JP2005110127A - Wind noise detecting device and video camera with wind noise detecting device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a wind noise detecting signal having high accuracy in detection by using not only a difference signal of two or more sound collecting means but a summation signal thereof as an information source for detecting the wind noise, and reducing incorrect determination even when the sound pressure level of the acoustic wave is high. <P>SOLUTION: A difference signal (L-R) is generated by a difference signal generator 8 from signals (L), (R) detected by two microphones, and a summation signal (L+R) is generated by a summation signal generator 9. Each signal is passed through low-pass filters 10, 12 and changed into each modulus by modulus detectors 11, 13. In a wind noise detecting signal generator 14, the modulus of the summation signal (L+R) is subtracted from the modulus of the difference signal (L-R) and the result is generated to a system control unit 7. In this case, 0 is generated if the resulted value is smaller than 0. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a wind noise detection technique for removing wind noise.

  In a video camera device, it is known that a diaphragm (diaphragm) of a microphone vibrates due to wind pressure and is detected as a wind noise signal. Several techniques for improving wind noise have already been proposed for this problem (for example, Patent Document 1).

  Hereinafter, this wind noise reduction technique will be described with reference to FIG.

  In FIG. 1, 10 is an Lch microphone that collects an Lch audio signal, 20 is an Rch microphone that collects an Rch audio signal, 30 is an Lch first stage microphone amplifier that amplifies the Lch microphone output signal, and 40 is an amplifier of the Rch microphone output signal. Rch first stage microphone amplifier 50, Lch matrix delay device 50 for delaying Lch signal, 60 Rch matrix delay device for delaying Rch signal, 70 Lch matrix attenuator for attenuating Lch signal, 80 Rch for attenuating Rch signal Matrix attenuator.

  Reference numeral 90 denotes an Lch matrix subtractor that subtracts the Rch signal that has passed through the Rch matrix delay unit 60 and the Rch matrix attenuator 80 from the output signal of the Lch first stage microphone amplifier. Reference numeral 100 denotes an Lch delay unit 50 that passes through the Lch attenuator 70. Rch subtractor that subtracts the Lch signal from the output signal of the Rch first stage microphone amplifier, 110 an Lch equalizer that corrects the frequency characteristic of the Lch subtractor output, 120 an Rch that corrects the frequency characteristic of the Rch subtractor output, etc. Is a generator.

  130 is an Lch output unit that outputs an Lch signal after signal processing, 140 is an Rch output unit that outputs an Rch signal after signal processing, 150 is an Lch HPF that extracts a high-frequency signal output from the Lch equalizer, and 160 is an Rch equalization RchHPF for extracting a high-frequency signal from the output of the generator, 170 is an Lch equalizer output attenuator for attenuating the output of the Lch equalizer 110, 180 is an LchHPF output attenuator for attenuating the LchHPF output, 190 is the Rch equalizer 120 Rch equalizer output attenuator for attenuating the output, 200 RchHPF output attenuator for attenuating the RchHPF output, 210 Lch adder for adding the output of the Lch equalizer output attenuator and the output of the LchHPF output attenuator, 220 Rch adder 230 adds the output of the Rch equalizer output attenuator and the output of the RchHPF output attenuator, 230 A difference signal generator that generates a difference signal (LR) from the Lch microphone output and the Rch microphone output, 240 is an LPF that extracts a low-frequency signal output from the difference signal generator 230, and 250 detects a peak value of the LPF output. However, this is a peak detection unit.

  In addition, in each part demonstrated so far, the part which is not related to wind noise reduction on a stationery is omitted, and operation | movement description is given below.

  The sound signals output from the two microphones 10 and 20 arranged in the vicinity have a strong correlation particularly in a low frequency region where the wavelength becomes long because the sound wave is a wave in the air. In other words, the signals output by the two microphones are almost similar.

  On the other hand, wind noise is generated when the wind randomly hits the microphone diaphragm, and has a characteristic that the correlation between the two microphone outputs is extremely thin and the level is high in the low frequency region.

Utilizing this characteristic, two microphone outputs are input to the difference signal generator 230 to generate a difference signal (LR). The difference signal (LR) is low in the case of a sound wave whose in-phase signal is a main component, and the level is high in the case of wind noise with low correlation. Thereafter, a low-frequency region, which is a main band of wind noise, is extracted by the LPF 240, and the value is detected by the peak detection unit 250, thereby generating a wind noise detection signal. Depending on the level of the wind noise detection signal, the attenuation amounts of the Lch equalizer output attenuator 170 (Rch equalizer output attenuator 190) and LchHPF output attenuator 180 (RchHPF output attenuator 200) are controlled, and the respective output signals are controlled. Addition by the Lch adder 210 (Rch adder 220) lowers the level of the low frequency region, which is the main band of wind noise, and reduces wind noise.
JP-A-5-7392 (page 6, FIG. 1)

  However, the above example has the following problems.

  In the case of sound waves, the output signals of the two microphones are highly correlated and have a very close level and phase relationship, but are not exactly the same, so the difference signal generator provides a certain level of difference signal (LR). May be output. In particular, when the sound source deviates greatly in the horizontal direction instead of in front of the microphone, and the sound pressure level of the sound wave applied to the microphone increases, the level of the difference signal (LR) also increases, and wind noise In some cases, the level of the low frequency region of the audio signal is attenuated even though sound waves are collected by erroneous detection.

  Accordingly, the present invention aims to provide a technique for preventing erroneous detection of wind noise even when the sound pressure level of sound waves applied to a microphone is very high, and generating a highly accurate wind noise detection signal. It is.

In order to solve this problem, for example, the wind noise detection apparatus of the present invention has the following configuration. That is,
A wind noise detecting device comprising a plurality of sound collecting means and detecting wind noise by signals from the plurality of sound collecting means,
Difference signal generating means for generating a difference signal from signals obtained by the plurality of sound collecting means;
Sum signal generating means for generating a sum signal from signals obtained by the plurality of sound collecting means;
First absolute value conversion means for converting the difference signal generated by the difference signal generation means into an absolute value signal;
Second absolute value conversion means for converting the sum signal generated by the sum signal generation means into an absolute value signal;
Wind noise detection means for obtaining a difference between the difference signal obtained by the conversion by the first and second absolute value conversion means and the sum signal and outputting it as a wind noise detection signal.

  According to the present invention, not only a difference signal of a plurality of sound collecting means but also a sum signal thereof is used as an information source for detecting wind noise, thereby reducing erroneous determination even when the sound pressure level of a sound wave is high. Thus, it is possible to obtain a wind noise detection signal with high detection accuracy.

  Embodiments according to the present invention will be described below in detail with reference to the accompanying drawings.

  FIG. 2 is a block diagram of the entire video camera in the embodiment.

  In the figure, 101 is a lens unit that captures an optical image, 102 is a CCD unit that forms an image of light incident through the lens unit 1 and converts it into an electrical signal of an image, and 103 is an electrical signal from the CCD unit 2. A camera signal processing unit for processing, 104 a microphone unit for collecting sounds, 105 an audio signal processing unit for processing audio signals, 106 a video signal processing unit for processing video signals, and 107 for controlling the entire video camera device. A system control unit 108, an input / output unit 108 for inputting / outputting signals with the video camera device, 109 a recording / reproduction processing unit for recording and reproducing data on a recording medium, and 110 for recording / reproducing data on / from the recording medium A mechanical unit for reproducing data, 111 a display signal processing unit for processing display data, and 112 a display unit for displaying data from the display signal processing unit (example) If a liquid crystal display unit).

  FIG. 3 is a detailed block diagram of the periphery of the microphone unit 104, the audio signal processing unit 105, and the system control unit 107 in FIG.

  In the figure, 1 is an Lch microphone that collects an Lch audio signal, 2 is an Lch first-stage amplifier that amplifies the output signal from the Lch microphone 1, and 3 is a control of the low frequency region level of the Lch audio signal from the wind noise detection result. LchHPF, 4 is an Rch microphone that collects the Rch audio signal, 5 is an Rch first-stage amplifier that amplifies the output signal from the Rch microphone 4, and 6 is a control of the low frequency region level of the Rch audio signal from the wind noise detection result. RchHPF, 7 is a system control unit for controlling the cutoff frequency of LchHPF3 and RchHPF6 from the output level of the wind noise detection signal generator 14, and 8 is a difference signal (L--) from each output of the Lch first stage amplifier 2 and Rch first stage amplifier 5. R), a difference signal generator 9 generates a sum signal (L + R) from the outputs of the Lch first stage amplifier 2 and the Rch first stage amplifier 5 10 is a difference signal LPF that extracts a low frequency component that is the main component of the wind in the difference signal, 11 is a difference signal absolute value detector that takes the absolute value of the difference signal LPF output, and 12 is a sum signal. A sum signal LPF for extracting a low frequency component which is a main component of the wind in the inside, 13 is a sum signal absolute value detector for taking the absolute value of the sum signal LPF output, and 14 is a wind noise detection signal generator for generating a wind noise detection signal. It is a vessel.

  The operation of each block will be described below with reference to FIG.

  The Lch microphone 1 and the Rch microphone 4 are arranged close to each other. In the case of sound waves, particularly in the low frequency region, the output of the Lch microphone 1 and the output of the Rch microphone 4 are similar signals having substantially the same phase. On the other hand, in the case of wind, since the change in wind pressure applied to the microphone diaphragm has random characteristics, the correlation between the output of the Lch microphone 1 and the output of the Rch microphone 4 is small, that is, a phase. appear.

  The output of the Lch microphone 1 and the output of the Rch microphone 4 are amplified and output as outputs of the Lch first stage amplifier 2 and the Rch first stage amplifier 5.

  Here, the output signals are expressed as (L) and (R), respectively.

  Signals (L) and (R) are input to LchHPF3 and RchHPF6 in the next stage, respectively. LchHPF3 and RchHPF6 are configured such that the cut-off frequency can be continuously varied by a control signal from the system control unit 7.

  The signals (L) and (R) are input to the difference signal generator 8 and the sum signal generator 9. Here, the signals (L) and (R) are converted into the signal (LR) by the difference signal generator 8 and converted into the signal (L + R) by the sum signal generator 9.

  From the generated signal (LR), the difference signal LPF 10 extracts the low frequency side, which is the main band of wind noise, and then the difference signal absolute delay detector 11 calculates the absolute value. The signal (L + R) produced in the same manner extracts the low frequency side, which is the main band of wind noise, from the sum signal LPF 12, and then calculates the absolute value by the sum signal absolute delay detector 13. The calculated detection values are respectively input to the wind noise detection signal generator 14, and a result obtained by subtracting the absolute value of the signal (L + R) from the absolute value of the signal (LR) is output. The wind noise detection signal generator 14 is a subtracter that is 0 and is limited, and has a minimum value of 0. That is, 0 is output when “absolute value (LR) −absolute value (L + R) <0”.

  The reason for this configuration will be briefly described.

  Assume that a sound wave is applied to the microphone and the applied level is not so high. In this case, since the detection signals (L) and (R) at the respective microphones are substantially similar signals, the level of the signal (LR) does not become a very large value. Accordingly, since the absolute value (L + R) having a positive value is originally subtracted from the absolute value (LR) having a small value, the output signal of the wind noise detection signal generator 14 outputs 0 or a small value. Promised to do.

  However, even if the sound pressure level is relatively high, for example, when the sound pressure level is relatively high, more specifically, a place where the sound source exists near the video camera and the sound source is shifted from the shooting direction of the video camera. (Hereinafter referred to as situation X), even if the signals (L) and (R) are substantially in phase, the difference signal (LR) (the absolute value thereof) is likely to have a difference of a certain degree or more. Become. Therefore, if it is determined whether or not “wind noise” is determined only by the absolute value signal (LR) (the absolute value thereof), it is erroneously determined as “wind noise” even in the situation X as described above. Or misjudgment.

  However, it should be noted that in the present embodiment, the sum signal (L + R) is also used as a criterion for determining whether there is wind noise. In the case of the situation X, since the signals (L) and (R) are substantially in phase, the sum signal (L + R) has a larger value than the difference signal (LR). That is, in the case of speech, the relationship of (LR) <(L + R) is practically promised regardless of the situation, and as a result, the output of the wind noise detection signal generator 14 is obtained. The signal is 0 or a small value.

  On the other hand, in the case of wind noise, as described above, the correlation between the signals (L) and (R) (low frequency components) obtained from the two microphones is small. Therefore, when the average levels of a certain arbitrary period are compared, the absolute value of the signal (LR) tends to be larger than the absolute value of the signal (L + R). That is, the output signal of the wind noise detection signal generator 14 is larger than 0, and wind noise detection can be performed within a range where there is no problem.

  If two microphones happen to receive wind pressure changes of the same phase at the same time and output only signals having different amplitudes, it may be determined that the noise is not wind noise according to the above processing. . However, such a situation is local on the time axis in the case of shooting with a video camera, and when viewed as a whole, the above-mentioned sum signal (L + R) is also referred to. The effect is more remarkable.

  As described above, the accuracy of the wind noise detection signal can be greatly improved by using not only the difference signal (LR) but also the sum signal (L + R) as a parameter of the detection unit.

  The wind noise detection signal generated in such a process is read by the system control unit 7, and the cutoff frequency of the LchHPF3 and the RchHPF6 arranged in the main route of the audio signal is continuously controlled. In particular, wind noise whose main component is a low frequency can be reduced with an optimum filter setting.

  In the above embodiment, the wind noise detection signal generator 14 is described as calculating “absolute value (LR) −absolute value (L + R)”, but the absolute value (LR) or the absolute value is calculated. One of (L + R) may be multiplied by a coefficient α and then subtracted, and this coefficient α may be appropriately changed by a switch (not shown) so that a determination criterion for whether or not wind noise is present may be adjusted.

  In the embodiment, the stereo microphone normally used by the video camera is used as the sound collecting means. However, the present invention is not limited to this. For example, it may be applied to a recording apparatus having a plurality of microphones.

<Second Embodiment>
Next, a second embodiment will be described with reference to FIG. FIG. 4 corresponds to FIG. 3 in the above embodiment, except that the wind noise low-pass filter changeover switch SW18 is provided, and that the system control unit 7 controls the display signal processing unit 11. The point is that the wind noise level of the display unit 12 is displayed.

  In the second embodiment, LchHPF3 and RchHPF6 do not change the cut-off frequency continuously, and are of the ON / OFF switching type by the wind noise reduction filter switching switch 18.

  In FIG. 4, the system control unit 7 sends display data corresponding to the wind noise level to the display signal control unit 15 in accordance with the output level of the wind noise detection signal generator 14. Receiving the data, the display signal control unit 15 displays a bar indicating the wind noise level on the display unit 16 as indicated by reference numeral 17 in the figure (the more bars, the greater the wind noise). When the photographer judges that it is necessary by looking at this level display, the wind noise is reduced by switching the wind noise reduction filter switch 18.

  Some photographers may not like the low frequency level to be reduced automatically depending on the wind noise level. It is possible to reduce the amount of the component according to the shooting state.

It is a figure which shows the structure of the conventional wind noise detection. It is a figure which shows the structure of the whole video camera apparatus which embodiment applies. It is a figure which shows the structure of the wind noise detection which 1st Embodiment applies. It is a figure which shows the structure of the wind noise detection which 2nd Embodiment applies.

Claims (5)

A wind noise detecting device comprising a plurality of sound collecting means and detecting wind noise by signals from the plurality of sound collecting means,
Difference signal generating means for generating a difference signal from signals obtained by the plurality of sound collecting means;
Sum signal generating means for generating a sum signal from signals obtained by the plurality of sound collecting means;
First absolute value conversion means for converting the difference signal generated by the difference signal generation means into an absolute value signal;
Second absolute value conversion means for converting the sum signal generated by the sum signal generation means into an absolute value signal;
Wind noise detection means comprising: wind noise detection means for obtaining a difference between a difference signal obtained by conversion by the first and second absolute value conversion means and a sum signal and outputting the difference as a wind noise detection signal. apparatus. A filter means for extracting a low frequency component of the difference signal and the sum signal generated by the difference signal generation means and the sum signal generation means respectively;
2. The first absolute value conversion unit and the second absolute value conversion unit respectively convert a difference signal and a sum signal from which low frequency components are extracted into absolute value signals. Wind noise detection device.   3. The wind noise according to claim 1, further comprising a low-frequency region continuous control unit that continuously controls a level of a low-frequency region of an audio signal using a signal from the wind noise detection unit. Detection device.   The wind noise detection apparatus according to any one of claims 1 to 3, further comprising notification means for notifying a wind noise level according to a level of a signal from the wind noise detection means.   A video camera device comprising the wind noise detection device according to any one of claims 1 to 4.

Images