close

Вход

Забыли?

вход по аккаунту

?

DESCRIPTION JP2010199689

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 JP2010199689
PROBLEM TO BE SOLVED: To calculate a filter coefficient sequence of an FIR filter 41 capable of
creating a desired reverberation effect without causing problems such as coloration and howling.
A CPU (42) generates a first index value f (n = 1, 1) indicating a reciprocal of a time difference Δt
between peaks in an impulse response waveform g (t) of an acoustic space 11 and a peak
position of an amplitude characteristic of the acoustic space 11. 2 ...). Furthermore, in each
critical band F (k = 1, 2...), The target number nadm (k of the second index value f ′) is adjusted
to the number napp (k = 1, 2...) Of the first index value f. = 1, 2 ...) are calculated, and the second
index values f '(n = 1, 2 ...) as many as the target number nadm (k = 1, 2 ...) are generated. Then,
the tap position Tap (i = 1, 2...) Of the FIR filter 41 is calculated based on the reciprocal of the
second index value f ′ (n = 1, 2...). [Selected figure] Figure 1
Filter coefficient calculation method, sound field support device and program
[0001]
The present invention relates to a technology for controlling the acoustic effect of acoustic space.
[0002]
There is a sound field support system based on the existing acoustic characteristics in the
acoustic space, enhancing and correcting the reflected sound characteristics including the
reverberation effect and the initial reflection sound in the acoustic space to control the acoustic
characteristics.
15-04-2019
1
This sound field support system is configured of a microphone and a speaker fixed to a ceiling or
a side wall of an acoustic space, and a sound field support device connected to them. In the
sound field support apparatus of this kind of sound field support system, FIR (Finite Impulse
Response: finite impulse) is applied to a filter coefficient sequence for applying a reverberation
effect or the like in a desired acoustic space to a collected sound signal input from a microphone.
A response waveform is convoluted by a filter, and the resulting signal is emitted from the
speaker. The sound emitted from the speaker returns to the microphone again via a plurality of
reflection paths in the acoustic space, and the speaker and the microphone constitute an acoustic
feedback system. According to this sound field support apparatus, by adjusting the filter
coefficient sequence to be convoluted to the sound pickup signal from the microphone according
to the desired acoustic space, reverberation as if performance is performed in the desired
acoustic space It can create effects. A technology related to the sound field support device of this
kind of sound field support system is disclosed, for example, in Patent Document 1.
[0003]
Unexamined-Japanese-Patent No. 07-240993 gazette
[0004]
Heinrich Kuttorf, Room acoustics-The sound of architecture and its theory-, Ichigaya Publishing
Co., August 8, 2003, p. 200-203
[0005]
By the way, in the sound field support system described above, the FIR filter is inserted in a
closed loop including a microphone and a speaker.
The gain of this closed loop depends on the product of the gain of the acoustic feedback system
from the speaker to the microphone and the gain of the FIR filter.
Therefore, when the band in which the gain peaks in the acoustic feedback system overlaps with
the band in which the gain peaks in the FIR filter, auditory problems such as howling and
coloration may be caused.
15-04-2019
2
[0006]
The present invention has been devised under such a background, and provides a technical
means that makes it possible to create a desired reverberation effect without causing auditory
problems such as howling and coloration. The purpose is to
[0007]
The present invention is a filter coefficient sequence of a filter which is inserted in a closed loop
including a speaker and a microphone provided in the same acoustic space and convolutes an
input signal with a filter coefficient sequence consisting of zero or nonzero filter coefficients. An
impulse response waveform acquiring process of acquiring an impulse response waveform in a
section from the microphone to the speaker in the closed loop, each peak appearing in the
impulse response waveform, and a next of each peak In the index value acquisition process of
acquiring the reciprocal of each time difference from the other peaks appearing in the table as
the first index value, dividing the frequency axis into a plurality of bands, and for each band, the
first index belonging to the band An aggregation process of aggregating the number of values,
and a smaller value as the aggregation value of the first index value is larger for each of the
plurality of bands Target number calculation process for calculating a target number having a
relationship such as to become smaller, and an index for generating, for each band of the
plurality of bands, a second index value equal in number to the target number belonging to the
band The time difference which is the reciprocal of the second index value generated in the value
generation process and the index value generation process is calculated respectively, and the
operation of accumulating each time difference in order from the largest one is repeated, and
each obtained sequentially by this repetition And a tap position calculating step of determining
tap positions of non-zero filter coefficients in the filter coefficient string based on an accumulated
value.
[0008]
In the present invention, the reciprocal of the time difference between the peaks in the impulse
response waveform in the acoustic space is calculated as the first index value.
The first index value indicates the frequency at which the gain peaks in the amplitude
characteristic of the frequency response of the section from the speaker to the microphone.
Then, in the above filter coefficient string calculation method, in each of the plurality of bands
15-04-2019
3
obtained by dividing the frequency axis, the number of second index values is decreased at each
portion where the number of first index values is large. Generate an index value of 2. The second
index value indicates the frequency at which the gain peaks in the amplitude characteristic of the
frequency response of the target filter. Then, the operation of accumulating the time difference,
which is the reciprocal of the second index value, in order from the largest one is repeated, and
the sequentially obtained accumulated value is used as the index value indicating the position,
and based on this index value Calculate the tap position of zero filter coefficient. When the tap
position of the nonzero coefficient in the filter coefficient sequence is calculated as described
above, the gain has a peak at each frequency indicated by the second index value in the
amplitude characteristic of the filter. And in the present invention, the number of second index
values is reduced in a band where the number of first index values is large, so the probability that
the second index value overlaps with the first index value is small. Become. Therefore, according
to the present invention, the overlap between the band where the gain peaks in the amplitude
characteristics of the section from the speaker to the microphone and the band where the gain
peaks in the filter amplitude characteristics is reduced, and coloration and howling occur. It can
be prevented.
[0009]
BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the whole structure of the
sound field assistance system containing the sound field assistance apparatus which is one
Embodiment of this invention. It is a flowchart which shows the process which CPU of the sound
field assistance apparatus performs. It is a figure which shows distribution of the appearance
frequency for every critical zone of the index value in acoustic space.
[0010]
Hereinafter, embodiments of the present invention will be described with reference to the
drawings. FIG. 1 is a diagram showing an entire configuration of a sound field support system
including a sound field support device 40 according to an embodiment of the present invention.
The microphone 10 and the speaker 20 in the sound field support system are fixed to the side
wall and the ceiling of the acoustic space 1 with a space. In this sound field support system, the
microphones 10 are connected to the sound field support apparatus 40 via the amplifier unit 31
and the speakers 20 via the power amplifier unit 32, and the acoustic space 1 → microphone 10
→ amplifier unit 31 → sound field A loop in which a path by an electric circuit from the
microphone 10 to the speaker 20 and an acoustic path from the speaker 20 to the microphone
10 are connected into one as a supporting device 40 → power amplifier unit 32 → speaker 20 →
15-04-2019
4
acoustic space 1 (simply Is formed.
[0011]
In the sound field support device 40, an analog signal input from the microphone 10, which has
collected the sound generated in the acoustic space 1 via the amplifier unit 31, is converted into
a digital format by an A / D converter (not shown). And is input to the FIR filter 41. The FIR filter
41 convolutes a filter coefficient series composed of zero or nonzero filter coefficients with an
input signal input to the FIR filter 41, and outputs the result as a reverberation sound signal. The
reverberation sound signal which is the output signal of the FIR filter 41 is converted to an
analog format by a D / A converter (not shown), and then input to the speaker 20 via the power
amplifier unit 32. Will be returned to The filter coefficient sequence of the FIR filter 41 is
determined based on the impulse response waveform of the acoustic space 1 and the desired
reverberation characteristic to be realized. The impulse sound source 45 is a sound source that
generates an impulse sound signal, and is used when measuring the impulse response of the
acoustic space 1.
[0012]
The CPU 42 is a control center of the sound field support device 40. The CPU 42 calculates the
filter coefficients stored in the ROM 44 while using the RAM 43 as a work area when a command
to set the filter coefficient sequence is given to the FIR filter 41 via the operation unit (not
shown). Run the program The filter coefficient calculation program includes tap positions Tapi (i
= 1, 2...) Of non-zero filter coefficients in a filter coefficient sequence of the FIR filter 41 and filter
coefficient values hi (i = 1, 2. It is a program which makes CPU42 perform processing which
computes i = 1, 2 ...).
[0013]
In the present embodiment, tap positions Tapi (i = 1, 2...) Of non-zero filter coefficients in the
filter coefficient string of the FIR filter 41 and filter coefficient values hi (i at each tap position
Tapi (i = 1, 2...) The size of (= 1, 2...) Is determined to satisfy the following condition. a. In the
amplitude characteristic of the frequency response of the FIR filter 41, the frequency at which
the gain peaks is the peak in the amplitude characteristic of the frequency response of the
impulse response of the acoustic space 1 (more specifically, the path from the speaker 20 to the
15-04-2019
5
microphone 10). Not overlap as much as possible. b. The envelope of the filter coefficient
sequence that the FIR filter 41 convolves corresponds to the desired reverberation characteristic.
The specific method of calculating the filter coefficient string will be clarified in the description
of the operation of the present embodiment in order to avoid the repetition of the description.
[0014]
Next, the operation of this embodiment will be described. FIG. 2 is a flowchart showing the
contents of processing that the filter coefficient calculation program causes the CPU 42 to
execute in the present embodiment. In FIG. 2, after switching the switch 46 to the OFF state
(S100), the CPU 42 causes the impulse sound source 45 to generate an impulse sound signal
(S110). The impulse sound signal generated by the impulse sound source 45 is emitted as an
impulse sound from the speaker 20 to the acoustic space 1 via the adder 47 and the power
amplifier unit 32. The impulse sound reaches the microphone 10 via a plurality of propagation
paths in the acoustic space 1. A signal indicating a sound collected by the microphone 10
(referred to as a “response signal”) is supplied from the microphone 10 to the CPU 42 of the
sound field support device 40 via the amplifier unit 31.
[0015]
The CPU 42 acquires the impulse response waveform g (t) of the acoustic space 1 from the
response signal obtained during a predetermined time from the generation time of the impulse
sound signal (S120). Next, the CPU 42 executes an index value acquisition process (S130). In this
index value acquisition process (S130), the CPU 42 detects each peak Pn (n = 1, 2...) Appearing in
the impulse response waveform g (t), and detects each peak Pn and the next appearing peak Pn +
1. Each time difference Δtn is obtained, and the reciprocal of the obtained time difference Δtn is
taken as a first index value fn. For example, when N peaks P1, P2... PN appear in the impulse
response waveform g (t), a time difference Δt1 between the first peak P1 and the second peak
P2 and a second peak P2 and the third peak P2 The time difference Δt N-1 between the time
difference Δt2 to P3 and the N-1st peak PN-1 and the Nth peak PN is calculated in order, and
their N-1 time differences Δt1, Δt2 ... ΔtN-1 Let each reciprocal be an index value f n (n = 1, 2...
N−1).
[0016]
Here, in the technical field of performing statistical analysis of the reverberation sound field, the
first index value fn (n = 1, 2...) Obtained in this way has a gain peak in the amplitude
15-04-2019
6
characteristic of the acoustic space 1 It is often said that the value is close to the frequency (for
details, refer to Non-Patent Document 1).
[0017]
Next, the CPU 42 executes an aggregation process (S140).
In this counting process (S140), the CPU 42 divides the frequency axis into critical bands Fk (k =
1, 2...), And the first index value fn (k = 1, 2...) The number nappk (k = 1, 2...) of n = 1, 2. Here, the
critical band is a band in which the amount of masking does not change even if the frequency
width is expanded, and is a band having a center frequency and a bandwidth as shown in Table 1,
for example.
[0018]
Next, the CPU 42 executes a target number calculation step (S150). As shown in FIG. 3, the total
number nappk of the first index values fn in the acoustic space 1 for each critical band Fk is low
in a certain band (in the example of FIG. 3, band F8: hereinafter referred to as “saturation
band”). On the band side, it increases while repeating large increases and decreases from the
low band to the high band, and on the high band side of the saturated band, it increases almost
linearly with a relatively gentle gradient d. In this target number calculation step (S150), the CPU
42 extrapolates the auxiliary line LINE-d of the gradient d from the saturation band to the lower
side thereof to set the central frequency of each of the critical bands Fk (k = 1, 2...) Let the
number on the corresponding auxiliary line LINE-d be the limit number nlimk (k = 1, 2...).
Furthermore, the CPU 42 sets the difference between the limit number nlimk (k = 1, 2...) And the
total number nappk (k = 1, 2...) Of the first index values fn obtained in the counting process
(S140) The target number nadmk (k = 1, 2...) Of the index value f′n of The second index value f
′ n is a value that indicates the frequency at which the gain peaks in the amplitude
characteristic of the frequency response of the FIR filter 41.
[0019]
Next, the CPU 42 executes an index value generation process (S160). In this index value
generation process (S160), the CPU 42 generates the second index value f'n as many as the
15-04-2019
7
target number nadmk for each of the critical bands Fk (k = 1, 2...) (S160). More specifically, the
CPU 42 first generates pseudo random numbers equal in number to the target number nadm1
determined for the critical band F1, and uses these pseudo random numbers to generate the
same number in the critical band F1 as the target number nadm1. A frequency is determined,
and each of the determined frequencies is set as a second index value f'n. Similarly, the CPU 42
generates second index values f'n of the same number as the target numbers nadm2, nadm3...
Also for the critical bands F2, F3,.
[0020]
Next, the CPU 42 executes a tap position calculation process (S170). In this tap position
calculation process (S170), the second index values f'n (n = 1, 2...) Generated in the index value
generation process (S160) and the envelope waveform c (t) prepared in advance are It becomes
an object of processing. Here, the envelope waveform c (t) is an envelope waveform prepared in
advance in accordance with an exponential decay curve having a desired reverberation
characteristic, for example, an impulse response waveform measured in an acoustic space having
a desired reverberation characteristic. It is an envelope waveform. In the tap position calculation
process (S170), first, time differences Δt'n (n = 1, 2...) Which are inverse numbers of the second
index values f'n (n = 1, 2...) Are calculated. Next, the largest time difference Δt′1 is selected
from the time differences Δt′n (n = 1, 2...), And the reference time (t = 0) of the time axis in the
envelope waveform c (t) corresponding to the desired reverberation characteristic ) Is set as the
first tap position Tap1 and the amplitude of the time t1 in the envelope waveform c (t) is set as
the filter coefficient value h1. Next, the CPU 42 selects the second largest time difference Δt2
out of the time differences Δt'n (n = 1, 2...) And accumulates the time difference Δt'1 to the time
difference Δt'1 selected up to that point. Time t'2 delayed by its accumulated value .DELTA.t'1 +
.DELTA.t'2 from the reference time (t = 0) of the second time as the second tap position Tap2,
and the amplitude of that time t'2 in the envelope waveform c (t) is filtered It is assumed that the
coefficient value h2. Thereafter, CPU 42 similarly repeats the operation of selecting and
accumulating the remaining time differences Δt 'n (n ≠ 1, 2) in order from the largest one, and
accumulating the time differences Δt' n sequentially obtained by this repetition. Based on the
values, third and subsequent tap positions Tap3, Tap4... TapN and filter coefficient values h3, h4.
[0021]
As described above, when the filter coefficient sequence of the FIR filter 41 is determined, the
FIR filter 41 has an amplitude characteristic in which the gain reaches a peak at the frequency
corresponding to the second index value. Therefore, overlapping of a band in which the gain
15-04-2019
8
peaks in the amplitude characteristic of the frequency response of the FIR filter 41 and a band in
which the gain peak in the frequency characteristic of the acoustic space 1 is avoided. Also, the
filter coefficient series of the FIR filter 41 has an envelope corresponding to the desired
reverberation characteristic. Therefore, when a performance or the like is performed in the
acoustic space 1, it is possible to realize desired reverberation characteristics while preventing
the occurrence of coloration and howling. Temporarily, a filter coefficient sequence obtained by
sampling an impulse response waveform sampled not from the filter coefficient sequence
obtained in this embodiment but from an acoustic space having a desired reverberation
characteristic (for example, the acoustic space A) is set as the FIR filter 41. In this case, if the
frequency at which the gain peaks in the amplitude characteristics of the acoustic space A and
the frequency at which the gain peaks in the acoustic space 1 overlap, coloration or howling may
occur at that frequency. However, according to the present embodiment, since overlapping of the
band in which the gain peaks in the amplitude characteristic of the FIR filter 41 and the band in
which the gain peaks in the acoustic space 1 is avoided, coloration and howling can be
prevented. it can.
[0022]
As mentioned above, although one embodiment of this invention was described, there may be
other embodiments in this invention. For example, it is as follows. (1) In the above embodiment,
the frequency axis is divided into critical bands Fk (k = 1, 2...), And the first index value fn (n = 1)
is obtained for each of the critical bands Fk (k = 1, 2. , 2... And the target numbers nadmk (k = 1,
2...) Of the second index values fn ′ (n = 1, 2...). However, the frequency axis is divided into a
band narrower or wider than the critical band, and the number of first index values fn (n = 1, 2,...)
And the second index value fn ′ (n = 1, n) for each band. The target number of 2 ...) may be
determined. (2) In the above embodiment, the sound field support device 40 has hardware such
as an application specific integrated circuit (ASIC) instead of the CPU 42, the RAM 43, and the
ROM 44, and this hardware performs the function of the filter coefficient calculation program.
The same process as the process by may be executed.
[0023]
DESCRIPTION OF SYMBOLS 1 ... acoustic space, 10 ... microphone, 20 ... speaker, 31 ... amplifier
part, 32 ... power amplifier part, 40 ... sound field assistance apparatus, 41 ... FIR filter, 42 ... CPU,
43 ... RAM, 44 ... ROM, 45 ... Impulse sound source, 46 switch, 47 adder.
15-04-2019
9
Документ
Категория
Без категории
Просмотров
0
Размер файла
20 Кб
Теги
description, jp2010199689
1/--страниц
Пожаловаться на содержимое документа