close

Вход

Забыли?

вход по аккаунту

?

DESCRIPTION JPH01282954

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 JPH01282954
[0001]
[Summary] The purpose of the present invention is to provide a howling suppressor for a
communication system in which communication is performed using a microphone and a speaker,
which has good communication quality, can cope with sudden changes in voice, and can reduce
the circuit size. The frequency where howling has occurred is detected, a delay equivalent to
rotating the frequency component by 180 degrees is given to the transmission signal, and then
the power level of the frequency component is detected before and after the delay section, and
the ratio is The looping gain of the howling is estimated, and the inverse of the gain is multiplied
by the delayed output signal, and then the feedback signal is configured to be subtracted from
the original signal to cancel the howling component. BACKGROUND OF THE INVENTION 1. Field
of the Invention The present invention relates to a howling suppression apparatus for a
communication system such as a teleconference system and a car telephone system, which
makes a call using a microphone and a speaker. [Prior Art] A conventional example will be
described below with reference to the drawings. Fig. 5 is a block diagram of a conventional
example, in which (A) is an echo suppressor system, (B) is an echo canceller system, and (C) is a
band pass filter bank system, both showing one station. There is also the other side
corresponding to this as. The echo suppressor system (A) turns on the switch 21 to enable signal
transmission from the microphone 1 when the signal to the speaker 2 detected by the detector
22 of the echo suppressor 25 is almost O, and the comparator 20 Then, compare the power level
of the signal from the microphone 1 and the power level of the signal to the speaker 2, and if the
comparison value is greater than 6 dB, for example, turn on the switch 21 to enable signal
transmission from the microphone 1, In this case, by turning off the switch 21, any line including
the other station is disconnected to suppress the howling. In the echo canceller method (B), the
pseudo echo generator 24 generates a pseudo echo equivalent to the echo, and the pseudo echo
is added to the microphone 1 from the speaker 2 by the adder 23, which is a difference from the
16-04-2019
1
echo Pull. At this time, the tap coefficient of the pseudo echo generator 24 is subtracted and
adaptive control is performed so that the remaining error approaches O, thereby suppressing
howling without disconnecting the line. In the band pass filter bank system of (C), a large number
of band pass filters having different band pass frequencies are provided in the signal
transmission line from the microphone 1 as 30.31, 32 ° ... 3 N, and the output power of each
band pass filter The power detection and comparison section 40 finds and compares the
frequency causing the howling, and the switch 41 turns off the switch of the output of the band
pass filter that passes the frequency causing the howling. It is suppressing.
[Problem to be Solved by the Invention] However, in the echo suppressor system, one line of the
reciprocation is disconnected during the call to block the echo, but the two-header disconnection
occurs because the two-way simultaneous call can not be performed. Problems occur. In the echo
canceller system, the signal from the microphone 1 and the signal to the speaker 2 are constantly
measured during a call, the path characteristic is identified and the pseudo echo is generated, so
that the rapid change of the echo path characteristic is followed. There is a risk that this will not
be possible, and the circuit scale also becomes large, and there is a problem that, for example, a
dedicated LSI 10 chip is required. In the band pass filter bank system, many band pass filters are
required, and even if a digital signal processor (hereinafter referred to as DSP) is used, about 3 to
4 chips are required and the circuit scale becomes large. An object of the present invention is to
provide a howling suppression apparatus which has good communication quality, can cope with
abrupt changes in echo path characteristics, and can reduce the circuit size. [Means for Solving
the Problems] FIG. 1 is a block diagram of the principle of the present invention. As shown in FIG.
1, the howling frequency and level detection unit 3 detects the howling frequency in which the
transmission signal from the microphone 1 passes through the speaker 2 through the speaker 2,
and the frequency is calculated as a frequency delay amount calculation unit 4 to calculate the
delay amount equivalent to rotating the component of the frequency by 180 degrees, and let the
delay amount of the delay unit 5 that delays the transmission signal from the microphone 1 be
the calculated delay amount. The howling frequency and level detection unit 3 detects the power
of components of the frequency before and after the delay unit 5, and the inverse of the
estimated value of howlin group gain is determined from the ratio. Then, the obtained inverse
number is multiplied by the multiplier 6 to the transmission signal delayed by the delay unit 5,
and the multiplied signal is subtracted from the transmission signal from the microphone 1 by
the addition unit 7. [Operation] According to the present invention, the howling frequency is
detected by the howling frequency and level detection unit 3, and the delay amount
corresponding to the rotation of the frequency component by 180 degrees is calculated by the
frequency delay amount calculation unit 4. The delay unit 5 delays the component of the
frequency of the transmission signal from the microphone 1 by a delay amount equivalent to
180 degrees rotation. Then, the howling frequency and level detection unit 3 determine the
reciprocal of the estimated value of howlin group gain, and the obtained reciprocal is multiplied
by the multiplier 6 by the transmission signal delayed by the delay unit 5 and the multiplied
16-04-2019
2
signal is obtained. By subtracting from the transmission signal from the microphone 1 in the
addition unit 7, the component of the howling frequency is canceled and the howling is
suppressed.
That is, in the present invention, since there is no disconnection of the line, communication
quality is not deteriorated, and since it is not identified as in the echo canceller system and echo
is not canceled gradually, even abrupt changes in echo bus characteristics are possible.
Correspondence is possible, and the circuit scale can be reduced. Also, the circuit scale can be
reduced as compared with the band pass filter bank method. The invention will now be described
by way of example with reference to the drawings. FIG. 2 is a block diagram of the embodiment
of the present invention, and FIG. 3 is a flowchart of the program of the embodiment of the
present invention. In FIG. 2, the signal from the speaker 1 is converted to digital by the A / D
converter 10, the howling is suppressed by the DSP 12 through the buffer 11, and the D / A
converter 14 through the buffer 13. Are converted to analog and transmitted to the other side.
The signal from the other party is converted to digital by the A / D converter 15, and the − unit
is input to the DSP 12 through the buffer 16, and the other is analog by the D / A converter 18
through the buffer 17. And output a signal from the speaker 2. The DSP 12 has a multiplication
unit for performing filter operation, power calculation, division, etc. in addition to the same
functions as the normal processor as the operation unit etc., and a ROM storing the program
shown in FIG. , And a RAM, and the DSP 12 performs the operation shown in the principle block
diagram of FIG. Next, the howling suppression operation of the DSP 12 will be described using
the program of FIG. 3 and FIG. In step 1 of FIG. 3, a signal corresponding to the point A of the
output of the microphone 1 of FIG. 1 and the point B of the input to the speaker 2 is processed.
Next, in step 2, for example, the power of the signal of one frame per second at point A is
calculated for three frames to obtain an average value, which is taken as the power PA at point A,
and one of these three frames The number of zero crossings per frame ZA-1, ZA-2, ZA-3 is
calculated. Next, in step 3, the power PA at point A is compared with a threshold TPA of, for
example, -33 dBm for detecting howling, and if it is larger than the threshold TPA, the process
proceeds to step 4; The power of the signal of one frame is calculated for three frames to obtain
an average value, which is taken as the power PB of point B, and the number of zero crossings
per frame of this three frames ZB-1, ZB-2, 2B Calculate -3. Next, in step 5, the power PB at point
B is compared with the dark value TPB of -33 dBm power, for example, to detect howling, and if
it is larger than the threshold TPB, the process proceeds to step 6 for 3 frames Difference
between the number of zero crossings at point A and the number of zero crossings at point B
(ZA-1)-(ZB-1), (ZA-2)-(ZB-2), (ZA-3)-(ZB-) 3) If both the frequency of the signal of point A and the
frequency of the signal of point B are almost equal and howling is recognized as equal to or less
than the threshold value TZ of 2, for example, step 7 Howling frequency (ZA + ZB) / 2T (where T
is the number of zero crossings ZA (average value of ZA-1, 2A-2, ZA-3), ZB (average value of ZB1, ZB-2 B ZB-3) Determine the frame period, and step 180 Determine T / (ZA + ZB) as the
appropriate delay value, and send it to step 14.
16-04-2019
3
At this time, in order to obtain the front and back loop gains of the delay unit 5 of FIG. 1 in step
1, the signal at point A and the signal at point A before time T / (ZA + ZB) passing through the
delay unit 5 are obtained. Buffer a signal of 0 point and pass a band pass filter having a peak at
frequency (ZA + ZB) / 2T in step 9.11. At step 10.12, output power PAF of each band pass filter,
The PCF is calculated, and in step 13, this ratio PAF / PCF is calculated, and the calculated
multiplier α is sent to step 14. In step 14, a signal obtained by multiplying A (n−T / (ZA + ZB))
by a multiplier α of a signal delayed by 180 degrees from the signal A (n) at point A is obtained
from the signal A (n) at point A Subtract and output at step 16. In this way, the howling signal is
almost canceled and immediately suppressed. If the dark value condition is not satisfied in step 3
and 5.6, it is determined that there is no howling, the multiplier α is set to 0 in step 15, and the
signal A (n) of point A is input to step 16 in step 14. Output as output. In digital signal
processing, (n−T / (ZA + ZB) must be a discrete value, and the value of T / (ZA + ZB) is, for
example, 3 KHz of the howling frequency and 8 KH2 of the sampling frequency, T / ( If the value
of ZA + ZB) becomes 8/3 = 2.67, the accuracy can not be obtained, so by oversampling, for
example, if 24KH2 is sampled, 24/3 = 8 and sufficient accuracy can be obtained. . In this way, if
howling occurs, it can be immediately suppressed, so there is no disconnection of the line as in
the echo suppressor system, without deteriorating the communication quality, and as in the echo
canceler system, it is identified and gradually echoed. Because it does not cancel out, it is possible
to cope with sudden changes in the corpus characteristics and to make the circuit scale smaller.
Also, the circuit scale can be reduced as compared with the band pass filter bank method.
Therefore, the howling suppressor in this case can be realized by a DSPI chip. Next, when the
looping gain of the howling is 2 or less, the howling can be suppressed even if the multiplier α =
1 if the signal delayed 180 degrees of the howling frequency component is canceled. An
embodiment in this case is shown in FIG. (A) is a principle block diagram corresponding to FIG. 1,
(B) is a flowchart of a program corresponding to FIG. Points different from FIGS. 1 and 3 in this
case will be described below.
In FIG. 4 (A), the difference from FIG. 1 is that it is not necessary to detect the loop gain by
detecting the levels before and after the delay unit 5 and to set the multiplier α = 1, except for
the multiplier 6 The point is that the howling frequency and level detection unit 3 is a howling
frequency detection unit 3 ′. Also, FIG. 4 (B) differs from FIG. 3 in that it is not necessary to
obtain the multiplier α, so that buffering for obtaining a 0 point signal is excluded from step 1 of
FIG. Since it is sufficient to obtain the delay value T / (ZA + ZB) equivalent to 180 degrees of this
frequency in step 8 without obtaining the howling frequency in step 7 of the figure, except for
step 7, those with howling are step 14 The signal A (n−T / (ZA + ZB)) delayed by 180 degrees
from the signal A (n) at point A is set as the subtraction output from the signal A (n) at point A,
step 3, 5.6 If the dark value condition is not satisfied, howling is not performed, and the signal A
(n) at point A shown in step 15 ° is to be output. As described above in detail, according to the
16-04-2019
4
present invention, it is possible to cope with abrupt changes in echo path characteristics without
deteriorating communication quality, and to obtain a howling suppressor with a small circuit
scale.
[0002]
Brief description of the drawings
[0003]
1 is a block diagram of the principle of the present invention, FIG. 2 is a block diagram of an
embodiment of the present invention, FIG. 3 is a flowchart of a program of the embodiment of
the present invention, and FIG. 4 is another embodiment of the present invention. FIG. 5 is a
block diagram of a conventional example.
In the figure, 1 is a microphone, 2 is a speaker, 3 is a howling frequency and level detection unit,
3 ° is a howling frequency detection unit, 4 is a frequency delay calculation unit, 5 is a delay
unit, 6 is a multiplier, 7.23 is Adder, 10.15 A / D converter, 11.13.16.17 buffer, 12 digital signal
processor, 14.18 D / A converter, 20 comparator, 21 switch, 22 Denotes a detector, 24 denotes a
false reconnaissance echo generation unit, 25 denotes an echo suppressor, 30.31.32.3 N denotes
a band pass filter, 40 denotes a power detection comparison unit, and 41 denotes a switch.
Invented / 1 original Chiri 7 °] · Tsukua No. 1 g /--\ \--/ 茅 5 Figure Uq 2-)
16-04-2019
5
Документ
Категория
Без категории
Просмотров
0
Размер файла
15 Кб
Теги
description, jph01282954
1/--страниц
Пожаловаться на содержимое документа