close

Вход

Забыли?

вход по аккаунту

?

JP2002135898

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 JP2002135898
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
sound image localization control headphone for realizing an outside-head localization of an audio
signal.
[0002]
2. Description of the Related Art Conventionally, in audio reproduction using headphones,
localization in the head of the sound image has been regarded as a problem, and various methods
for solving the problem have been proposed.
[0003]
Basically, using a dummy head or the like, the filter characteristics for sound image localization
control are determined from the head transfer function from an actual sound source such as a
speaker to the dummy head and the transfer function of headphones mounted on the dummy
head. It is common to process an audio signal as a filter coefficient.
[0004]
Hereinafter, conventional sound image localization control headphones will be described with
reference to the drawings.
[0005]
10-05-2019
1
FIG. 20 is a block diagram showing a conventional sound image localization control headphone.
[0006]
In FIG. 20, 1a to 1b are FIR filters for performing sound image localization control of input
signals, 33a to 33b are memories for setting coefficients in the FIR filters 1a to 1b, and 8 is a
sound image localization controlled input signal through an amplifier not shown. It is
headphones to play.
[0007]
The operation of the sound image localization control headphone shown in FIG. 20 will be
described below.
[0008]
The input signal from the sound source is subjected to convolution processing with the
coefficients by the FIR filters 1 a to 1 b and input to the headphone 8.
The coefficients of the FIR filters 1a to 1b are stored in the memories 33a to 33b, and the
coefficients are previously set in the FIR filters 1a to 1b before performing convolution with the
input signal.
[0009]
Next, how to obtain this coefficient will be described.
[0010]
First, a head-related transfer function as a target of sound image localization is measured.
In FIG. 21, the measurement signal from the measurement signal generator 17 is reproduced
from the speaker 18 which is an actual sound source.
10-05-2019
2
The reproduced sound is detected by the microphones 9 c to 9 d installed in the ear holes of the
dummy head 19 and input to the transfer function measuring device 34.
Since the measurement signal from the measurement signal generator 17 is also input to the
transfer function measuring instrument 34, the transfer function from the speaker 18 to the
microphones 9c to 9d (if performed in the anechoic chamber) by the measurement signal and the
microphone detection signal This becomes the head related transfer function).
FIG. 21 shows the case where the speaker 18 is installed in front of the dummy head 19. The
transfer function from the speaker 18 to the microphone 9c is Hr, and the transfer function from
the speaker 18 to the microphone 9d is Hl.
[0011]
Next, the acoustic characteristics of the headphones 8 are measured.
In FIG. 22, the headphone 8 is properly mounted on the dummy head 19, and the measurement
signal from the measurement signal generator 17 is reproduced from the headphone 8.
The reproduced sound is detected by the microphones 9 a to 9 b installed in the ear holes of the
dummy head 19 and is input to the transfer function measuring device 34. Since the
measurement signal from the measurement signal generator 17 is also input to the transfer
function measuring instrument 34, the transfer function from the headphone 8 to the
microphones 9a to 9b is measured by the measurement signal and the microphone detection
signal. In FIG. 22, the transfer function from the right speaker 8 a of the headphone 8 to the
microphone 9 a is Cr, and the transfer function from the left speaker 8 b of the headphone 8 to
the microphone 9 b is Cl.
[0012]
Each transfer function measured in FIG. 21 and FIG. 22 is put on a computer such as a work
station to obtain the FIR filter coefficients in FIG. Assuming that the transfer function of the FIR
filter 1a is Xr and the transfer function of the FIR filter 1b is Xl, it suffices to reproduce Hr and Hl
10-05-2019
3
shown in FIG. 21 at the ear of the listener 10,
[0014]
The FIR filters 1a to 1b have transfer functions
[0016]
It is sufficient to find out the coefficient that becomes.
[0017]
If the coefficients obtained in this manner are stored in the memories 33a to 33b of FIG. 20, the
coefficients are set in the FIR filters 1a to 1b as necessary, and sound image localization is
performed by performing convolution processing with the input signal. Control is performed.
[0018]
Although the center channel signal has been described above as an example, the concept is the
same whether it is a two-channel signal of Lch signal and Rch signal like a stereo signal or a
multi-channel signal of more than that.
[0019]
However, in the conventional sound image localization control headphone shown in FIG. 20,
since the measured transfer functions Hr, Hl, Cr, and Cl differ from the actual listener 10, the
sound image localization effect becomes worse. .
This is because the measured transfer functions Hr, Hl, Cr, and Cl contain a lot of information
specific to individuals, so individual differences are large during actual sound image localization
control in FIG. 20 (in particular, a dummy head is used in this conventional example). Because the
difference with the actual human being is large), the effect is deteriorated.
Since each transfer function at the time of sound image localization control by the actual listener
is different from the transfer function at the time of measurement, the transfer functions of the
FIR filters 1a to 1b are defined as Hr ', Hl', Cr 'and Cl'.
10-05-2019
4
[0021]
Thus, it differs greatly from the original (Equation 2).
[0022]
In order to solve this problem, it has also been proposed to measure a large number of human
head transfer functions and classify them into several types of groups, and use a plurality of
transfer functions corresponding to the groups.
However, in this case, it is necessary to hold many transfer characteristics, that is, a large number
of coefficients of the FIR filters 1a to 1b, resulting in a new problem that the memory capacity is
increased.
[0023]
By the way, even if the transfer functions Hr, Hl, Cr, Cl which are small in individual difference
and excellent in the sound image localization effect are measured, the listener 10 does not always
wear the headphones 8 in the same manner without any gap.
In other words, individual differences among the transfer functions Hr, Hl, Cr, and Cl are also a
problem, but even if they are the same person who actually measured the transfer functions Hr,
Hl, Cr, and Cl, the transfer function Cr depends on the way of wearing the headphones 8. , Cl will
be different.
That is, depending on the wearing condition of the headphones 8, the transfer function of the FIR
filters 1a to 1b is
[0025]
This is also different from (Equation 2). For this reason, even if it is the same person who actually
measured the transfer functions Hr, Hl, Cr, and Cl, it is difficult to always keep the sound image
10-05-2019
5
localization effect at the maximum.
[0026]
In addition, while performing sound image localization control of the transfer functions Cr and Cl
and the FIR filter coefficients Xr and Xl, it is not possible to optimally measure and calculate for
each listener.
[0027]
In view of the above problems, the present invention can measure the transfer functions Cr and
Cl of the headphone and the ear for each listener or for each headphone wearing of the listener,
and calculate the filter coefficients Xr and Xl for sound image localization control as well. It is an
object of the present invention to provide a sound image localization control headphone that can
Moreover, it is possible to execute the measurement of the transfer functions Cr and Cl and the
calculation of the filter coefficients Xr and Xl while performing sound image localization control.
[0028]
As a result, the sound image localization control effect is not influenced by the individual
difference or the headphone wearing state, and a good effect can be obtained.
[0029]
A sound image localization control headphone according to claim 1 comprises a digital filter for
processing an input signal to perform sound image localization control, a target characteristic
filter for processing the input signal, and the input signal. A transfer function corrector that
processes a signal, a headphone that reproduces the output of the digital filter, a microphone
attached to the headphone, and a subtractor that subtracts the output of the target characteristic
filter from the output of the microphone; It is characterized by comprising a coefficient updater
which updates the coefficient of the digital filter by the output from the transfer function
corrector and the output from the subtractor.
[0030]
Preferably, the digital filter is characterized in that the coefficient is set from a memory storing
an initial coefficient initialized in advance or an update coefficient updated by the coefficient
10-05-2019
6
updater.
[0031]
Preferably, the digital filter is characterized by storing the update coefficient updated by the
coefficient updater in a memory.
[0032]
Preferably, the target characteristic filter is characterized in that a transfer function from a sound
source, which is a sound image localization target, to the ear of the listener is approximated as a
coefficient.
[0033]
Preferably, the target characteristic filter is characterized in that the coefficient is set from a
memory storing a coefficient approximating the transfer function from the sound source which is
the sound image localization target to the ear of the listener.
[0034]
Preferably, the transfer function corrector is characterized by approximating a transfer function
between the headphone and the microphone as a coefficient.
[0035]
Preferably, the transfer function corrector is characterized in that the coefficient is set from a
memory storing a coefficient approximating the transfer function between the headphone and
the microphone.
[0036]
Preferably, the memory stores a plurality of initial coefficients or update coefficients of the
digital filter.
[0037]
Preferably, the memory stores a plurality of target characteristic coefficients of the target
characteristic filter.
10-05-2019
7
[0038]
Preferably, the memory stores a plurality of coefficients of the transfer function corrector.
[0039]
Preferably, the coefficient updater updates the coefficients of the digital filter so as to minimize
the output signal from the subtractor.
[0040]
Preferably, the memory is characterized in that the initial coefficient or the update coefficient of
the digital filter can be downloaded from a storage medium or a personal computer.
[0041]
Preferably, the memory is downloadable from the storage medium or a personal computer from
the target characteristic coefficient of the target characteristic filter.
[0042]
Preferably, the memory is characterized in that the coefficients of the transfer function corrector
can be downloaded from a storage medium or a personal computer.
[0043]
The sound image localization control headphone according to claim 2 reproduces an output of
the digital filter, a digital filter that performs sound image localization control by processing an
input signal, an identification digital filter that performs signal processing of an output from the
digital filter, and Headphones, a microphone attached to the headphones, an identification
subtractor for subtracting the output of the identification digital filter from the output of the
microphone, and the output of the digital filter and the output of the identification subtractor for
identification The identification coefficient updater identifies the transfer function between the
headphone and the microphone as a coefficient while the digital filter performs sound image
localization control, and the identification coefficient updater updates the coefficients of the
digital filter for identification. It is characterized by
[0044]
Preferably, the digital filter is characterized in that the coefficient is set from a memory storing
an initial coefficient initialized in advance or an update coefficient updated by the coefficient
updater.
10-05-2019
8
[0045]
Preferably, the coefficient identified by the identification coefficient updater is stored in a
memory.
[0046]
Preferably, the identification coefficient updater updates the coefficients of the identification
digital filter so as to minimize the output signal from the identification subtractor.
[0047]
Preferably, the memory stores a plurality of initial coefficients or update coefficients of the
digital filter.
[0048]
Preferably, the memory is characterized in that the initial coefficient or the update coefficient of
the digital filter can be downloaded from a storage medium or a personal computer.
[0049]
BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be
described below with reference to FIGS.
[0050]
(First Embodiment) FIG. 1 shows a block diagram of a sound image localization control
headphone in the first embodiment.
[0051]
In FIG. 1, 1a to 1b are FIR filters for performing sound image localization control of audio input
signals, 2a to 2b target characteristic filters, 3a to 3b Fx filters as transfer function correctors,
and 4a to 4b coefficient updaters LMS arithmetic unit, 5a-5b is a subtractor, 6a-6f is a memory
RAM, 7 is a coefficient setting write control circuit for controlling reading and writing of
coefficient data of RAM6a-6f, 8 is a sound image by FIR filters 1a-1b Headphones for
reproducing localization-controlled audio signals through an amplifier (not shown), 8a to 8b are
speakers attached to the headphone 8, 9a to 9b are microphones attached to the headphone 8,
10 is a listener wearing the headphone 8 It is.
10-05-2019
9
[0052]
FIG. 1 realizes the sound image localization of the center sound source similarly to FIG. 20
described in the conventional example.
[0053]
First, the initial coefficients stored in the RAMs 6a to 6b are set in the FIR filters 1a to 1b by the
control signal from the coefficient setting write control circuit 7.
The input audio center signal is convoluted with the coefficients by the FIR filters 1 a to 1 b and
input to the headphone 8.
The initial coefficients set in the FIR filters 1a to 1b are obtained by calculating in advance
coefficients having characteristics for realizing sound image localization control. Therefore, when
the listener 10 listens to the reproduced sound from the headphones 8, the sound It can be felt
that it is reproduced from the center speaker like 21.
[0054]
By the way, in the conventional sound image localization control headphones, due to the
individual difference, the wearing condition of the headphones, or the characteristic variation of
the speaker unit of the headphones, it is not always possible to always listen in the optimum
condition. However, in the sound image localization control headphones of the present invention,
it is an object to control so as to always be able to listen in an optimal state, which will be
described below.
[0055]
The reproduced sound from the headphones 8 is detected by the microphones 9a to 9b and
input to the subtractors 5a to 5b.
On the other hand, the RAMs 6c to 6f set the stored target characteristic coefficients and Fx
10-05-2019
10
coefficients in the target characteristic filters 2a to 2b and the Fx filters 3a to 3b, respectively, by
control signals from the coefficient setting write control circuit 7.
The input audio center signal is convoluted with the coefficients by the Fx filters 3a to 3b and
input to the LMS computing units 4a to 4b, and the target characteristic filters 2a to 2b are also
convoluted to the subtractor 5a. Is input to .about.5b.
In the subtractors 5a to 5b, the outputs of the target characteristic filters 2a to 2b are subtracted
from the outputs of the microphones 9a to 9b, and the subtraction results are input to the LMS
arithmetic units 4a to 4b.
The LMS computing units 4a to 4b use the output signals from the Fx filters 3a to 3b as
reference signals, and the output signals from the subtractors 5a to 5b as error signals, using the
least squares method. The coefficients of the FIR filters 1a to 1b are updated so as to minimize.
Here, the transfer function Cr between the right speaker 8a of the headphone 8 to the
microphone 9a in the Fx filter 3a and the transfer function Cl between the left speaker 8b of the
headphone 8 to the microphone 9b in the Fx filter 3b are approximated as coefficients. ing.
This method is called Filtered-x LMS method (for example, B. Widrow and S. Stearns, "Adaptive
Signal Processing" (Prentice-Hall, Englewood Cliffs, NJ, 1985)).
As a result, the FIR filters 1a to 1b update their coefficients so as to realize the target
characteristics indicated by the target characteristic filters 2a to 2b at the microphones 9a to 9b,
that is, the ears of the listener 10.
That is, assuming that the transfer function of the FIR filter 1a is Xr, the transfer function of the
FIR filter 1b is Xl, the transfer function of the target characteristic filter 2a is Hr, and the transfer
function of the target characteristic filter 2b is Hl, the transfer functions of the FIR filters 1a to
1b Is
[0057]
【0057】となる。
10-05-2019
11
[0058]
If the transfer functions of the headphones (speakers 8a to 8b) and the ears (microphones 9a to
9b) when the listener 10 actually wears the headphones 8 are the same as the coefficients that
the RAMs 6e to 6f set in the FIR filters 3a to 3b. For example, assuming that the audio input
signal is x, the reproduced sound yr, yl at the listener 10 is
[0060]
Thus, the target playback sound can be heard and listened.
[0061]
By the way, as described in the conventional example, it is assumed that the transfer functions Cr
and Cl of the headphones and the ear become Cr 'and Cl' due to the individual difference, the
headphone wearing state, or the characteristic variation of the speaker unit of the headphones.
In this case, transfer functions Cr 'and Cl' of the headphones and the ear are newly measured.
[0062]
FIG. 2 shows a block diagram of a sound image localization control headphone in the case of
measuring the transfer function of the headphones and the ear (that is, the coefficients of the Fx
filters 3a to 3b).
[0063]
In FIG. 2, 1a to 1b are FIR filters for controlling sound image localization of audio input signals,
6a, 6b, 6e and 6f are memories RAM, and 7 are reading and writing coefficient data of RAMs 6a,
6b, 6e and 6f Coefficient setting write control circuit to control, 8 is a headphone to reproduce
an audio signal whose sound image localization is controlled by FIR filters 1a to 1b through an
amplifier not shown, 8a to 8b are speakers attached to headphone 8, 9a to 9b are Microphone
attached to the headphone 8, 10 is a listener wearing the headphone 8, 11a to 11b are FIR filters
for identification, 12a to 12b are LMS operators as identification coefficient updaters, and 13a to
13b are for identification It is a subtractor.
10-05-2019
12
[0064]
Initial coefficients are set to the FIR filters 1a to 1b from the RAMs 6a and 6b as described in FIG.
Therefore, the listener 10 listens to the audio reproduction sound of which the sound image
localization is controlled.
[0065]
On the other hand, the outputs of the FIR filters 1a-1b are input to the FIR filters 11a-11b and
the LMS arithmetic units 12a-12b, and at the same time, the output signals of the FIR filters 11a11b reproduced by the headphones 8 are detected by the microphones 9a-9b, It is input to the
subtractors 13a to 13b.
In the subtractors 13a to 13b, the output signals of the FIR filters 11a to 11b are subtracted
from the detection signals of the microphones 9a to 9b, and the results are input to the LMS
computing units 12a to 12b.
That is, in the LMS arithmetic units 12a to 12b, the least squares method is executed with the
outputs of the FIR filters 1a to 1b as reference signals and the outputs of the subtractors 13a to
13b as error signals.
By this, the coefficients of the FIR filters 11a to 11b approximate transfer characteristics from
the speakers 8a to 8b of the headphones 8 to the microphones 9a to 9b.
That is, the transfer functions Cr 'and Cl' of the headphones and the ear are to be measured.
Then, the measured coefficients are stored in the RAMs 6e and 6f.
[0066]
10-05-2019
13
Returning to FIG. 1, the newly measured coefficients are set in the RAMs 6e and 6f to the Fx
filters 3a to 3b, and the Filtered-x LMS method described above is executed using the
coefficients.
Then (Equation 5),
[0068]
Although the actual headphone and ear transfer functions are also Cr 'and Cl' (Eq. 6),
[0070]
Thus, even if there are individual differences, headphone wearing conditions, or characteristic
variations of the speaker unit of the headphones, it is possible to hear the target reproduced
sound.
[0071]
To summarize the above operation contents as a flowchart, first, the measurement of the
coefficients of the FIR filters 11a to 11b of FIG. 2 (= coefficients of Fx filters 3a to 3b of FIG. 1) is
as shown in FIG.
[0072]
First, the initial coefficient or the previous update coefficient is set in the FIR filters 1a to 1b for
sound image localization control, and sound image control of the audio signal is performed.
At the same time, coefficient measurement of the FIR filters 11a to 11b for Fx filter coefficient
measurement is performed.
First, the coefficients of the FIR filters 11a and 11b are initialized, and coefficient measurement
(= identification) is performed.
10-05-2019
14
When the coefficient measurement is completed, the coefficients are stored in the RAMs 6e and
6f.
During this time, the audio signal is normally subjected to sound image localization control, and
the listener 10 is not particularly aware of the Fx filter coefficient measurement.
[0073]
Next, the sound image localization control operation of FIG. 1 is as shown in FIG.
[0074]
As in the previous identification, first, the initial coefficients or the previous update coefficients
are set in the FIR filters 1a to 1b for sound image localization control, and sound image control of
the audio signal is performed.
At the same time, the coefficients stored in the RAMs 6e and 6f are set in the Fx filters 3a to 3b.
Next, the coefficients stored in the RAMs 6c and 6d are also set in the target characteristic filters
2a to 2b.
When these settings are completed, the coefficients of the FIR filters 1a to 1b are updated
according to the Filtered-x LMS method.
When the coefficients converge, the coefficients are stored in the RAMs 6a and 6b.
Then, this coefficient can be used at the next start of sound image localization control.
[0075]
From the above, it is possible to measure the transfer function of the headphones and the ear for
10-05-2019
15
each listener or for each wearing of the headphones of the listener, and to calculate the filter
coefficient for sound image localization control in the same manner.
Moreover, it is possible to execute the measurement of the transfer function and the calculation
of the filter coefficient while performing sound image localization control.
[0076]
As a result, the sound image localization control effect is not influenced by the individual
difference or the headphone wearing state, and a good effect can be obtained.
[0077]
By the way, how to obtain a target characteristic which has not been described yet will be
described.
[0078]
FIG. 5 shows a block diagram of sound image localization control headphones in the case of
measuring a target characteristic.
[0079]
In FIG. 5, 6c and 6d are RAMs as memories, 7 is a coefficient setting write control circuit for
controlling reading and writing of coefficient data in the RAMs 6c and 6d, and 9c to 9d are
listeners or dummy heads 19 assuming listeners. Microphones 14a to 14b are FIR filters, 15a to
15b are LMS operators, 16a to 16b are subtracters, 17 is a measurement signal generator for
measuring target characteristics, 18 is a measurement signal generator 17 A speaker for
reproducing the measurement signal through an amplifier (not shown), and a dummy head 19
assuming a listener or a listener.
[0080]
First, a measurement signal is output from the measurement signal generator 17 and reproduced
from the speaker 18.
Here, in the case where the center sound source is used as the target characteristic as in the
10-05-2019
16
present embodiment, the listener or the dummy head 19 is installed facing the speaker 18 on the
front axis of the speaker 18.
[0081]
On the other hand, the measurement signals of the measurement signal generator 17 are input to
the FIR filters 14a-14b and the LMS arithmetic units 15a-15b, and at the same time, the output
signals of the measurement signal generator 17 reproduced by the speaker 18 are detected by
the microphones 9c-9d. , Subtracters 16a to 16b.
The subtractors 16a to 16b subtract the output signals of the FIR filters 14a to 14b from the
detection signals of the microphones 9c to 9d, and the result is input to the LMS computing units
15a to 15b.
That is, in the LMS computing units 15a to 15b, the least squares method is executed with the
measurement signal of the measurement signal generator 17 as a reference signal and the
outputs of the subtractors 16a to 16b as an error signal.
By this, the coefficients of the FIR filters 14a to 14b approximate transfer characteristics Hr and
Hl from the speaker 18 to the microphones 9c to 9d.
Then, the measured coefficients are stored in the RAMs 6c and 6d.
[0082]
The target characteristics thus obtained are stored in the RAMs 6c and 6d, and are used at the
time of sound image localization control shown in FIG.
[0083]
At this time, if a plurality of target characteristics are measured and stored, for example, an
anechoic chamber characteristic, a characteristic of a certain room (a room with good acoustic
characteristics, etc.) or a characteristic when the speaker 18 is changed, the listener 10 The
desired target characteristics can be set in the target characteristic filters 2a to 2b at the time of
the sound image localization control of FIG. 1, and the desired sound image, the sound quality,
10-05-2019
17
the feeling of the sound field and the like can be enjoyed.
[0084]
In the first embodiment, the RAMs 6a to 6f are described as separate memories, but of course,
each may be mapped to one memory.
[0085]
Second Embodiment FIG. 6 is a block diagram of a sound image localization control headphone
according to a second embodiment.
[0086]
In FIG. 6, 1a to 1b are FIR filters that perform sound image localization control of audio input
signals, 2a to 2b target characteristic filters, 3a to 3b Fx filters that are transfer function
correctors, and 4a to 4b coefficient updaters LMS arithmetic unit, 5a-5b is a subtractor, 6a-6f is a
memory RAM, 7 is a coefficient setting write control circuit for controlling reading and writing of
coefficient data of RAM6a-6f, 8 is a sound image by FIR filters 1a-1b Headphones for
reproducing localization-controlled audio signals through an amplifier (not shown), 8a to 8b are
speakers attached to the headphone 8, 9a to 9b are microphones attached to the headphone 8,
10 is a listener wearing the headphone 8 , 20 is a storage medium such as a floppy (registered
trademark) disk, a memory card, an optical disk, etc. There is a personal computer.
[0087]
FIG. 6 shows the configuration of FIG. 1 to which a storage medium or a personal computer 20 is
added.
Therefore, since the basic sound image control operation is the same, only the storage medium or
the personal computer 20 will be described.
[0088]
When the storage medium or the personal computer 20 is, for example, a storage medium, a
plurality of respective coefficients used in the RAMs 6a to 6f are stored in the storage medium
10-05-2019
18
20, respectively. Can be entered.
For example, in the case where the listener 10 uses various headphones, storing coefficients in
the storage medium 20 adapted to a plurality of assumed headphones enables the listener 10 to
cope with the replacement of the headphones.
Further, by storing these coefficients in the storage medium 20, the memory capacity of the
RAMs 6a to 6f can be minimized.
[0089]
Next, consider the case where the storage medium or the personal computer 20 is a personal
computer.
Since a floppy disk drive, a CD-ROM drive, etc. are usually attached to the personal computer 20,
similar effects can be obtained if the floppy disk or the CD-ROM disk is used as the storage
medium described above.
Furthermore, if it is possible to download the coefficients used in the RAMs 6a to 6f from the
homepage of the Internet by using the network function, even without using a storage medium
such as a memory card, a floppy disk or a CD-ROM disk Similar effects can be obtained.
[0090]
Third Embodiment FIG. 7 is a block diagram of a sound image localization control headphone
according to a third embodiment.
[0091]
In FIG. 7, 1a to 1b are FIR filters that perform sound image localization control of audio input
signals, 2a to 2b target characteristic filters, 3a to 3b Fx filters that are transfer function
correctors, and 4a to 4b coefficient updaters LMS arithmetic unit, 5a-5b is a subtractor, 6a-6f is a
memory RAM, 7 is a coefficient setting write control circuit for controlling reading and writing of
coefficient data of RAM6a-6f, 8 is a sound image by FIR filters 1a-1b Headphones for
reproducing localization-controlled audio signals through an amplifier (not shown), 8a to 8b are
10-05-2019
19
speakers attached to the headphone 8, 9a to 9b are microphones attached to the headphone 8,
10 is a listener wearing the headphone 8 , 21a to 21b are convergence monitoring circuits, and
22a to 22b are switches.
[0092]
FIG. 7 shows the configuration of FIG. 1 to which convergence monitoring circuits 21a to 21b
and switches 22a to 22b are added.
Therefore, since the basic sound image control operation is the same, only the operations of the
convergence monitoring circuits 21a to 21b and the switches 22a to 22b will be described.
[0093]
As described in FIG. 1, the FIR filters 1 a to 1 b perform a coefficient update operation to perform
sound image localization control optimal for the listener 10.
When the coefficients of the FIR filters 1a to 1b converge, the coefficients are stored in the RAMs
6a to 6b.
[0094]
Here, if the coefficients of the FIR filters 1a to 1b converge, it is not necessary to continue the
coefficient update operation thereafter.
Therefore, the convergence monitoring circuits 21a to 21b monitor the degree of convergence of
the output signals of the subtractors 5a to 5b, that is, the error signal of the Filterd-x LMS
method, and when converged, the switches 22a to 22b are turned off to stop the coefficient
updating operation. (The switches 22a to 22b have been turned on until then).
[0095]
10-05-2019
20
The convergence monitoring circuits 21a to 21b constantly monitor the output signals of the
subtractors 5a to 5b. Therefore, if the output signal levels of the subtractors 5a to 5b are
increased due to some reason and the optimal sound image localization control state is lost,
immediately The switches 22a to 22b can be turned on to enter the coefficient update operation.
[0096]
The above applies to the identification operation for measuring the coefficients of the Fx filters
3a to 3b.
[0097]
FIG. 8 shows the configuration of FIG. 2 to which convergence monitoring circuits 21a to 21b
and switches 22a to 22b are added.
Therefore, the basic operations of the convergence monitoring circuits 21a to 21b and the
switches 22a to 22b are the same, and thus detailed description will be omitted.
[0098]
As described above, when the convergence monitoring circuits 21a to 21b and the switches 22a
to 22b are added, it is necessary to measure the coefficients of the Fx filters 3a to 3b, or it is
necessary to update the coefficients of the FIR filters 1a to 1b. Each of the operations can be
executed when a occurs, and unnecessary identification operation or coefficient update operation
can be stopped if each coefficient converges.
[0099]
Although in the third embodiment, the convergence monitoring circuits 21a to 21b monitor the
output signals of the subtractors 5a to 5b, other than this, the convergence monitoring circuits
21a to 21b perform subtraction on the input signals of the Fx filters 3a to 3b or the FIR filters 1a
to 1b. The configuration may be such that two signals of the output signals of the units 5a to 5b
are monitored, or two signals of the output signals of the Fx filters 3a to 3b or the FIR filters 1a
to 1b and the output signals of the subtractors 5a to 5b.
[0100]
(Fourth Embodiment) FIG. 9 is a block diagram of a sound image localization control headphone
10-05-2019
21
in the fourth embodiment.
[0101]
In FIG. 9, 1a to 1b are FIR filters that perform sound image localization control of audio input
signals, 2 is a target characteristic filter, 3 is an Fx filter that is a transfer function corrector, 4 is
an LMS operator that is a coefficient updater, and 5 is A subtractor, 6g to 6h are RAMs as
memories, 7 is a coefficient setting write control circuit for controlling reading and writing of
coefficient data of the RAMs 6g to 6h, and 8 is an audio signal whose sound image localization is
controlled by FIR filters 1a to 1b. Headphones 8a to 8b are speakers attached to the headphones
8, microphones 9a to 9b are microphones attached to the headphones 8, listeners wearing the
headphones 8, and 23a to 23d switches .
[0102]
9 uses the switches 23a to 23d, the target characteristic filters 2a to 2b in the configuration of
FIG. 1 as the target characteristic filter 2, the Fx filters 3a to 3b as the Fx filter 3, and the LMS
arithmetic units 4a to 4b as LMS operations. The subtractors 5a to 5b are combined in the
subtracter 5 in the unit 4, and the RAMs 6a, 6c and 6e are combined in the RAM 6g, and the
RAMs 6b, 6d and 6f are combined in the RAM 6h.
Therefore, since the basic sound image control operation is the same, only the contents regarding
the operation of the switches 23a to 23d will be described (the grouping of the RAMs 6a to 6f
has already been described in the first embodiment).
[0103]
In FIG. 1, the FIR filter 1a and the FIR filter 1b independently and simultaneously perform the
coefficient updating operation, but the configuration of FIG. 9 is configured to alternately execute
one by one.
[0104]
For example, assuming that the switches a to d are first turned on, the FIR filter 1a enters a
coefficient updating operation.
10-05-2019
22
Then, when the coefficient of the FIR filter 1a converges and is stored in the RAM 6g, next, the
switches 23a to 23d are turned on at the b side, and the coefficient updating operation of the FIR
filter 1b is started.
When the coefficients of the FIR filter 1b converge, the coefficients are stored in the RAM 6h.
[0105]
Thus, the configuration of FIG. 9 causes the FIR filter 1a and the FIR filter 1b in FIG. 1 to perform
the coefficient updating operation alternately.
As a result, the update portion required for the coefficient update operation becomes one each of
the Fx filter 3, the LMS operator 4 and the subtractor 5.
For this reason, the amount of calculations and memory capacity required for updating the
coefficients can be reduced.
[0106]
Naturally, the same applies to the identification of the Fx filter 3 in FIG.
[0107]
10 uses the switches 23d to 23e, the FIR filters 11a to 11b in the configuration of FIG. 2 are the
FIR filter 11, the LMS operators 12a to 12b are the LMS operators, and the subtractors 13a to
13b are the subtractor 13. The RAMs 6a and 6e are stored in the RAM 6g, and the RAMs 6b and
6f are stored in the RAM 6h.
[0108]
Therefore, as in the case of FIG. 9, the identification of the Fx filter 3a and the Fx filter 3b in FIG.
1 can be alternately performed, whereby the update portion necessary for the identification
operation is the FIR filter 11 and the LMS operator. 12 and one for each of the subtractors 13.
10-05-2019
23
For this reason, the amount of calculations and memory capacity required for identification can
be reduced.
[0109]
Furthermore, the same applies to the measurement of the target characteristic of the target
characteristic filter 2 in FIG.
[0110]
11 uses the switch 23d to set the FIR filters 14a to 14b in the configuration of FIG. 5 to the FIR
filter 14, the LMS operators 15a to 15b to the LMS operator 15, and the subtracters 16a to 16b
to the subtractor 16, respectively. In addition, the RAM 6c is summarized in the RAM 6g, and the
RAM 6d is summarized in the RAM 6h.
[0111]
Therefore, as in the case of FIGS. 9 and 10, target characteristic measurement of the target
characteristic filter 2a and the target characteristic filter 2b in FIG. 1 can be alternately
performed, whereby the update portion required for target characteristic measurement is , One
each of the FIR filter 14, the LMS computing unit 15 and the subtractor 16.
For this reason, the amount of calculations and memory capacity required for target
characteristic measurement can be reduced.
[0112]
(Fifth Embodiment) FIG. 12 is a block diagram of a sound image localization control headphone
in the fifth embodiment.
[0113]
In FIG. 12, 1c to 1f are FIR filters that perform sound image localization control of audio input
signals, 2 is a target characteristic filter, 3 is an Fx filter that is a transfer function corrector, 4 is
an LMS operator that is a coefficient updater, and 5 is A subtractor 6i is a memory RAM, 7 is a
coefficient setting write control circuit for controlling reading and writing of coefficient data of
10-05-2019
24
the RAM 6i, 8 is an audio signal whose sound image localization is controlled by the FIR filters 1c
to 1f through an amplifier not shown. 8a to 8b are speakers attached to the headphones 8, 9a to
9b are microphones attached to the headphones 8, 10 is a listener wearing the headphones 8,
24a to 24c are switches, and 25a to 25b are additions It is
[0114]
FIG. 12 shows an arrangement in which audio input signals are front Lch and front Rch, and
sound image localization control of the signals is performed.
[0115]
The coefficients of the FIR filters 1c to 1f are updated sequentially by the switches 24a to 24c.
[0116]
For example, when updating the coefficient of the FIR filter 1c, the switch 24a is turned ON to
select the front Rch, the switch 24b is turned ON to select the FIR filter 1c, and the switch 24c is
turned ON. The microphone 9a may be selected.
When updating the coefficient of the FIR filter 1e, the switch 24a is turned on to select the front
Lch, the switch 24b is turned on to select the c filter, the FIR filter 1e is selected, and the switch
24c is turned on. The microphone 9 b may be selected.
[0117]
As a result, the FIR filters 1c to 1f can be sequentially updated with coefficients, and the updated
portions required for the coefficient updating operation become one each of the Fx filter 3, the
LMS arithmetic unit 4 and the subtracter 5.
For this reason, the amount of calculations and memory capacity required for updating the
coefficients can be reduced.
[0118]
10-05-2019
25
Further, in FIG. 12, the Fx coefficient, the target characteristic coefficient, or the coefficients of
the FIR filters 1c to 1f are collectively stored in one RAM 6i.
[0119]
As described above, even when the audio input signal is the front Lch and the front Rch, the
coefficients of the FIR filters 1c to 1f performing the sound image localization control while
performing the sound image localization control can be executed as in the case of the center ch.
[0120]
Next, identification of the Fx filter 3 in FIG. 12 will be described.
[0121]
FIG. 13 shows a block diagram in the case of identifying the Fx filter 3 in FIG.
[0122]
In FIG. 13, 1c to 1f are FIR filters that control sound image localization of audio input signals, 6i
is a RAM that is a memory, 7 is a coefficient setting write control circuit that controls reading
and writing of coefficient data in the RAM 6i, and 8 is an FIR filter Headphones for reproducing
an audio signal whose sound image localization is controlled by 1c to 1f through an amplifier not
shown, 8a to 8b are speakers attached to the headphone 8, 9a to 9b are microphones attached to
the headphone 8, 10 is a headphone 8 A listener who is attached, 11 is an FIR filter for
identification, 12 is an LMS computing unit which is a coefficient updating unit for identification,
13 is a subtractor for identification, 24a to 24c are switches, and 25a to 25b are adders.
[0123]
As apparent from FIG. 13, by using the switches 24a and 24c, identification of the Fx filter 3 can
be sequentially performed while performing sound image localization control by the FIR filters 1c
to 1f.
In order to identify the Fx coefficient of the Fx filter 3 used for updating the coefficients of the
FIR filters 1c and 1f, the switches 24a to 24c may be turned on to the side a.
10-05-2019
26
In order to identify the Fx coefficient of the Fx filter 3 used for updating the coefficients of the
FIR filters 1d and 1e, the switches 24a to 24c may be turned on to the b side.
[0124]
As described above, even when the audio input signal is the front Lch and the front Rch,
identification of each Fx coefficient of the Fx filter 3 can be performed while performing sound
image localization control as in the case of the center ch.
[0125]
Next, measurement of the target characteristic of the target characteristic filter 2 in FIG. 12 will
be described.
[0126]
FIG. 14 shows a block diagram in the case where the target characteristic measurement of the
target characteristic filter 2 in FIG. 12 is performed.
[0127]
In FIG. 14, 6i is a RAM as a memory, 7 is a coefficient setting write control circuit for controlling
reading and writing of coefficient data in the RAM 6i, and 9c to 9d are attached to the listener or
the ear of the dummy head 19 assuming the listener. 14 is an FIR filter, 15 is an LMS operator,
16 is a subtractor, 17 is a measurement signal generator for measuring target characteristics,
and 18a to 18b are measurement signals from the measurement signal generator 17 via an
amplifier not shown. A speaker 19 reproduces the sound, 19 is a listener or a dummy head
assuming the listener, and 24c to 24d are switches.
[0128]
As apparent from FIG. 14, by using the switches 24c to 24d, it is possible to sequentially measure
the target characteristics Hrr and Hrl from the front Rch speaker 18a and the target
characteristics Hlr and Hll from the front Lch speaker 18b.
The target characteristic coefficients thus measured are stored in the RAM 6i.
10-05-2019
27
[0129]
Sixth Embodiment FIG. 15 is a block diagram of a sound image localization control headphone
according to a sixth embodiment.
[0130]
In FIG. 15, 1a to 1l are FIR filters that perform sound image localization control of multi-channel
audio input signals, 2 is a target characteristic filter, 3 is an Fx filter that is a transfer function
corrector, and 4 is an LMS operator that is a coefficient updater , 5 denotes a subtractor, 6i
denotes a RAM which is a memory, 7 denotes a coefficient setting write control circuit which
controls reading and writing of coefficient data of the RAM 6i, and 8 does not illustrate an audio
signal whose sound image localization is controlled by FIR filters 1a to 1l. Headphones 8a to 8b
are speakers attached to the headphones 8, 9a to 9b are microphones attached to the
headphones 8, 10 is a listener wearing the headphones 8, 25a to 25b is an adder, 26a ∼26c is a
switch.
[0131]
FIG. 15 shows a configuration for performing sound image localization control of audio input
signals in a sound source of multi-channel signals such as DVD.
FIG. 16 shows an example of a speaker arrangement for reproducing multi-channel signals of the
DVD 27. As shown in FIG.
The transfer characteristics of the left and right ears of the listener 19 are stored as target
characteristics in the RAM 6i from the speakers 18, 18a to 18e, and are set in the target
characteristic filter 2 in accordance with the coefficients of the FIR filters 1a to 1l.
[0132]
Similar to the case of FIG. 12, the coefficient updating of the FIR filters 1a to 1l is configured to
be sequentially executed by the switches 26a to 26c.
[0133]
10-05-2019
28
As a result, the update portion required for the coefficient update operation becomes one each of
the Fx filter 3, the LMS operator 4 and the subtractor 5.
For this reason, the amount of calculations and memory capacity required for updating the
coefficients can be reduced.
[0134]
Here, an Fx coefficient corresponding to the state of the switches 26a to 26c is set in the Fx filter
3.
In the state shown in FIG. 17, since the switch 26a is on the f side, the switch 26b is on the l side,
and the switch 26c is on the b side, the coefficients of the FIR filter 1l are updated.
Therefore, an Fx coefficient indicating a transfer function from the speaker 8 b to the
microphone 9 b is set in the Fx filter 3.
Similarly, in the target characteristic filter 2, target characteristic coefficients corresponding to
the states of the switches 26a to 26c are set from the RAM 6i.
[0135]
As described above, even in the case where the audio input signal is multi-channel, it is possible
to perform the coefficient update of the FIR filters 1a to 1l performing the sound image
localization control while performing the sound image localization control.
[0136]
Seventh Embodiment FIG. 17 is a block diagram of a sound image localization control headphone
according to a seventh embodiment.
[0137]
In FIG. 17, 1a to 1l are FIR filters that perform sound image localization control of multi-channel
audio input signals, 2 is a target characteristic filter, 3 is an Fx filter that is a transfer function
10-05-2019
29
corrector, and 4 is an LMS operator that is a coefficient updater , 5 denotes a subtractor, 6j
denotes a RAM which is a memory, 7 denotes a coefficient setting writing control circuit which
controls reading and writing of coefficient data of the RAM 6j, and 8 denotes an audio signal
whose sound image localization is controlled by FIR filters 1a to 1l. Headphones 8a to 8b are
speakers attached to the headphones 8, 9a to 9b are microphones attached to the headphones 8,
10 is a listener wearing the headphones 8, 25a to 25b is an adder, 26a 26c is a switch, and 28a
to 28m are reflected sound generation circuits.
[0138]
In FIG. 17, the reflected sound generation circuits 28 a to 28 l are cascaded to the FIR filters 1 a
to 1 l of FIG. 15, so the coefficient update unit also has the reflected sound generation circuit 28
m cascaded to the target characteristic filter 2. is there.
The reflected sound generation circuits 28a to 28l are circuits for simulating the reflected sound
characteristics of a certain room, and the configuration thereof is shown as in FIG. 18 or FIG.
By this, for example, it is possible to simulate reflected sound and reverberation when the
speakers 18, 18a to 18e shown in FIG. 16 are installed in a listening room or the like.
Then, the FIR filters 1a to 11 are only required to simulate only the direct sound from the
speakers 18, 18a to 18e to the left and right ears of the listener 19. For example, the target
characteristic of the target characteristic filter 2 is measured in an anechoic chamber And the
coefficients of the target characteristic filter 2 and the FIR filters 1a to 1l can be shortened.
Therefore, the amount of calculations and memory capacity can be reduced accordingly.
[0139]
In the coefficient updating operation of the FIR filters 1a to 1l, the reflected sound generation
circuit 28m is connected to the target characteristic filter 2 in a cascade connection
configuration, and similarly to the Fx filter 3 and the target characteristic filter 2, reflected sound
characteristics corresponding to the states of the switches 26a to 26c. That is, the delay time of
the delay devices 29a to 29N of FIG. 18 or FIG. 19, the gain values of the level adjusters 30a to
30N, and the frequency characteristics of the f characteristic adjusters 31, 31a to 31N are set by
10-05-2019
30
the RAM 6j. It is executed in the same manner as in the case of 15.
[0140]
Naturally, the reflected sound generation circuits 28a to 28m may be applied to the first to fifth
embodiments.
[0141]
As described above, even in the case where the audio input signal is multi-channel, it is possible
to perform the coefficient update of the FIR filters 1a to 1l performing the sound image
localization control while performing the sound image localization control.
[0142]
According to the sound image localization control headphone of the present invention, the
microphone is mounted in the headphone so as to be disposed in the vicinity of the ear of the
listener, and the target characteristic filter in which the coefficient serving as the target
characteristic of sound image localization control is set Subtractor to subtract target
characteristic filter output from output, transfer function corrector to correct transfer
characteristic between headphone and speaker, coefficient update to calculate coefficient of
sound image localization control by output from transfer function corrector and subtractor By
using the digital signal processor and the digital filter that processes the input signal by using the
coefficient obtained by the coefficient updater and outputs the signal to the headphone, it is
possible to simultaneously update the coefficient while performing sound image localization
control with the digital filter.
As a result, the sound image localization control effect is not influenced by the individual
difference or the headphone wearing state, and a good effect can be obtained.
[0143]
In the sound image localization control headphone of the present invention, the input signal is
subjected to signal processing using a digital filter in which coefficients for performing sound
image localization control are set, the output is reproduced from the headphone, and the
reproduced sound is in the vicinity of the listener's ear It detects with the microphone attached in
10-05-2019
31
the headphone so that it is arranged, processes the output of the digital filter for sound image
localization control with the digital filter for identification, subtracts the digital filter output for
identification from the microphone output with the subtractor, The coefficients of the
identification digital filter are calculated by the identification coefficient updater based on the
control digital filter output and the output from the subtracter, and the coefficients of the
identification digital filter are controlled by the digital filter for sound image localization control
while the sound image localization control is performed. Updates can also be performed
simultaneously.
For example, if the input signal is a music signal, the listener can identify the digital filter for
identification without being aware of the identification operation while listening to music.
When the coefficient updating of the digital filter for sound image localization control described
above is executed by setting the identified coefficient as the coefficient of the transfer function
corrector, the sound image localization control effect is not influenced by the individual
difference or the headphone wearing state. , Good effect can be obtained.
[0144]
In particular, by storing the coefficients of the digital filter for sound image localization control
and the digital filter for identification in the memory, optimal coefficients such as the previously
updated coefficient are set as initial values in addition to the standard initial coefficients stored in
advance. can do.
Further, since a plurality of coefficients can be stored, coefficients respectively corresponding to
various headphones can be obtained and stored.
[0145]
Also, by storing the target characteristic coefficient of the target characteristic filter in the
memory, it is possible to store a plurality of coefficients corresponding to, for example, an
anechoic chamber characteristic, a room characteristic with good acoustic characteristics, various
speaker characteristics, etc. The target characteristic filter can be set according to
10-05-2019
32
That is, the sound image localization control effect, the sound field control effect, the sound
quality and the like can be obtained each time according to the preference of the listener.
[0146]
In addition, the digital filter for sound image localization control, the digital filter for
identification, and the coefficients of the target characteristic filter can be downloaded from a
storage medium or a personal computer, so the memory in the body of the sound image
localization control headphone is minimally required. It does not matter by the capacity.
Also, the fact that a new coefficient can be set in the memory means that a coefficient to be
provided in the future even if it is not supported at that time (for example, the acoustic
characteristic of the latest concert hall is made a target characteristic filter coefficient) Can also
be used to improve effects or add new effects.
10-05-2019
33
Документ
Категория
Без категории
Просмотров
0
Размер файла
47 Кб
Теги
jp2002135898
1/--страниц
Пожаловаться на содержимое документа