close

Вход

Забыли?

вход по аккаунту

?

DESCRIPTION JPH07147700

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 JPH07147700
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
howling detection apparatus and an adaptive equalizer to which the same is applied.
[0002]
2. Description of the Related Art A conventional example will be described with reference to the
drawings. Howling is a major cause of the sound quality deterioration in the loudspeakers and
the telephonic conference equipment used in theaters, event halls and other meeting rooms. FIG.
1 is a block diagram showing the operation of a one-channel loudspeaker. In A (f), the audio
signal input to the microphone 1 is amplified by the amplifier 3 to be the gain of the light
emitted from the speaker 2, and B (f) is the sound emitted from the speaker 2 sneaking into the
microphone 1 Let it be the gain of the wolf to be picked up again. At this time, the one-round
gain G (f) until the voice input to the microphone 1 is emitted from the speaker 2 and enters the
microphone 1 is expressed by G (f) = A (f) × B (f), When G (f) is larger than 1, the probability of
howling to occur is extremely high. In a conventional loudspeaker and its installation
environment, A (f) is greater than one and B (f) is less than one. And A (f) and B (f) are not
constant with frequency. Therefore, at the stage of increasing the gain of the loudspeaker, G (f)
becomes greater than 1 at a certain frequency, and howling occurs at that frequency, which
becomes an obstacle in designing the loudspeaker gain.
[0003]
15-04-2019
1
FIG. 2 shows the frequency characteristics of the open loop gain of a loudspeaker that may cause
howling. G (f) exceeds 1 at frequency f6, where it causes howling and can not increase the gain of
the loudspeaker any further. For this reason, many loudspeakers use an equalizer to flatten the
frequency characteristics to suppress howling.
[0004]
FIG. 3 is a block diagram for explaining the operation of a one-channel loudspeaker using the
equalizer 4. By using the equalizer 4 with gain E (f), the new loop gain G '(f) is G' (f) = A (f)
.times.E (f) .times.B (f) = G (f) .times.E ( f). FIG. 4 shows an equalizer gain E (f) suitable for
correcting the characteristics of FIG. 2, and FIG. 5 shows a round trip gain after correction. By
correcting the frequency characteristics with the equalizer 4, the loudspeaker can further
increase the gain over the entire frequency band, and as a result, freedom in gain design
increases.
[0005]
An adaptive equalizer is known as a device that automatically adjusts the parameters of the
equalizer while using a loudspeaker and performs howling suppression. FIG. 6 is a block diagram
for explaining the operation of the one-channel loudspeaker using the adaptive equalizer 7. The
adaptive equalizer 7 constantly monitors the input of the microphone 1 by the howling detection
device 5 during use of the loudspeaker system, and suppresses howling by changing the
parameter of the equalizer 4 when the howling occurs.
[0006]
The howling detection device 5 needs a device that determines whether or not the sound input to
the microphone 1 is causing howling, and further in which frequency band the howling is
occurring. At present, a method is employed which analyzes the frequency of the acoustic signal,
and when the output level of the highest peak is significantly larger than the average output of
the entire band, it is regarded as howling is occurring. FIG. 7 is a view for explaining a method of
detecting howling based on the peak height of the frequency analysis result. If the level of the
frequency f6, which is the highest peak, is higher than the average level of the entire band by a
15-04-2019
2
threshold or more, it is considered that howling occurs at this frequency f6.
[0007]
FIG. 8 shows the frequency characteristics of normal speech, and FIG. 9 shows the frequency
characteristics of howling is occurring. Also in the frequency characteristic of the voice signal,
when the voice is a vowel, the signal exhibits periodicity and peaks occur. Taking this into
consideration, in order to detect howling only by looking at the level difference, it is necessary to
set a large detection threshold. Setting a large detection threshold results in a problem that it can
not be detected unless howling grows sufficiently.
[0008]
The present invention provides a howling detection device that solves the problems as described
above and an adaptive equalizer that applies the same.
[0009]
A frequency analysis unit 10 is provided for frequency analysis of an input voice signal, and a
frequency characteristic signal obtained by frequency analysis is controlled from the low
frequency range to the high frequency range over the frequency band to be controlled on the
frequency axis. A differential operation unit 12 is provided to obtain a signal differentiated to a
region, and a differential value of the maximum at a higher frequency side differential value zero
in the vicinity of a frequency having a maximum differential value in the control target frequency
band A howling detection apparatus is provided that includes a howling detection unit 11 that
detects a minimal differential value equivalent to
[0010]
In the howling detection device described above, when the maximum value of the frequency
characteristic signal obtained by frequency analysis matches the howling frequency detected by
the howling detection unit 11, the howling detection unit 11 outputs this as the howling
frequency The howling detection apparatus is configured to have the following configuration.
In the above-described howling detection apparatus, the howling detection unit 11 is configured
to output the howling frequency when the local maximum value of the differential value is
15-04-2019
3
sufficiently large compared with the dispersion value of the differential value in the control
target frequency band. It also constitutes a howling detection device.
[0011]
Furthermore, in the adaptive equalizer 7 that changes parameters by the howling detection
device 5 that monitors the input of the microphone 1, the howling detection device includes a
frequency analysis unit that analyzes the frequency of the input audio signal, and is obtained by
frequency analysis. A differential operation unit for obtaining a signal obtained by differentiating
a frequency characteristic signal from a low frequency range to a high frequency range over a
control frequency band on a frequency axis is provided, in the vicinity of a frequency having a
maximum differential value in the control frequency band. The adaptive equalizer is configured
to include a howling detection unit that detects a minimum differential value equal to the
maximum differential value with respect to a higher frequency side differential value of zero.
[0012]
In the above-described adaptive equalizer, the howling detection unit outputs the howling
frequency when the maximum value of the frequency characteristic signal obtained by frequency
analysis matches the howling frequency detected by the howling detection unit. To construct an
adaptive equalizer.
In the above-described adaptive equalizer, the howling detection unit has a configuration for
outputting the howling frequency when the local maximum value of the differential value is
sufficiently large compared with the dispersion value of the differential value in the control
target frequency band. I configured an adaptive equalizer.
[0013]
Embodiments of the present invention will be described with reference to the drawings.
As described above, the frequency characteristic of FIG. 9 shows the frequency characteristic
causing howling. FIG. 10 is a diagram showing the result of differentiating the frequency
characteristic causing the howling shown in FIG. 9 with respect to the frequency. The acoustic
signal causing howling as shown in FIG. 9 has a general feature that the width of the peak is
15-04-2019
4
narrow compared to a normal audio signal. Therefore, a signal obtained by differentiating the
power frequency characteristic on the frequency axis is used as a clue to distinguish between the
frequency characteristic in the normal state and the frequency characteristic in the howling state,
and the high frequency side of the frequency where the differential signal takes the maximum
value If there is a local minimum equal to the local maximum across the zero-crossing point that
exists in, consider a narrow peak showing howling characteristics near the frequency of the zerocrossing point, and focus on this to detect howling be able to.
[0014]
The signal in the howling state is characterized in that the absolute value of the differential
signal, that is, the slope is uniform and small in a band other than the peak, as shown in FIG. If
the slope of the peak maximum, ie, the maximum value of the derivative of the power frequency
characteristic, is much larger than the dispersion over the entire band of the derivative, the audio
signal may be The number of false detections can be reduced.
[0015]
In addition, if the frequency of the howling peak obtained based on the differential value is the
same as the frequency at which the power is at a maximum, howling detection with less
erroneous detection is performed by considering that howling occurs. Can. Here, the howling
detection apparatus and the operation thereof will be specifically described. FIG. 11 shows a
howling detection apparatus according to the present invention. This howling detection
apparatus comprises a frequency analysis unit 10 for an acoustic signal, a differentiation
calculation unit 12 for frequency characteristics, and a howling detection unit 11. The howling
detection device receives the acoustic signal x (t), and outputs the howling frequency when the
howling occurs.
[0016]
As shown in FIG. 12, the frequency analysis unit 10 analyzes the frequency of the input acoustic
signal x (t) and outputs a power frequency characteristic P (f). The differential calculator 12
calculates and outputs the slope of the power frequency characteristic P (f), that is, the
differential value D (f) of the frequency characteristic. The howling detection unit 11 detects
howling based on the outputs of the frequency analysis unit 10 and the differential calculation
15-04-2019
5
unit 12. FIG. 13 and FIG. 14 are a series of diagrams showing an example of an algorithm that
the howling detection unit 11 holds.
[0017]
First, the power frequency characteristic P (f) and its differential value D (f) are read. The
frequency F1 at which P (f) is maximal is determined. The variance M of the slope D (f) is
determined. The frequency F2 at which the slope D (f) is maximal is determined. The frequency
F3 at which the slope D (f) is minimized is determined. The frequencies F2 and F3 become the
two slopes of the peak at the maximum when F3-F2> 0, so if the width (F3-F2) of the peak is
smaller than a certain value W, proceed to the next . If the peak width (F3 -F2) is greater than or
less than the value W, it is determined that howling has not occurred.
[0018]
Next, it is determined whether D (F2) at which the slope is maximum is larger than C times the
variance M of the slope. If it is larger, it proceeds to the next, and if it is smaller, it is judged that
the howling has not occurred. Although the value C is a constant larger than 1, if the value C is
too small, a normal voice signal is also erroneously detected as howling. If the value C is too
large, then the howling state can not be detected. Therefore, the value C needs to be set to an
appropriate value. Finally, it is determined whether the peak of the power maximum matches the
peak of the slope maximum. Since the peak of the slope maximum is between the frequency F2
and the frequency F3, it is judged whether or not the frequency F1 at which P (f) becomes the
maximum is in this range, and if it is not, it is judged that howling has not occurred. If it is, it is
determined that howling is occurring at the frequency F1, and the frequency F1 is output.
[0019]
As described above, according to the present invention, it is possible to more accurately and
quickly perform howling detection necessary when performing howling suppression using an
adaptive equalizer in a loud sound field. Therefore, howling suppression by the equalizer can be
made more effective.
15-04-2019
6
Документ
Категория
Без категории
Просмотров
0
Размер файла
14 Кб
Теги
description, jph07147700
1/--страниц
Пожаловаться на содержимое документа