JPH0530585

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 JPH0530585
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
noise reduction headphone device which is preferably used in places where external noise is
severe.
[0002]
2. Description of the Related Art In a headphone device for listening to audio signals such as
voice and music in general, when it is used in a place where the external noise is intense,
listening to the voice and music (hearing) is obstructed due to the mixed noise. Problem of being
The following techniques are known to solve such a problem.
[0003]
The first technique is to change the frequency characteristics of a voice or music signal to make
the noise less noticeable. This technique not only has the potential to damage the original voice
and music, but also has the disadvantage that it is effective only on specific noises.
[0004]
08-05-2019
1
In the second technique, as shown in FIG. 4 and FIG. 5, external noise is picked up by the
microphone 2 attached to the outside of the headphone case 1, and this picked up signal is sent
to the equalizer 3, The signal is adjusted so that the amplitude is equal and in opposite phase to
the noise mixed into the inside of the headphone 1 in 3 and then sent to the (speaker of) the
headphone 5 via the adder 4 for reproduction. The noise reaching the ear (ear canal) 6 is to be
canceled. The adder 4 is supplied with an audio signal or a music signal to be listened to from the
signal input terminal 7.
[0005]
FIG. 5 shows the configuration of FIG. 4 in the form of a block using a transfer function. The
external noise (noise) N from the input terminal 11 is acousto-electrically converted by the
microphone 2 whose transfer function is M and transmitted Electro-acoustic conversion is
performed on (the speaker of) the headphone 5 whose transfer function is H via the equalizer 3
whose function is −γ, and is sent to the adder 12. The external noise (noise) N is supplied to the
adder 12 through the headphone casing 1 of the sound insulation characteristic F, and is
additively mixed in the headphone internal space, and the human ear (ear canal 6 through the
output terminal 13 To reach).
[0006]
In such a configuration, since the addition output from the output terminal 13 is N · F + N · M ·
(−γ) · H = N (F−M · γ · H), the transfer function −γ of the equalizer 3 is If adjustment is
made so that γ = F / (H · M), F−M · γ · H = 0, and noise can be canceled. In addition, although
illustration is abbreviate ¦ omitted in FIG. 5, the electrical signal of the audio ¦ voice and music
which it is going to listen to is supplied and superimposed between the equalizer 3 and the
headphones 5 (speaker). According to this, it has the advantage that it does not affect the original
voice and music, and in principle does not depend on the nature of the noise.
[0007]
However, in fact, if the adjustment by the equalizer 3 is not accurately performed, the noise
cancellation effect is thin, and the characteristics of the headphones 5 are mainly determined by
the shape of the ear canal and the way of wearing the headphones after adjustment. Electroacoustic conversion characteristics) H may change to lower the effect.
08-05-2019
2
[0008]
Next, as a third technique, as shown in FIGS. 6 and 7, one in which the adaptive filter 16 replaces
the equalizer 3 configured as shown in FIGS. 4 and 5 is known.
That is, in FIG. 6, the first microphone 14 for noise collection provided at a position outside the
headphone housing 1 and near the ear when the headphones are mounted, and when the
headphones are mounted inside the headphone housing 1 The adaptive filter 16 filters the input
from the first microphone 14 and sends it to the adder 4 using the second microphone 15
provided at a position between the headphone (the speaker) 5 and the ear canal. The filter
characteristic (transfer function −γ) at this time is adaptively controlled so that the input from
the second microphone 15 is minimized. FIG. 7 is a block diagram showing the transfer function
of each part. The transfer function of the first microphone 14 is M1, and the transfer function of
the second microphone 15 is M2. The other parts of the configuration are the same as in FIG. 4
and FIG. Further, the signal to be listened to, such as voice and music, has no correlation with the
noise to be reduced or canceled, and does not affect the noise reduction operation, so the
description will be omitted.
[0009]
According to this configuration, the signal from the adder 4 is sent to (the speaker of) the
headphone 5 to reduce the noise component, and the noise-reduced sound in the headphone is
collected by the second microphone 15 and the so-called The residual is sent to the adaptive
filter 16, which learns to minimize the power of the residual signal and adjusts its own filter
characteristics.
[0010]
The internal configuration of the adaptive filter 16 comprises a filter unit 21 and an adaptive
algorithm unit 22 as shown in FIG.
The input from the first microphone 14 is supplied to the terminal 17 and sent to the filter unit
21 and the adaptive algorithm unit 22 through the input terminal 16a of the adaptive filter 16 as
a so-called reference input x. The output y from the filter unit 21 is taken out as the output of the
adaptive filter 16 through the terminal 16 b and sent to the adder 18. The adder 18 represents
08-05-2019
3
acoustic mixing inside the headphone, and a component d of which external noise (noise) has
reached the inside of the headphone via the sound insulation characteristic functional block such
as a headphone case is connected via the terminal 19 A so-called noise residual e (= dy) is
obtained by being sent to the adder 18 and subtracting the above-mentioned adaptive filter
output y from the noise component d. The residual e from the adder 18 is picked up by the
second microphone 15 and supplied to the adaptive algorithm unit 22 through the terminal 16c.
The adaptive algorithm unit 22 changes the filter coefficient of the filter unit 21 to change the
filter characteristics, thereby minimizing the power of the residual e with respect to the output y
obtained by filtering the input x. It controls.
[0011]
According to the third technique, since adjustment is performed by learning, the adjustment is
accurate, the noise reduction effect can be enhanced, and it is easy to re-adjust it even if the
characteristics of the headphone change later. There is. Therefore, among the three techniques
described above, this third technique is considered to be the most effective.
[0012]
By the way, in the noise reduction using the adaptive filter described as the third technique, there
is a problem in timing deviation when using a sequential adaptive algorithm. This problem is
explained below.
[0013]
FIG. 9 shows an internal configuration of the adaptive filter 16, and shows a specific example
using a so-called FIR (finite impulse response) filter as the filter unit 21. As shown in FIG. In FIG.
9, the reference input x from the input terminal 16a is sent to a series circuit of delay elements
231, 232,..., 23L according to the number of taps. The input x0 from the input terminal 16a and
the outputs x1, x2, ..., xL from the delay elements 231, 232, ..., 23L are coefficient multipliers
240, 241, 242, ..., 24L, respectively. , And are multiplied by the filter coefficients w 0, w 1, w 2,...,
W L and sent to the adder 25. The filter coefficients w0, w1, w2,..., WL are corrected by the
coefficient correction signal from the adaptive algorithm unit 22, and the output y from the
adder 25 is taken out from the output terminal 16b.
08-05-2019
4
[0014]
As the adaptive algorithm used in the adaptive algorithm section 22, many methods have been
proposed, and one specific example thereof is the LMS (least mean square, least mean square)
algorithm which is a kind of sequential adaptive algorithm. Will be explained.
[0015]
The input data at the k-th sample cycle time point (time k) of the data series of the input x and
the delayed output data from each of the delay elements 231, 232, ..., 23L are xk0, xk1, xk2, ... ..,
XkL, an input vector Xk subjected to FIR filter processing is set as Xk = [xk0 xk1 xk2 ... xkL] T (1).
T in this equation (1) represents a transpose. Assuming that the above filter coefficients
(weighting coefficients) are wk0, wk1, wk2,..., WkL for this input vector Xk, and the FIR filter
output is yk, the relationship between input and output is given by the following equation (2) It
will be.
[0016]
yk = wk0xk0 + wk1xk1 +... + wkLxkL (2) Further, if the filter coefficient vector (weighting vector)
Wk is defined as Wk = [wk0 wk1 wk2 wkL] T (3), input / output The relationship is described as
yk = XkTWk (4). Assuming that the desired response is dk, the error .epsilon.k from the output is
expressed as .epsilon.k = dk -yk = dk -Xk T Wk (5). Using these, the LMS algorithm is expressed
as Wk + 1 = Wk + 2μεk Xk (6). (6) is a gain factor that determines the speed and stability of
adaptation.
[0017]
When the adaptive filter 16 in which this LMS algorithm is an adaptive algorithm is applied to
the configuration of FIG. 6 and FIG. 7, it will be represented as shown in FIG. In FIG. 10, an
adaptive algorithm is applied for a time delay which occurs between reproduction (electroacoustic conversion) by the headphone (speaker) 5 and sound collection (acoustic-electrical
conversion) by the microphone 15. The residual input to the unit 22 becomes uncorrelated with
08-05-2019
5
the reference input supplied to the filter unit 21 at the same time, and the condition of the
equation (6) is not satisfied. That is, assuming that the delay time in the headphone 5 is virtually
d1 and the delay time in the microphone 15 is d2, the remaining reference input to the adaptive
algorithm unit 22 for the reference input xk supplied to the filter unit 21. The difference is
delayed by the delay times d1 and d2 as in ek-d1-d2, for example.
[0018]
As described above, when the timing of each input supplied to the adaptive filter deviates,
effective noise reduction can not be performed because the condition of the above equation (6) is
not satisfied. There is a disadvantage that the adaptive algorithm can not be used. That is, since
the amount of calculation can be small, it is impossible to use a sequential adaptation algorithm
suitable for practical use by an LSI or the like.
[0019]
The present invention has been made in view of such circumstances, and in a noise reduction
method using an adaptive filter, a noise reduction headphone that enables use of a so-called LMS
algorithm or a sequential adaptation algorithm such as a learning identification method. The
purpose is to provide a device.
[0020]
SUMMARY OF THE INVENTION A noise reduction headphone device according to the present
invention comprises a first acoustic-electrical conversion means provided in the vicinity of the
ear at the time of wearing the headphone to pick up external noise, and the ear canal at the time
of wearing the headphone. From the second acoustic-electrical conversion means, and the second
acoustic-electrical conversion means located between the electrical-acoustic conversion means
and the ear canal when the headphone is attached First adaptive filter means for adaptively
filtering and outputting the input from the first acousto-electrical conversion means so as to
minimize the power of the first input, and the output of the first adaptive filter means Second
adaptive filter means that adaptively filters and outputs so as to minimize the power of the
residual between the supplied output from the filter unit and the output from the second
acousto-electrical conversion means; This second The filter coefficient of the adaptive filter
means is provided, and filter means processing the input from the first acousto-electrical
conversion means and sending it to the first adaptive filter means is provided to solve the abovementioned problems. Do.
08-05-2019
6
[0021]
Here, a microphone can be used for the first and second acoustic-electrical conversion means,
and a speaker can be used for the electro-acoustic conversion means.
The adaptive filter means comprises a filter part and an adaptive algorithm part, and the output
from the filter means is sent to the adaptive algorithm part of the first adaptive filter means, and
the filter part of the first adaptive filter means The output from is sent to the filter part and the
adaptive algorithm part of the second adaptive filter means, and the output from the second
acousto-electrical conversion means is sent to the adaptive algorithm part of the first adaptive
filter means.
[0022]
The characteristic of the filter section of the second adaptive filter means approximates the
characteristic through the electro-acoustic conversion means and the second acousto-electrical
conversion means, in particular the delay, and the coefficient of this filter part is By being set as
the coefficient of the filter means, the characteristics of the filter means also approximate the
delay.
Therefore, regarding the reference input and the residual input to the adaptive algorithm unit of
the first adaptive filter means, the output from the filter means (reference input) and the output
from the second acousto-electrical conversion means (residue) Are the same amount of delay,
and effective adaptive processing can be realized.
[0023]
FIG. 1 is a block circuit diagram showing an example of the configuration of an adaptive filter
used in a noise reduction headphone device according to an embodiment of the present
invention. FIG. 1 shows an example of a configuration used in place of the adaptive filter 16 of
the noise reduction headphone device as shown in FIG. 6 and FIG. 7 described above, and each of
the terminals 16a, 16b and 16c is the same as that in FIG. The eight adaptive filters 16
correspond to the terminals 16a, 16b and 16c, respectively.
08-05-2019
7
[0024]
In FIG. 1, the reference input x from the terminal 16a is sent to the filter section 31 of the first
adaptive filter 30 and a filter 51 for delay compensation to be described later, and the output
from the filter 51 is the first. It is sent to the adaptive algorithm unit 32 of the adaptive filter 30.
The residual e from the terminal 16c is supplied to the adaptive algorithm unit 32, and the
adaptive algorithm unit 32 modifies the filter coefficients of the filter unit 31 so as to minimize
the power of the residual e. The output y from the filter unit 31 is taken out through the terminal
16 b and sent to the filter unit 41 and the adaptive algorithm unit 42 of the second adaptive
filter 40. The output from the filter unit 41 is sent as a subtraction signal to the adder 52 and
subtracted from the residual e from the terminal 16c, and the output from the adder 52 is sent to
the adaptive algorithm unit 42. The adaptive algorithm unit 42 modifies the filter coefficients of
the filter unit 41 so as to minimize the power of the output from the adder 52. The filter
coefficients of the filter unit 41 are sent to the delay compensation filter 51 and copied to realize
the same characteristics (in particular, delay characteristics).
[0025]
The other configuration is the same as that of the noise reduction headphone device shown in
FIG. 6, FIG. 7 and the like described above, so the illustration is omitted, but the reference input
terminal 16a is provided in the vicinity of the ear when wearing the headphone The input from
the microphone, which is the first acoustic-electrical conversion means for picking up noise, is
supplied, and the output from the output terminal 16b is provided in the vicinity of the ear at the
time of wearing the headphone and the ear canal Are sent to the headphone (the speaker of the
electro-acoustic conversion means for outputting sound), and between the headphone (the
speaker) and the ear canal at the time of wearing the headphone on the residual input terminal
16c. An input is provided from a microphone, which is a second acoustic-electrical conversion
means located. Furthermore, the specific internal configuration of each of the first and second
adaptive filters 30, 40 may be the configuration as shown in FIG. 8 or 9 described above, and the
filter 51 is similar to the filter 21 of FIG. It may be a FIR filter configuration. Here, the filter unit
41 of the adaptive filter 40 and the filter 51 use the same filter structure, and by copying the
filter coefficients, the same characteristics (especially the same delay characteristics) can be
realized.
[0026]
08-05-2019
8
Next, the operation will be described with reference to FIGS. 2 and 3. The operation of this
embodiment can be roughly divided into an operation around the adaptive algorithm unit 31 of
the adaptive filter 30 and an operation around the adaptive algorithm unit 41 of the adaptive
filter 40. In the following description, it is assumed that the adaptive processing by the adaptive
algorithm unit 41 is first performed (FIG. 2), and the operation of the adaptive algorithm unit 31
is performed after this is completed (FIG. 3). However, it is also possible to advance these
operations simultaneously by setting the conditions appropriately.
[0027]
In FIGS. 2 and 3, as in the example of FIG. 7 described above, the sound insulation characteristic
of the headphone case 1 is represented by the transfer function F, provided near the ear when
the headphones are worn, and external noise (noise) is The transfer function of the microphone
14 which is the first acousto-electrical conversion means provided outside the headphone for
picking up sound is M1, and it is provided near the ear when the headphone is worn, and is used
to output the sound to the ear canal -A second acoustic-electrical conversion is provided between
the headphone 5 and the ear canal, provided that the transfer function of the headphone 5 as the
acoustic conversion means is H and is provided inside the headphone when the headphone is
worn The transfer function of the microphone 15, which is a means, is M2, and the acoustic
mixing in the headphone is represented as addition in the adder 12.
[0028]
That is, in FIGS. 2 and 3, the external noise (noise) N from the input terminal 11 is acousticallyelectrically converted by the microphone 14 of the transfer function M1 and supplied to the filter
unit 31 and the filter 51. The output from the unit 31 is electro-acoustically converted by (the
speaker of) the headphone 5 of the transfer function H, and is sent to the adder 12.
The external noise N is supplied to the adder 12 through the headphone casing 1 of the sound
insulation characteristic F, is additively mixed in the headphone internal space, reaches the
human ear, and is acoustically detected by the microphone 15 of the transfer function M2.
Electrical conversion is sent to the adaptive algorithm unit 32 and the adder 52. Note that the
voices and music signals that are intended to be played back with headphones (the user wants to
listen to) are not correlated with the noise to be reduced, and the noise reduction operation is not
affected, so the explanation is omitted. doing.
08-05-2019
9
[0029]
In FIG. 2, a kind of noise is reproduced by the headphone 5 and collected by the microphone 15,
and the same kind of noise is supplied to the filter unit 41 and the adaptive algorithm unit 42.
This kind of noise may use the noise collected by the microphone 14, but it is preferable to use
the noise generated from a white noise generator or the like. The residual obtained by
subtracting the output of the filter unit 41 from the signal collected by the microphone 15 is
input to the adaptive algorithm unit 42. The adaptive algorithm unit 42 learns to minimize the
power of the residual, and corrects the coefficients of the filter unit 41. As a result, the filter unit
41 approximates the characteristics of the headphone 5 and the microphone 15, in particular,
the time delay described above. Each filter coefficient of the filter unit 41 is copied as each filter
coefficient of the filter 51. That is, the filter unit 41 and the filter 51 have, for example, an FIR
filter configuration with the same number of taps, and copying the respective coefficients of the
filter unit 41 to the filter 51 can realize the same characteristics (delay characteristics).
[0030]
Next, in FIG. 3, the noise signal collected by the microphone 14 is filtered by the filter unit 2 to
be pseudo noise, and then sent to (the speaker of) the headphone 5 to be reproduced. The
reproduced pseudo noise acoustically cancels out the noise mixed in the headphone, and the
residual is picked up by the microphone 15 and input to the adaptive algorithm unit 32. The
residual signal is delayed by the delay of the headphone 5 and the microphone 15. However,
since this delay is realized by the filter 51, the noise signal supplied to the adaptive algorithm
unit 32 as the reference input also has the same delay. It has become.
[0031]
Therefore, when the LMS algorithm as described above is used for the adaptive algorithm unit
32, the requirement of the equation (6) is satisfied, and effective adaptive processing is
performed. Therefore, the use of a sequential adaptation algorithm requiring a small amount of
operation, such as the LMS algorithm and the learning identification method, enables practical
use with an LSI or the like, and further facilitates practical use.
[0032]
The present invention is not limited to the above embodiment. For example, the sequential
08-05-2019
10
adaptive algorithm is not limited to the LMS method, and various other sequential adaptive
algorithms can be used.
[0033]
As is apparent from the above description, according to the noise reduction headphone device of
the present invention, the input from the first acousto-electrical conversion means for picking up
external noise is adaptively filtered A first adaptive filter means for outputting the sound, an
electro-acoustic conversion means for supplying an output from the first adaptive filter means
and outputting a sound to the ear canal at the time of wearing the headphone, and the electroacoustic A second acousto-electrical conversion means located between the conversion means
and the ear canal, an output from the filter unit to which the output of the first adaptive filter
means is supplied, and the second acousto-electrical conversion means A second adaptive filter
means that adaptively filters and outputs so as to minimize the power of the residual with the
output of the first filter, and the filter coefficient of the second adaptive Filter means for filtering
the input from the electrical conversion means and sending it to (the algorithm part of) the first
adaptive filter means, the first adaptive filter means comprising the second acousto-electrical
conversion means The filter coefficients are modified to minimize the power of the input from.
Therefore, the characteristic of the filter section of the second adaptive filter means
approximates the characteristic through the electro-acoustic conversion means and the second
acousto-electrical conversion means, especially the delay, and the coefficient of this filter part is
By being set as the coefficient of the filter means, the characteristics of the filter means also
approximate the delay, and the reference input and the residual input to the adaptive algorithm
section of the first adaptive filter means are the filter The output from the means (reference
input) and the output (residue) from the second acousto-electrical conversion means become the
same delay amount, and effective adaptive processing can be realized by the sequential adaptive
algorithm.
[0034]
Therefore, the use of a sequential adaptation algorithm requiring a small amount of operation,
such as the LMS algorithm or learning identification method, enables practical use with an LSI or
the like, and further facilitates practical use.
08-05-2019
11