DESCRIPTION JP2011135485

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 JP2011135485
PROBLEM TO BE SOLVED: To provide a howling suppression device which does not erroneously
determine that howling is occurring due to input of sudden sound generation. SOLUTION: A
means for performing frequency analysis of an analog input signal using a digital input signal, a
means for calculating an accumulated average value of amplitude values for each frequency from
a frequency analysis result, and occurrence of howling is determined by the accumulated average
value. Means, means for selecting a filter coefficient for reducing the amplitude value of the
corresponding frequency when it is determined that howling is occurring, a filter comprising a
plurality of digital filters according to a plurality of selected coefficients, howling A method for
suppressing howling, comprising: a digital signal processing device having means for performing
silence determination when it is determined that no occurrence occurs. [Selected figure] Figure 1
Howling suppression device
[0001]
The present invention relates to a howling suppression device used to prevent howling.
[0002]
Conventionally, graphic equalizers are known as an apparatus for suppressing howling.
By combining the graphic equalizer and the mixer, it is possible to suppress the howling by
lowering the level of the frequency band at which the howling seems to have occurred. In the
15-04-2019
1
method using such a graphic equalizer, it is difficult to accurately determine the frequency band
in which howling seems to occur, and a complicated process is required to effectively suppress
howling. Therefore, after howling occurs, it takes time to reduce the level of a specific frequency.
[0003]
There has been known a howling suppression apparatus capable of automatically performing the
determination of howling and accurate and quick howling determination and the identification
and suppression of the howling occurrence frequency by eliminating the complexity as described
above (e.g. reference).
[0004]
Patent No. 3134557 gazette
[0005]
The howling suppression apparatus described in Patent Document 1 performs frequency analysis
by Fourier transform on an input sound, determines the occurrence of howling from the level for
each frequency, and suppresses the level of the frequency causing the howling. By setting the
notch filter, howling is automatically suppressed.
As a general cause of howling, it is known that the level of the sound input from the microphone
is excessive, and the speaker and the microphone are too close and the output of the speaker is
picked up by the microphone .
The howling suppression apparatus described in Patent Document 1 compares the frequency
analysis result of the input signal with a predetermined threshold value, and howling occurs
when the amplitude value of each frequency obtained from the frequency analysis result is larger
than the threshold value. It is determined that In such a judgment method, although the
suppression effect for the general cause of howling is exhibited, the level of the input signal is
rapid, for example, when the input sound is sudden sound, or when the input sound temporarily
becomes large. Even when it changes to, it is determined that howling has occurred. That is, even
if howling does not actually occur, the occurrence of howling is erroneously determined. If the
notch filter is set based on such an erroneous determination of howling occurrence, the level of
the frequency component which does not need to be suppressed originally will be suppressed.
15-04-2019
2
[0006]
In order to avoid such an erroneous determination, a plurality of filters may be further set, but
unnecessary filter processing is increased, which results in deterioration of the output sound
from the speaker.
[0007]
The present invention has been made in view of the above problems, and a howling suppression
device that operates without erroneously determining the occurrence of howling even if the input
of sudden sound generation or the input sound temporarily increases. Intended to provide.
[0008]
The present invention comprises a microphone, an A / D converter for converting an analog
input signal from the microphone into a first digital signal, and a digital for performing a
predetermined signal processing on the first digital signal to generate a second digital signal. The
digital signal processing apparatus includes a signal processing device, a D / A converter that
converts a second digital signal generated by the digital signal processing device into an analog
output signal, and a speaker that outputs the analog output signal. Frequency analysis means for
performing frequency analysis of digital signal, amplitude value for each frequency of analog
input signal obtained by frequency analysis, and average value of amplitude values for each
frequency obtained by past frequency analysis The accumulated average calculation means for
calculating the accumulated average value for each and the occurrence of howling by comparing
each calculated accumulated average value with a predetermined threshold value When it is
determined that the howling is occurring by the determining means and the determining means,
the selection of a plurality of coefficients used in the filter for reducing the amplitude value of the
frequency corresponding to the accumulated average value larger than the predetermined
threshold is selected Means that the analog input signal is silent when it is determined that the
howling does not occur by the determination means, the filter means including the plurality of
digital filters in which the plurality of coefficients selected by the coefficient selection means are
set, and the determination means It is most important to have silence detection means for
determining, and filter initialization means for invalidating the function of the filter provided in
the filter means when the analog input signal is determined to be silence by the silence detection
means. It is a feature.
[0009]
The present invention is characterized in that, in the howling suppression apparatus, the filter
initialization means invalidates at least one digital filter among a plurality of digital filters
provided in the filter means.
15-04-2019
3
[0010]
The present invention is characterized in that, in the howling suppression device, the filter
initialization means invalidates the digital filter when the time determined to be silence in the
silence detection means reaches a predetermined length. .
[0011]
In the howling suppression apparatus according to the present invention, the accumulated
average calculating means may include an amplitude value obtained by individually weighting an
amplitude value for each frequency of an analog input signal component obtained by frequency
analysis, and a frequency obtained by past frequency analysis. It is characterized in that the
accumulated average value for each frequency is calculated by adding the average value of each
amplitude value.
[0012]
In the howling suppression apparatus according to the present invention, the accumulated
average calculation means may use an average value of an amplitude value for each frequency of
an analog input signal component obtained by frequency analysis and an average value of
amplitude values for each frequency obtained by past frequency analysis. A weighted average
value is added to calculate an accumulated average value for each frequency.
[0013]
In the howling suppression apparatus according to the present invention, the howling
determination unit determines that howling is occurring when it is determined that each
accumulated average value calculated in a plurality of successive comparisons is larger than a
predetermined threshold value. It is characterized by
[0014]
The present invention is characterized in that, in the howling suppression apparatus, the
coefficient selection means selects a coefficient corresponding to any one of a notch filter, a dip
filter, and an all-pass filter.
[0015]
15-04-2019
4
According to the present invention, even when a sudden voice is input to the microphone, the
howling determination means is less likely to make an erroneous determination, and unnecessary
filter processing is not performed in the filter processing means. It becomes possible to reduce
the deterioration of the sound quality.
Also, by setting the filter processing once set again under a predetermined condition, only the
necessary filter processing can be performed, and unnecessary filter processing can be omitted,
so that unnecessary deterioration of the sound quality can be reduced. Will be able to
[0016]
It is a functional block diagram showing an example of a howling suppression device concerning
the present invention.
It is a flowchart which shows the example of a process of the howling suppression apparatus
which concerns on this invention.
[0017]
Hereinafter, an embodiment of a howling suppression device according to the present invention
will be described.
FIG. 1 is an example of a functional block diagram of a howling suppression device according to
the present invention.
In FIG. 1, the howling suppression device 100 converts a microphone (MIC) 40 and an audio
signal (analog input signal) picked up by the microphone 40 and amplified by an amplifier (not
shown) into a digital signal (first digital signal) A / Digital signal processing means (DSP) 10 that
executes predetermined signal processing on the first digital signal output from the D converter
20, the A / D converter 20, and outputs the second digital signal; And D / A converter 30 for
converting the second digital signal into an analog output signal, and speaker 50 for outputting
the analog output signal converted by D / A converter 30.
15-04-2019
5
Although not shown, an amplifier for amplifying an analog input signal may be provided at the
front stage of the A / D converter 20, and an amplifier may be provided at the rear stage of the D
/ A converter 30.
[0018]
The DSP 10 performs predetermined filtering on the input first digital signal.
The filtering process performed by the DSP 10 is performed by changing the setting of
parameters (coefficients) used for a program (software) executable in the DSP 10, and this
filtering process is performed on a low pass filter, high pass filter, band pass in the analog filter
circuit. Signal processing equivalent to various filters such as a filter and a notch filter can be
performed.
Each processing unit included in the DSP 10 is executed by information processing by software
executable by the DSP using hardware resources included in the DSP.
[0019]
In FIG. 1, the DSP 10 includes a frequency analysis unit 11, an accumulated average calculation
unit 12, a howling determination unit 13, a coefficient selection unit 14, a silence detection unit
15, a filter initialization unit 16, and a filter processing unit 17.
[0020]
The frequency analysis unit 11 performs spectrum analysis by discrete Fourier transform
processing on the first digital signal input to the DSP 10, and stores amplitude levels (values) of
respective frequencies in a storage area (not shown).
[0021]
The accumulated average calculation unit 12 is an average value of the amplitude level of each
frequency obtained by the frequency analysis unit 11 and the amplitude level of each frequency
obtained by the past spectrum analysis, and is stored in a storage area (not shown) A process of
adding a stored average value to calculate a new stored average value and updating the stored
average value stored in a storage area (not shown) is performed.
15-04-2019
6
[0022]
The howling determination unit 13 determines that the howling is caused by the input signal
when there is a value larger than the preset first threshold value among the accumulated average
values of the respective frequencies calculated by the accumulated average calculation unit 12. If
the stored average value of each frequency calculated by the stored average calculation unit 12
is smaller than the first threshold, it is determined that howling has not occurred, and data
indicating the determination result in a storage area (not shown) Perform processing to store
The first threshold is stored in a storage area of storage means (not shown) provided in the
howling suppression apparatus 100.
[0023]
When the howling determination unit 13 determines that the howling is occurring, the
coefficient selection unit 14 performs a process of selecting a filter coefficient for suppressing
the amplitude level of the frequency exceeding the threshold.
The filter coefficients are stored for each frequency band in a storage area (not shown).
For example, the sampling frequency in the A / D converter 20 is 48 kHz, and a value obtained
by dividing this by 256 (about 187.5 Hz) divides one frequency band of the input signal every
187.5 Hz as one bandwidth. The filter coefficients corresponding to each of them are made into a
table in advance and stored in a storage area provided in the DSP 10.
That is, in the coefficient selection unit 14, the howling determination unit 13 performs a process
of reading from the storage area (not shown) the filter coefficient corresponding to the frequency
band in which the frequency determined to be howling is generated.
[0024]
15-04-2019
7
The filter processing unit 17 includes a plurality of filters continuously connected in multiple
stages so that a plurality of filter processes can be performed.
The filter coefficient selected by the coefficient selection unit 14 is set in each filter process, and
the filter process according to the filter coefficient is performed on the first digital signal. Since
the filter processing possessed by the filter processing unit 17 is continuously connected, the
filter processing in which the coefficient for disabling the filter function is set and the filter
processing in which the coefficient for enabling the filter function is set simultaneously exist It
has become.
[0025]
When the howling determination unit 13 determines that the howling does not occur, the silence
detection unit 15 counts the number of times it is determined that the howling does not occur
continuously and stores the count in a storage area (not shown). Do.
[0026]
The filter initialization unit 16 initializes at least one of the filter coefficients set in the filter
processing unit 17 when the value counted up by the silence detection unit 15 exceeds a
predetermined second threshold. Process to reduce the number of filters.
[0027]
As described above, according to the howling suppression apparatus according to the present
embodiment, howling occurring at a specific frequency is determined using the accumulated
average value of the amplitude level for each frequency of the input signal to generate howling.
Howling can be suppressed by providing a filter that suppresses the frequency.
Also, among the set filters, when the time during which the input signal is silent (the state where
howling does not occur) exceeds a predetermined time, at least one of the previously set filters is
initialized. Thus, unnecessary filtering can be reduced, thereby preventing deterioration of the
sound quality of the output signal from the speaker.
[0028]
15-04-2019
8
Next, a howling suppression method using the howling suppression apparatus according to the
present embodiment will be described.
FIG. 2 is a flowchart showing an example of processing of the howling suppression method
according to the present invention. In FIG. 2, the steps of each process are represented as S10,
S11,. First, based on the first digital signal digitally converted by the A / D converter 20, the
frequency analysis unit 11 performs spectrum analysis. This spectrum analysis is performed by
discrete Fourier transform processing. The amplitude level for each frequency component
obtained by the discrete Fourier transform process is stored for each frequency component in a
storage area (not shown) of the DSP 10 (S10).
[0029]
Next, in the accumulated average calculation unit 12, an average value for each frequency
component of the amplitude level obtained in step S10 and for each amplitude component
obtained by the past frequency analysis processing already calculated (the DSP 10 A storage
average value stored in a storage area (not shown) is added to calculate a new storage average
value, and the value of the predetermined storage area is updated (S11).
[0030]
Next, the howling determination unit 13 determines whether or not there is a value larger than a
first predetermined threshold value among the new accumulated average values (S12).
When there is a value larger than the first threshold (Y in S12), the filter used in the filter for
suppressing the amplitude level of the frequency band including the frequency having the value
larger than the first threshold in the coefficient selection unit 14 A coefficient is selected (S13).
[0031]
Next, the filter coefficient selected in the coefficient selection unit 14 is set to a digital filter in
which no coefficient is set among the plurality of digital filters possessed by the filter processing
unit 17, and the process ends (S14). In this way, it is possible to determine the frequency at
15-04-2019
9
which howling occurs and to automatically set the filter that suppresses the level of the
frequency.
[0032]
In addition, in processing step S12, when there is no value larger than the first predetermined
threshold value among the new accumulated average values (N in S12), the silence detection unit
15 determines that the input signal is silence, It is determined whether the number of times
determined to be silent in succession exceeds the second threshold (S15). If the second threshold
is exceeded in processing step S15 (Y in S15), processing to invalidate at least one of the digital
filters set in the filter processing unit 17 by the filter initialization unit 16 is performed Then, the
process ends (S16). In the filter invalidation processing (S16) in the filter initialization unit 16, at
least one of the filter coefficients set in the plurality of filters included in the filter processing
unit 17 is set as a filter coefficient that invalidates the filter function. Processing. If the threshold
is not exceeded in processing step S15 (N of S15), the processing ends.
[0033]
As described above, according to the howling suppression method according to the present
embodiment, howling occurring at a specific frequency is determined using the accumulated
average value of the amplitude level for each frequency of the analog input signal to generate
howling. Howling can be suppressed by providing a filter that suppresses the frequency being
used. In addition, when the state of silence of the input signal (state in which howling does not
occur) exceeds a predetermined time, unnecessary filter processing is reduced by initializing at
least one of the previously set filters. This can prevent deterioration of the output signal from the
speaker.
[0034]
In the howling suppression method according to the present embodiment, the accumulated
average calculation process (S11) weights each level of each frequency obtained by the spectrum
analysis, and then weights the accumulated average calculation process (S11). S12) may be
performed. Alternatively, the storage average value may be updated by adding new spectrum
analysis results after performing predetermined weighting on the past storage average value
stored in a storage area (not shown). As described above, by weighting the level or accumulated
15-04-2019
10
average value at each frequency, it is possible to reduce erroneous determination of howling
occurrence due to sudden sound generation or rapid volume change, and to perform more
accurate howling suppression processing. Become.
[0035]
In the howling suppression method according to the present embodiment, the howling
determination process (S12) counts the number of times the calculation result by the storage
average calculation process (S11) continuously exceeds the threshold, and the number is
predetermined. It may be determined that howling has occurred when another threshold of is
exceeded. As described above, by providing a plurality of threshold values in the determination
process (S12), it is possible to more accurately determine the occurrence of howling.
[0036]
Further, in the howling suppression method according to the present embodiment, the coefficient
selection process (S13) includes a notch filter, a dip filter, or the frequency band for suppressing
the level of the band including the frequency determined to be the howling. The coefficients of
the all-pass filter may be selected to change the phase of. A plurality of notch filters, dip filters or
all-pass filters can be set by sequentially applying the coefficients to a plurality of digital filters
included in the filter processing unit 17. As a result, it is possible to suppress howling with a dip
filter with little attenuation, and it is possible to reduce change in the frequency characteristic of
the input signal, and to perform more accurate howling suppression while preventing
deterioration in sound quality.
[0037]
10 DSP 20 A / D converter 30 D / A converter 40 microphone 50 speaker
15-04-2019
11