close

Вход

Забыли?

вход по аккаунту

JP2011071665

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 JP2011071665
The present invention provides an acoustic apparatus capable of sound image localization at an
arbitrary angle with high accuracy while reducing the amount of calculation. An arithmetic unit
(20) refers to stored contents of a head related transfer function storage part (12) and reads out
head related transfer functions for right and left ears of two angles sandwiching a given angle,
and stores an initial delay sample number Reading the initial delay sample number for the left
and right ears with respect to the given angle with reference to the stored contents of the part
14, interpolating the read out two head transfer functions for each of the left ear and the right
ear, and giving the given angle The head related transfer functions for the left ear and the right
ear for. Since the delay processing corresponding to the read initial delay sample number
corresponding to each of the left ear and the right ear is performed on the head related transfer
functions of the left ear and the right ear calculated by the interpolation, the amount of
calculation is reduced. While being able to localize the virtual sound source accurately. [Selected
figure] Figure 1
Acoustic device
[0001]
The present invention relates to an improvement of an acoustic device that makes it possible to
virtually localize a sound source using a head-related transfer function that is an impulse
response in the time domain.
[0002]
10-05-2019
1
The head-related transfer function is an acoustic transfer function from the sound source to the
left and right ears, and specifically refers to the response characteristic when an acoustic
impedance is given to the listener of the acoustic signal from a certain direction.
This head related transfer function is called HRTF (Head Related Transfer Function) in the
frequency domain, and the one expressed in the time domain by inverse Fourier transform of this
is called HRIR (Head Related Impulse Response). FIG. 7 shows that the head transfer function at
the angle A and the angle B is known, assuming that the front right of the listener is the
reference angle "0 degree" and the opposite angle to the rear right just behind is "180 degrees".
It is a schematic explanatory drawing of the prior art which calculates ¦ requires the head related
transfer function in the angle C pinched ¦ interposed by two angles. Here, the head related
transfer function at angle A consists of impulse responses Al and Ar for each of the left ear LE
and the right ear RE of the listener, and similarly, the head related transfer function at angle B is
the listener's Impulse responses Bl and Br for the left ear LE and the right ear RE, respectively.
Then, a method has been proposed in which responses Cl and Cr to the listener's left ear LE and
right ear RE at angle C are determined by "interpolation between Al and Bl" and "interpolation
between Ar and Br", respectively. . In general, since there is a difference in arrival time depending
on the angle, the head transfer function has a state in which a plurality of samples 0 from
the head is reached with a later arrival, and angles A and B used for interpolation occur. In the
rising of the impulse response of the above, there is also a problem that a shift of several samples
of "0" occurs. Therefore, an apparatus has been proposed that corrects the time axis and converts
it into the frequency domain to correct the rise time difference of the impulse response. As an
apparatus for performing this time axis correction, for example, an audio signal is converted to a
digital signal, a convolution operation with a tap coefficient is performed, and the audio signal is
converted to an audio signal and reproduced again. The tap coefficients are represented by the
interpolation values obtained by the internal division method using discrete tap coefficients, and
as a result, the characteristics are continuously changed smoothly and the direction of the sound
source is continuously changed. A sound reproduction apparatus has been proposed (see, for
example, Patent Document 1). Further, as an apparatus for converting to a frequency domain and
correcting, an acoustic apparatus has been proposed in order to perform binauralization at an
arbitrary angle by calculation (for example, see Patent Document 2). In this apparatus, signals
from the control circuit are supplied to a memory storing the left and right head transfer
functions measured at a plurality of angles at predetermined angular intervals, and any desired
sound image localization is desired. The transfer function of the angle is read, and further, the
read head transfer function is written to the respective registers, and the signals from these
registers are respectively given to the interpolation / interpolation arithmetic circuit.
Then, a signal for controlling the interpolation / interpolation ratio from the control circuit is
supplied to the arithmetic circuit, and an operation according to this ratio is performed. Since the
10-05-2019
2
head related transfer function calculated by this is supplied to the conversion circuit to generate
left and right sounds, only by measuring the head related transfer function corresponding to a
plurality of angles, binauralization of an arbitrary angle is performed by calculation. realizable.
[0003]
Japanese Patent Application Laid-Open No. 60-9212 (Page 1-2, FIG. 1) Japanese Patent
Application Laid-Open No. 5-300599 (Page 2-3, FIG. 1)
[0004]
However, in the second apparatus described above that performs conversion into the frequency
domain and performs correction, it is generally necessary to convert the calculation result into
the time domain in the calculation in the frequency domain, and the amount of calculation at that
time is It becomes extremely large, and as a result, a device with a simple configuration can not
be realized.
In addition, even in the first apparatus described above that performs correction on the time axis,
it is necessary to hold "0" which continues several samples from the beginning of the filter
coefficient, and this extra memory capacity is required. The Moreover, in order to carry out linear
interpolation of the delay between two filter coefficients, the case where the amount of shift ¦
offset ¦ differences of an actual delay amount and a delay interpolation result became large also
occurred.
[0005]
The present invention has been made to solve such conventional problems, and it is an object of
the present invention to provide an acoustic device capable of sound image localization at any
angle with high accuracy while reducing the amount of calculation.
[0006]
In order to achieve the above object, according to the present invention, an initial delay sample
number storage unit storing information (initial delay sample number) indicating an initial delay
of an arrival signal to a listener at a first plurality of angles; A head-related transfer function
storage unit storing a head-related transfer function that is an impulse response in the time
domain for each of the left and right ears at a second plurality of angles, excluding the number of
initial delay samples And a computing means for obtaining a head related transfer function with
10-05-2019
3
respect to an angle given with reference to the memory contents of the storage means, wherein
the first plurality of angles are the second plurality The calculation means is set to have a larger
number of angles including the angle of. The arithmetic means refers to the stored contents of
the head related transfer function storage unit, and the left and right sides of two angles
sandwiching the given angle Head against ear The readout function reads out the transfer
function and the initial delay sample number for the left and right ears of the given angle with
reference to the stored contents of the initial delay sample number storage unit, and the two
head transfer functions read out Interpolation means for interpolating for each ear and right ear
to obtain head related transfer functions for left and right ears for a given angle, and for head
related transfer functions of left and right ears determined by this interpolation And delay means
for performing delay processing according to the corresponding read initial delay sample
number for each of the left ear and the right ear.
[0007]
According to the present invention, the readout means refers to the stored contents of the head
related transfer function storage unit and reads out the head related transfer functions for the
left and right ears of two angles sandwiching a given angle, and the initial delay sample With
reference to the stored contents of the number storage unit, the number of initial delayed
samples of the left and right ears for a given angle is read out, and the two head related transfer
functions read out by the interpolation means are interpolated for each left ear and right ear, The
HRTF for the left ear and the right ear for a given angle is determined, and the delay means
corresponds to the HRTF for the left ear and the right ear obtained by this interpolation for each
of the left ear and the right ear Since the delay processing according to the number of initial
delay samples read out is performed, the virtual sound source can be localized with high
accuracy, and the processing on the time axis is performed, so that the amount of calculation is
reduced and as a result, the memory capacity used is larger than that of the conventional device
Also It becomes no.
[0008]
Further, in this acoustic device, the readout means refers to the contents stored in the initial
delay sample number storage unit, and when there is no initial delay sample number for both the
left and right ears matching the given angle, The number of initial delay samples for left and
right ears of two angles sandwiching a given angle is read out, and the interpolation means
further interpolates the number of two initial delay samples read out for each left ear and right
ear. If the number of initial delay samples for the left ear and the right ear for a given angle is
determined, it is not necessary to store the number of initial delay samples at every small angle.
10-05-2019
4
Preferably, the first plurality of angles is 360 degrees per degree, and the second plurality of
angles is 12 degrees per 30 degrees in terms of accuracy, and the number of initial delay
samples is Specifically, this is the number of "0" continuing from the top of the sound source
arrival signal.
[0009]
According to the present invention, it is possible to obtain an effect that sound image localization
can be performed at any angle with high accuracy while reducing the amount of calculation.
[0010]
FIG. 1 is a block diagram of a sound device 1;
FIG. 2 is an explanatory diagram of storage contents of a memory 10;
It is explanatory drawing of the number of initial delay samples.
It is explanatory drawing which shows the relationship between an angle and the number of
initial delay samples. It is explanatory drawing of operation ¦ movement. 5 is a flowchart
illustrating the operation of the audio device 1; It is explanatory drawing of a prior art. It is
explanatory drawing of operation ¦ movement.
[0011]
Embodiments of the present invention will be described below with reference to the drawings.
[0012]
(Configuration) FIG. 1 is a configuration diagram of an acoustic device 1 according to an
embodiment of the present invention.
10-05-2019
5
The acoustic device 1 includes a memory 10 (storage means), an operation unit 20, an operation
unit 30, FIR filters 50 and 55, and an input signal source 40. Signal is supplied to two FIR filters
50, 55. Each of the FIR filters 50 and 55 is supplied with an L channel coefficient and an R
channel coefficient from the operation unit 20, and the signal from the input signal source 40 is
subjected to processing such as multiplication by the supplied coefficient. A channel sound and
an R channel sound are output, and when both sounds are listened by headphones etc., desired
virtual sound source localization is performed. The acoustic device 1 can be realized by executing
a required process using a RAM or the like as a work area according to a program recorded by a
CPU or DSP in a recording medium such as a ROM.
[0013]
The memory 10 has a head related transfer function storage unit 12 and an initial delay sample
number storage unit 14. Note that the head-related transfer function in the present embodiment
refers to an impulse response in the time domain, not a transfer function in the frequency
domain. As shown in FIG. 2 (a), in the head related transfer function storage unit 12, with the
reference angle "0 degree" directly in front of the listener, the head related transfer function for
the left ear and the right for each 30 degrees in the clockwise direction Ear head transfer
functions are measured and stored in advance. In this example, it is set every 30 degrees, but it
may not be 30 degrees if it is measured at an angle of 90 degrees or less. As shown in FIG. 2 (a),
the head transfer function for the left ear is "HL0, HL1, HL2, ..., HL6, ..., HL9, HL10, HL11", and
the head transfer function for the right ear is " HR0, HR1, HR2, ..., HR6, ..., HR9, HR10, HR11 ". In
addition, since the head-related transfer functions of 180 degrees to 360 degrees are obtained
when the head-related transfer functions of 0 degrees to 180 degrees for left and right ears are
interchanged, "HR11, HR10, HR9, ..." are "HL1, HL2 "," And "HL11, HL10, HL9, ..." are "HR1, HR2,
...". Here, what is important here is that these head-related transfer functions (impulse response)
stored in the head-related transfer function storage unit 12 are the ones obtained by removing
the number of initial delay samples described below.
[0014]
FIG. 2 (b) shows the initial delay sample count stored in the initial delay sample count storage
unit 14, with the reference angle "0 degree" in front of the listener as the reference angle "0
degree" for the left ear initial delay sample every one degree The number of samples and the
number of initial delayed samples for the right ear are previously measured and stored. In this
example, the angle is set to one degree, but if the measurement angle is set at an angle within
one degree, the accuracy is further improved. As shown in FIG. 2 (b), the number of initial delay
10-05-2019
6
samples for the left ear is "TL0, TL1, TL2, ..., TL179, ..., TL357, TL358, TL359", and the number
of initial delay samples for the right ear is " TR0, TR1, TR2, ..., TR179, ..., TR357, TR358, TR359 ".
Note that if the initial delay sample numbers for 0 ° to 180 ° for the left and right ears are
interchanged, the initial delay sample number for 180 ° to 360 ° is obtained, so TR359,
TR358, which is the initial delay sample number for the right ear "TR357, ..." becomes "TL1,
TL2, ..." and "TL359, TL358, TL357, ... is" TR1, TR2, ... ".
[0015]
Now, FIG. 3 is a view showing an impulse response reaching one ear (for example, the left ear) at
two angles, and the horizontal axis is a time axis t. Since there is a distance difference (in other
words, a phase difference) at the two angles, the signal WA2 is delayed with respect to the signal
WA1, and an initial delay occurs. As described above, when the initial delay occurs, the signal
0 continues from the beginning by the delay amount, which is referred to as the initial delay
sample number. FIG. 4 shows the number of initial delay samples actually measured by the
inventor (the number of samples in which "0" continues from the beginning). Fig. 4 shows the
number of initial delay samples (the number of samples for which "0" continues from the
beginning) of the head transfer function (impulse response) from the front of the listener (0
degrees) to the rear of the listener (180 degrees) It is the figure which plotted, the horizontal axis
is an angle, and a vertical axis is the number of initial delay samples. The graph S1 shows the
graph for the right ear, and the graph S2 shows the graph for the left ear. As can be seen with
reference to FIG. 4, the relationship between the angle and the number of initial delay samples is
not necessarily linear. Therefore, linear time interpolation may deviate greatly from the actual
value. Therefore, in the present embodiment, the initial delay sample number is measured and
stored for both the left and right ears at every one degree, and the head-to-peak transfer function
without the initial delay sample number is used to raise the rise between the signals. Aligned,
started out aligned to improve localization accuracy.
[0016]
(Operation) It is assumed that the listener gives an angle of 75 degrees through the operation
unit 30 in the state shown in FIG. Note that the front right of the listener is "0 degrees" and the
rear right is "180 degrees". LE indicates the left ear, RE indicates the right ear, and H indicates
the head. Then, the angle increases in the clockwise direction from the reference angle 0 degree,
and the angle is set to return to the reference angle 0 degree at 360 degrees. First, as shown in
step S600 in FIG. 6, the operation unit 20 refers to the stored contents of the head related
transfer function storage unit 12 and sets two angles 60 degrees and 90 degrees sandwiching
10-05-2019
7
the given angle 75 degrees. The head related transfer functions for the left and right ears are
read out, and the contents stored in the initial delay sample number storage unit 14 are
referenced to read out the initial delay sample numbers for the left and right ears at a given
angle of 75 degrees. Thereby, four head transfer functions (60 degrees left and right and 90
degrees left and right) and two (75 degrees left and right) initial delay sample numbers are read.
[0017]
Next, the interpolation processing in step S610 of FIG. 6 is performed. The difference between
the angles of 90 degrees and 60 degrees is 30 degrees, and the difference between the angles of
60 degrees and 75 degrees and between the angles of 90 degrees and 75 degrees is 15 degrees,
so one to one In proportion, the two head related transfer functions read out are interpolated
for each of the left ear and the right ear. First, add one half of the head related transfer functions
at an angle of 60 degrees and an angle of 90 degrees to the left ear and multiply them by "1/2"
for the number of initial delay samples of 75 degrees to the left ear. It is added to the top (step
S620 in FIG. 6). Similarly, each of the head transfer functions at an angle of 60 degrees and an
angle of 90 degrees with respect to the right ear is multiplied by 1/2 and added, and this is
added to the number of initial delay samples for 75 degrees with respect to the left ear "" Is
added to the top (step S620 of FIG. 6). In general, assuming that two angles sandwiching a given
angle θ are θ1 and θ2, H1 × ((θ−θ1) / (θ2−θ1) + H2 × (θ2−θ) / (θ2−θ2):
However, H1 and H2 perform head transfer functions at angles θ1 and θ2, θ2> θ1 on the left
ear and the right ear, and then 0 at the beginning by the number of corresponding initial
delay samples Add
[0018]
Then, the head transfer function (impulse response) to which "0" is added at the beginning by the
number of initial delay samples obtained by this arithmetic processing is the FIR filter 50, as an L
channel coefficient and an R channel coefficient, respectively. The FIR filters 50 and 55
respectively multiply the given coefficients by the signal from the input signal source 40 to
output an L channel signal and an R channel signal. FIG. 8 is an explanatory view of this
operation. In each case, the horizontal axis is a time axis t. As shown in FIGS. 8 (a) and 8 (b), the
impulse responses with angles of 60 degrees and 90 degrees except the initial delay (number of
initial delay samples) are shown in FIGS. 8 (d), and the signals shown in FIGS. 8 (c) and 8 (d) are
interpolated (specifically, to the signals shown in FIGS. 8 (c) and 8 (d)). The result obtained by
adding the initial delay sample number for a given angle of 75 degrees to the head thereof is as
shown in FIG. 8 (e).
10-05-2019
8
[0019]
As described above, according to the embodiment of the present invention, the virtual sound
source can be localized with high accuracy, and processing on the time axis is performed, so that
the amount of computation is reduced, and as a result, the memory capacity to be used is
reduced.
[0020]
When there is no initial delay sample number at an angle matching the given angle, the same
interpolation processing as when the head related transfer function is obtained may be
performed.
In the embodiment described above, the head transfer functions are stored at equal angular
intervals. However, it is not necessary to store the head transfer functions at equal angular
intervals, for example, 0 The same process can be performed even if the head related transfer
functions are stored at angular intervals such as 45 degrees, 90 degrees, 120 degrees, 180
degrees, and so on.
[0021]
As described above, the present invention can be applied to a music apparatus for performing
sound image localization on a given sound signal.
[0022]
DESCRIPTION OF SYMBOLS 1 acoustic apparatus 10 memory 12 head-related transfer function
storage part 14 initial delay sample number storage part 20 calculating part 30 operation part
40 input signal source 50 FIR filter 55 FIR filter
10-05-2019
9
1/--страниц
Пожаловаться на содержимое документа