the variable phase shifter 120 at this set angle is sent to the output buffer 114 to
10-04-2019
9
update the data of the block. In step S105, it is determined again whether or not a clip is
generated for the updated data of one block. The setting angle may be changed not only in the
positive direction but also in the negative direction, and instead of preparing a plurality of
options in advance as described above, a certain angle (for example, 15 °) may be used. The
current setting angle may be added to or subtracted from the current setting angle to obtain the
next setting angle. The weighting gains Ga and Gb of the multipliers 126 and 123 are determined
in accordance with the setting angle, but a plurality of sets of options of the gains Ga and Gb are
prepared in advance, or the setting angle is constant Of course, the values of the gains Ga and Gb
to be added or subtracted may be determined by calculation.
[0032]
In the next step S105, it is determined whether or not the clip has been canceled for the output
from the variable phase shifter 120 at the changed set angle, and if YES (no clip), the process
proceeds to step S119, and if NO The process proceeds to step S106. In step S106, it is
determined whether or not all the preset angles prepared in advance have been selected
(whether or not they have totally hit). If NO, the process returns to step S104, and if YES, the
process proceeds to step S107. When the set angle is changed by being added or subtracted by a
constant angle, it may be determined in step S106 whether or not the original set angle is
returned. Further, the range of change of the set angle or the number of times of change is
determined in advance, and when exceeding the range or number of change of this range, for
example, the range of the first set angle ± 45 ° before change, YES in step S106 And the
process may proceed to step S107.
[0033]
In step S107, in the processing up to this point, the clip has been generated at any of the
changed set angles, but at the set angle at which the amount of clip among these is the smallest,
the variable phase shifter 120 Output to the output buffer for storage (updating the data in the
block). When a large buffer capacity can be obtained, all block data at each setting angle changed
so far may be accumulated, and block data of the setting angle at which the clip amount is the
smallest may be used. . In the next step S108, if necessary, clipping processing such as limiter
processing and compressor processing (dynamics control) similar to those in the related art is
performed, and then the process proceeds to step S119.
[0034]
10-04-2019
10
In step S119, D / A conversion is performed by reading data from the output buffer 114 for a
block where there is no clipping, a block where there is no clip due to the phase operation, and a
block where the clip does not disappear even with phase operation. Then, the sound is
reproduced by the speaker 117.
[0035]
The routine R110 in FIG. 8 is a process to put the phase operation within a predetermined range
(for example, -90 ° to 90 °) as much as possible so as to place emphasis on sound quality, and
if adopted, it can be used as necessary. Good.
[0036]
In this routine R110, when NO (no clip) is determined in step S103, first, in step S111, the
rotational phase angle (set angle) of the variable phase shifter 120 is other than 0 ° during
processing of the previous block. It is determined whether or not it was.
When it is determined as NO (0 °) in step S111, the process proceeds to step S119 to perform
reproduction, but when it is determined as YES (other than 0 °), the process proceeds to step
S112.
In step S112, it is determined whether or not a change equal to or greater than a predetermined
angle (for example, 15 °) is required in order to return the rotational phase angle to 0 °. If it is
determined as YES in this step S112, the process proceeds to step S113, and after returning the
set angle to the 0 ° direction by the predetermined angle, the process proceeds to step S115,
and when it is determined as NO, the process proceeds to step S114. After setting the set angle
to 0 °, the process proceeds to step S115.
[0037]
In step S115, recalculation is performed such that the output data filtered by the variable phase
shifter 120 is written to the output buffer 114 at the set angle at which the phase is changed in
the 0 ° direction as described above. It is determined whether or not it has occurred. When it is
determined as NO (no clip) in step S115, the process proceeds to step S119, and when it is
determined as YES (with clip), the process proceeds to step S116. In step S116, the inside of the
10-04-2019
11
output buffer 114 is updated with the output data from the variable phase shifter 120 at the
initial rotational phase angle before changing the phase in step S113 or S114 to the 0 °
direction, and the process proceeds to step S119. . The output data at the above-mentioned initial
angle may be left in the output buffer 114 and used.
[0038]
Here, if a clip occurs, the steps S104, S105, and S106 may be repeated to avoid the clip, although
a large change in rotational phase angle may occur as compared to the previous block. If it does
not occur, the above-mentioned routine R110 is used to gradually return (for example, every 15
°) gradually to 0 ° when returning the changed rotational phase angle to 0 °. The effect of
reducing deterioration can be expected. In applications where the routine R110 is not required,
the process may proceed from step S103 to step S119. The set phase in this case is constantly
changing, for example, in the range of -180 ° to 0 ° to 180 °.
[0039]
In the output buffer 114, although the case where data for one block is mainly stored has been
described, data for a plurality of blocks can be stored, and block data for a plurality of blocks
when the set angle is changed is stored. Of course, the required block data may be retrieved
without recalculation. In addition, it is also preferable to be able to realize smooth phase change
(for example, due to fading processing etc.) by storing block data not only for the block currently
being processed but also for other blocks that differ in time. .
[0040]
FIG. 9 is a timing chart for explaining, for example, a case where data of three temporally
consecutive blocks are prefetched to smoothly perform phase change. In FIG. 9, the processing
block BL is processed and reproduced in the order of BL0, BL1, BL2, BL3,. At the moment, the
block BL0 is reproduced, and the three blocks BL1, BL2 and BL3 continuous to the block BL0 are
read ahead. Here, no clipping occurs in the blocks BL1 and BL2, and a clipping occurs in the
block BL3. In order to avoid this, a phase operation is required, and the weighting gains Ga of the
multipliers 126 and 123, respectively, It is assumed that Gb needs to be Ga3 and Gb3,
respectively. Here, when the weighting gains Ga1 and Gb1 in the block BL1 are rapidly switched
to the Ga3 and Gb3, if the phase difference between them is large, a sense of discomfort may
10-04-2019
12
occur when switching the blocks. Therefore, in the block BL2 between the blocks BL1 and BL3,
the weighting gains Ga and Gb are gradually changed from Ga1 and Gb1 to Ga3 and Gb3,
respectively, in order to gradually change the phase. As a result, when moving from the block
BL1 to the block BL3 via the block BL2, the rotational phase changes gradually, the sense of
incongruity in hearing is suppressed, and the sound quality can be improved.
[0041]
By the way, in the variable phase shifter 120 of FIG. 2 described above, the delay unit (delay
circuit 124) and the shift Hilbert filter 121 are separately provided, and these filter output
results are obtained as the optimum gains Ga of the multipliers 126 and 123. And Gb are used as
the final phase shifter output signal, but both the delay unit and the Hilbert unit can be realized
as the same FIR filter, and both can be added linearly as a system block. Therefore, the
configuration shown in FIG. 10 can also be used.
[0042]
The variable phase shifter 130 used in FIG. 10 corresponds to the variable phase shifter 113 in
FIG. 1 and the variable phase shifter 120 in FIG. The phase of the input signal (the output signal
from the filter 112) is changed and output.
The source 111, the filter 112, the output buffer 114, the control unit 115, the D / A converter
116, and the speaker 117 are the same as those shown in FIGS.
[0043]
Similar to the variable phase shifter 120 shown in FIG. 2, the variable phase shifter 130 adds a
shift type Hilbert filter output and a delay filter (delay circuit) output at an appropriate
component ratio, and thereby obtains a desired phase rotation angle. In order to realize these
filters, the coefficients of the FIR filter for realizing these filters are first combined in an
appropriate component ratio, and the resulting FIR coefficients are loaded (applied) to the FIR
filter 131. There is. That is, the coefficients for realizing the shift Hilbert filter with the FIR filter
are generated by the shift Hilbert filter coefficient generator 133 and sent to the multiplier 134,
and the coefficients for realizing the delay filter (delay circuit) with the FIR filter It is generated
by the delay coefficient generation unit 135 and sent to the multiplier 136. The weighting gains
ga and gb which are multiplication coefficients of the multiplication units 136 and 134 are
10-04-2019
13
variably controlled by the control signal from the control unit 115, and the weighted coefficients
from the multiplication units 136 and 134 are FIR coefficient generation units By sending them
to 137, FIR coefficients are created such that the weighting gains of the final delay output and
the shift Hilbert filter output become Ga and Gb as described above. By loading the FIR
coefficients from the FIR coefficient creation unit 137 into the FIR filter 131, the FIR filter 131
realizes input / output characteristics similar to those of the variable phase shifter 120 of FIG.
[0044]
Next, FIG. 11 is a block diagram showing an example in which a signal is divided into n frequency
bands and the phase is determined for each frequency band as another embodiment of the
present invention. In the example of FIG. 11, the configuration as described in conjunction with
FIG. 2 is used for each band, and addition is performed by selecting and outputting a
combination having the smallest amplitude for each band. The overall amplitude is also expected
to be small.
[0045]
In FIG. 11, an audio signal from a source (such as a sound source) 111 via a filter 112 such as an
equalizer is input to a variable phase shifter 150, and is divided into n frequency band signals by
a frequency division circuit 151. . The cutoff frequency fc at this time is f0, f1, f2, ..., fn-2, and fn1, respectively. The signals of the frequency bands divided into bands are respectively sent to the
variable phase shifters 120_0, 120_1,. These variable phase shifters 120_0, 120_1,...
Respectively have the same configuration as the variable phase shifter 120 of FIG. 2 described
above, and from the respective variable phase shifters 120_0, 120_1,. The outputs are sent to the
output buffers 114_0, 114_1,..., And the outputs from these output buffers 114_0, 114_1,... Are
sent to the general output buffer 114 ′ as the output from the variable phase shifter 150. Be
The general control unit 115 ′ determines whether or not the data stored in the general output
buffer 114 ′ is clipped, and the control signals are controlled by the control units 115_0 and
115_1 of the variable phase shifter 50 according to the determination result. , ... sent to. Each
control unit 115_0, 115_1,... Selects, for each band, a combination of weighting gains such that
the amplitude decreases. The variable phase shifters 120_0, 120_1,..., The output buffers 114_0,
114_1,..., The control units 115_0, 115_1,. The same as the output buffer 114 and the control
unit 115, the description will be omitted. The output from the total output buffer 14 ′ is
converted to an analog audio signal by the D / A converter 116 and reproduced as an acoustic
signal by the speaker 117.
10-04-2019
14
[0046]
As for the variable phase shifters 120_0, 120_1,... For each frequency band in FIG. 11, the FIR
coefficients are first obtained as in the variable phase shifter 130 in FIG. The following
configuration may be used.
[0047]
In the above embodiment, as the variable phase shifter 113 in FIG. 1 above, the desired rotational
phase angle can be obtained by weighting and adding the delay filter (delay circuit) output and
the shift type Hilbert filter output with an appropriate component ratio. Although what is realized
is used, a phase transition device may be used which can directly obtain a desired rotational
phase angle as an input-output characteristic.
This phase shift device is configured by a multistage all pass filter (APF) when configured by
analog. In the case of digital configuration, it is configured of a multistage cyclic (Infinite Impulse
Response: IIR) filter or a non-cyclic (Finite Impulse Response: FIR) filter with a sufficient number
of taps. Also, instead of the shift-type Hilbert filter, it is possible to use a phase shifting device
having a characteristic that the phase for each frequency f is π / 2 + nπ [rad] (n is an integer) as
the input / output characteristic. Good.
[0048]
FIG. 12 shows a phase shift device 13 in which 10 stages of IIR filters are connected as an
example of a phase shift device which can be used instead of the variable phase shifter 113 of
FIG. 1 or the shift Hilbert filter 121 of FIG. FIG. The input speech signal S0 (the output signal
from the filter 112 in FIG. 1) input through the input terminal 15 is filtered through 10 stages of
IIR filters 161, 162, 163... 169 and 1610. Each IIR filter is, for example, a second-order IIR
designed as an all-pass filter (APF). The amplitude (gain) characteristic is constant regardless of
the frequency, but the phase is characterized by shifting according to the frequency. Therefore,
by connecting IIR filters having different phase characteristics individually in multiple stages, for
example 10 stages, the phase can be shifted constantly over almost the entire frequency band,
and this phase shifted output is output from the output terminal 17 Can.
[0049]
10-04-2019
15
FIGS. 13 and 14 show, for example, the amplitude characteristics and the phase characteristics of
the IIR filter 161 and the IIR filter 162 of FIG. In the vertical axis, the left side is the amplitude
Gain [dB] on the left side, and the right side is the phase shift amount [rad] from 0 ° to π (pi).
[0050]
FIGS. 13 and 14 show the expressions of the transfer functions H (s) of the IIR filter 161 and the
IIR filter 162, respectively. In these examples, the coefficients of A3, A4, B3 and B4 are zero. That
is, as described above, this is an example of a second-order IIR filter.
[0051]
In each characteristic of the IIR filter 161 shown in FIG. 13, for example, the amplitude
characteristic Gain is constant at 0 [dB], and the phase characteristic phase changes depending
on the frequency. That is, although the amplitude is constant, it can be seen that the phase shift
amount is changing. The phase shift amount is π * (9/10) [rad] at 10 Hz, π * (6/10) [rad] at
100 Hz, π * (1/10) [rad] at 1000 Hz, 0 [rad] at 10000 Hz. ]. Also in each of the characteristics of
the IIR filter 162 shown in FIG. 14, for example, the amplitude characteristic Gain is constant at 0
[dB], and the phase characteristic phase changes with frequency. The phase shift amount is π
[rad] at 10 Hz, π * (9/10) [rad] at 100 Hz, π * (6/10) [rad] at 1000 Hz, and π * (1/10) at
10000 Hz. ]. Therefore, it can be seen that the phase characteristics of the IIR filter 161 and the
IIR filter 162 are different from each other.
[0052]
Of course, the APF by the IIR filter is not limited to the second order, and may be the first order
or the third or higher order, and by connecting in multiple stages, for example, phase
characteristics as shown in FIG. 13 and FIG. To design the desired phase characteristic.
[0053]
FIG. 15 shows, as shown in FIG. 12, a plurality of IIR filters having phase characteristics different
from each other with respect to the frequency, such as the IIR filter 161 and the IIR filter 162
10-04-2019
16
shown in FIG. 13 and FIG. 10 shows the delay characteristics of the phase shift device obtained
by connecting 10 stages.
It can be formed by superposing the phase characteristics of the respective IIR filters 163... 169
and 1610, including the phase characteristics of FIGS. The group delay time [t] shown on the
vertical axis of FIG. 15 changes monotonically with the frequency [Hz] on the horizontal axis. In
this case, the delay time decreases monotonically as the frequency increases. That is, the group
delay amount (time) changes according to the frequency. By using this characteristic, it is
possible to correct the phase difference between the input signal to the phase shift device 13 and
the phase shift output signal.
[0054]
In the case of using the phase transition device shown in FIG. 15 instead of the shift-type Hilbert
filter 121 of FIG. 2 described above, ideally, the phase for each frequency f is π / 2 + nπ [rad] It
becomes. n is an integer, and if it is desired to increase the delay amount, this may be a larger
integer value. However, as the delay effect, there is a delay effect with respect to the low band
component having a large group delay amount, and since the period per wavelength is short with
respect to the high band, the delay effect is small.
[0055]
In addition, even if the above n is not exactly an integer, and the phase difference between the
input signal and the phase shift output signal of the phase shift device 13 is slightly shifted to 90
°, such as ± 10 ° or ± 20 °, practically There may be no problem.
[0056]
FIG. 16 shows an example of the configuration of an nth-order IIR filter.
The input signal X from the input terminal 21 is multiplied by the coefficient c by the coefficient
multiplier 22 and then supplied to the adder 23. The adder 23 is also supplied with the addition
output from the adder 36 described later. The addition output T of the adder 23 is supplied to a
delay unit 26 for one clock cycle and an adder 24 described later. The delay unit 26 supplies an
output signal delayed by one clock period to the coefficient multiplier 27, the coefficient
10-04-2019
17
multiplier 28 and the delay unit 29. The coefficient multiplier 27 multiplies the delay output
signal of the delay unit 26 by the coefficient a 1 and supplies the product to the adder 36. The
adder 36 is also supplied with the addition output from the adder 35 described later. The
coefficient multiplier 28 multiplies the delay output signal of the delay unit 26 by the coefficient
b 1 and supplies the result to the adder 38. The adder 38 is also supplied with the addition
output from the adder 37 described later. The delay unit 29 further delays the delayed output
signal of the delay unit 26 by one clock cycle, and supplies the delayed output signal to the
coefficient multiplier 30, the coefficient multiplier 31, and the delay unit of the next stage. The
coefficient multiplier 30 multiplies the delay output signal of the delay unit 29 by the coefficient
a 2 and supplies the product to the adder 35. The adder 35 is also supplied with the addition
output from the adder. The coefficient multiplier 31 multiplies the delay output signal of the
delay unit 29 by the coefficient b 2 and supplies the product to the adder 37. The adder 37 is
also supplied with the addition output from the adder. Similarly, the delay unit 32 is supplied
with the delayed output signal of the previous stage delay unit. The delay unit 32 further
supplies a delayed output signal delayed by one clock period to the coefficient multiplier 33 and
the coefficient multiplier 34. The coefficient multiplier 33 multiplies the delay output signal of
the delay unit 32 by the coefficient an, and returns it to the previous stage adder. Further, the
coefficient multiplier 34 multiplies the delay output signal of the delay unit 32 by the coefficient
bn, and returns the multiplication result to the former stage adder. Then, the addition output
from the adder ···, the adder 35 and the adder 36 is returned to the adder 23. Also, the addition
output from the adder ···, the adder 37 and the adder 38 is returned to the adder 24. The adder
24 is connected to the output terminal 25, and an output signal Y is output from the output
terminal 25. The nth-order IIR filter is, of course, deeper by the order (n).
[0057]
FIG. 17 shows a configuration in which second-order IIR filters 411, 412... And 41m are
connected in multiple stages (m stages) based on the n-order IIR filter whose configuration is
shown in FIG. According to the multistage connection configuration of this second-order IIR filter,
it is possible to obtain group delay characteristics as shown in FIG. Further, the transfer function
may be partially fractionated to connect IIR filters in parallel.
[0058]
Here, the amount of phase shift can be changed by controlling each coefficient or the like of the
phase shift device corresponding to the variable phase shifter 113 by the control unit 115 of FIG.
1, and the phase shift amount (rotational phase angle) It can be controlled to any angle. Further,
10-04-2019
18
as described above, the phase shift amount (rotational phase angle) of the phase shift device is
fixed to 90 ° (or -90 °) or a value before or after it, and used instead of the Hilbert filter 121 of
FIG. It goes without saying that the rotational phase angle may be controlled by adjusting the
weighting gain between the output from the phase shift device and the output from the delay
circuit 124. It should be noted that any rotational phase angle other than 90 ° can also be
obtained by performing weighted addition in combination with 0 ° of the simple delay.
[0059]
In the embodiment of the present invention described above, although one audio signal
(monaural signal) is described to simplify the description, an output buffer corresponding to each
channel also in the case of stereo or multi-channel. The same effect can be obtained by
performing the phase operation by the same angle for all channels.
[0060]
In the above embodiment, digital signal processing is performed assuming digital signal input,
but the present invention is also applicable to analog signal processing performed on an analog
input audio signal. Of course.
[0061]
According to the embodiment of the present invention, even when there is a large output such
that it is difficult to avoid the clip, the effect of suppressing the amplitude of the entire waveform
makes it possible to use the conventional type even when combined with conventional clipping
processing means such as limiter processing. The degree of suffering distortion and deterioration
of sound quality can be reduced.
[0062]
The present invention is not limited to the above-described embodiment, and it goes without
saying that various modifications can be made without departing from the scope of the present
invention.
[0063]
FIG. 1 is a block diagram showing a schematic configuration of an audio signal processing
apparatus according to an embodiment of the present invention.
10-04-2019
19
It is a block diagram which shows the other example of the audio signal processing apparatus
which becomes embodiment of this invention.
It is a figure which shows the impulse response of the transfer function of Hilbert transform.
It is a figure which shows the impulse response of the delay circuit which performed cyclic shift.
It is a figure which shows the impulse response of the Hilbert transform filter which performed
cyclic shift. It is a figure which shows the impulse response of the Hilbert transform filter which
varied the sample number of cyclic shift. It is a figure for demonstrating weighting composition
with delay circuit output vector a and Hilbert filter output vector b. It is a flowchart for
demonstrating the specific example of the control action for phase angle determination in the
control part of an audio signal processing apparatus. FIG. 10 is a timing chart for explaining a
case where data of three blocks continuous in time is read ahead and phase change is smoothly
performed. FIG. FIG. 6 is a block diagram showing an example of using an FIR filter in which filter
coefficients are loaded in a variable phase shifter of an audio signal processing apparatus
according to an embodiment of the present invention. FIG. 16 is a block diagram showing an
example of dividing an input signal into n frequency bands and determining a phase for each
frequency band as still another embodiment of the present invention. FIG. 7 is a diagram showing
a phase shift device 13 in which 10 stages of IIR filters are connected as an example of the phase
shift device used for the variable phase shifter 113 of FIG. 1. FIG. 6 is a diagram showing an
amplitude characteristic and a phase characteristic of the IIR filter 161. FIG. 6 is a diagram
showing an amplitude characteristic and a phase characteristic of the IIR filter 162. It is a figure
which shows the delay characteristic of a phase transition apparatus. It is a figure which shows
the structural example of the IIR filter of the n-th. It is a figure which shows the structure which
connected the multi-order IIR filter 411, 412 ... and 41 m in multiple stages (m level).
Explanation of sign
[0064]
13 phase transition device, 113, 120, 130, 150 variable phase shifter, 114 output buffer, 115
control unit, 121 shift type Hilbert filter, 123, 126, 134, 136 multiplier, 124 delay circuit, 131
FIR filter, 133 Shifted Hilbert filter coefficient generator, 135 delay coefficient generator, 137
FIR coefficient generator
10-04-2019
20
10-04-2019
21
Документ
Категория
Без категории
Просмотров
0
Размер файла
36 Кб
Теги
description, jp2007067578
1/--страниц
Пожаловаться на содержимое документа