JP2008245254A - Audio processing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To remove wind noise contained in a low frequency component in an input sound. <P>SOLUTION: A wind noise detector (309) detects a level L of wind noise from audio signals outputted from microphones (301 to 304). A wind noise removing unit (306) removes wind noise from the audio signals outputted from the microphones (301 to 304) in accordance with the wind noise level L. A sound field converter (307) converts the audio signals outputted from the wind noise removing unit (306) to 5.1 channel audio signals. A sound volume adjusting unit (310) adjusts a sound volume level of a low frequency channel (LF) in accordance with the wind noise level L. An automatic level controller (308) controls the levels of sound signals from the sound field converter (307) and those from the sound volume adjusting unit (310) in accordance with one of the sound signals from the sound field converter (307). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a speech processing apparatus, and more particularly to a speech processing apparatus that reduces noise in input speech.

  The video camera records the surrounding sound with a stereo microphone while shooting a moving image of the subject.

  As a method for removing noise (wind noise) generated in an audio signal when wind strikes a microphone, a method for removing low-phase bass from 2-channel audio data is known (see, for example, Patent Document 1). ). There is also known a technique for changing a frequency range for removing low-frequency bass in accordance with a wind noise level.

In DVD video, a 5.1ch surround sound signal composed of a plurality of sound channels and sound channels centered on a low frequency is recorded as sound data. In recent years, video cameras that record such 5.1 channel surround sound signals have also appeared. Specifically, audio signals obtained by a plurality of omnidirectional microphones are subjected to matrix calculation, converted into 5.1ch audio signals, and recorded.
JP-A-4-270599

  In the noise removal technique described in Patent Document 1, when a sound signal of a plurality of channels is converted to 5.1ch sound after wind noise removal, low frequency components of wind noise that could not be removed may be emphasized. is there. When wind noise that cannot be completely removed is included in the low-frequency channel after 5.1 ch conversion, there is a problem that unpleasant sound is recorded.

  In view of the above problems, an object of the present invention is to present a speech processing device that removes noise contained in low frequency components in input speech.

  The speech processing apparatus according to the present invention includes a plurality of speech input means, a noise detection means for detecting a magnitude of noise included in a low frequency band of a plurality of speech signals output from the plurality of speech input means, Noise removing means for removing the noise from the plurality of voice signals output from the plurality of voice input means based on the output of the noise detecting means, and the plurality of voice signals output from the noise removing means, Conversion means for converting into audio data of a plurality of channels including a channel and other channels, and adjustment for controlling the level of the audio data of the low frequency channel according to the magnitude of noise detected by the noise detection means And a level of the audio data of the other channel output from the converting unit and the audio data of the low frequency channel output from the adjusting unit. Characterized in that it comprises a level control means for adjusting the.

  The speech processing apparatus according to the present invention includes a plurality of speech input means, a noise detection means for detecting a magnitude of noise included in a low frequency band of a plurality of speech signals output from the plurality of speech input means, A noise removing unit that removes the noise from the plurality of audio signals output from the plurality of audio input units based on an output of the noise detecting unit, and a directivity different from each other from the plurality of audio signals output from the noise removing unit A conversion unit that generates audio data of a plurality of channels, and a calculation unit that generates audio data of a plurality of channels having different directivities by calculating a plurality of audio signals output from the noise removing unit; A synthesizing unit that extracts and synthesizes low frequency components of a plurality of audio signals output from the noise removing device, and a magnitude of noise detected by the noise detecting unit A conversion unit that adjusts a level of an output signal of the synthesis unit according to the adjustment unit and outputs it as low-frequency channel audio data; and the plurality of channels of audio data and the low-frequency channel output from the conversion unit And level control means for adjusting the level of the audio data.

  According to the present invention, it is possible to remove noise included in low-frequency components of an input audio signal such as wind noise.

  FIG. 1 is a block diagram showing a schematic configuration of a video camera 100 in which an embodiment of a sound processing apparatus according to the present invention is mounted. FIG. 2 is an external perspective view of the video camera 100.

  First, the external appearance will be described with reference to FIG. The microphone unit 201 includes four microphones that convert surrounding sounds into electric signals. The photographing lens 202 forms an optical image of the subject on the image sensor. The display panel 203 displays captured images, reproduced images, and various other information. The display panel 203 is attached to the main body of the video camera 100 so as to be opened and closed by a hinge mechanism. In this specification, the direction in which the photographing lens faces is referred to as the front of the video camera 100.

  With reference to FIG. 1, the basic configuration and operation of the present embodiment will be described.

  The operation during shooting will be described. When the power is turned on by the power switch of the operation unit 110, the recording pause state is set. Alternatively, when the recording mode is selected with the mode dial, the recording pause state is set.

  In the recording pause state, the control unit 109 controls the imaging unit 101 to start capturing a subject image (capturing an image). The imaging unit 101 includes a photographing lens 201, an imaging element that converts an optical image obtained by the photographing lens 201 into an image signal, and a camera signal processing unit that converts an output image signal of the imaging element into a predetermined video signal format. A moving image signal from the imaging unit 101 is sent to the display control unit 104. The control unit 109 controls the display control unit 104 and causes the display unit 105 to display an image related to the moving image signal obtained by the imaging unit 101. The display unit 105 includes a display panel 203 and an electronic viewfinder on the back of the camera 100. The display control unit 104 can control which display means displays an image.

  The audio input unit 102 includes a microphone unit 201 and an audio processing unit that generates 5.1 channel audio data from the output audio signal of the microphone unit 201. In the recording pause state, the audio processing unit of the audio input unit 102 is paused. Details of the voice input unit 102 will be described later.

  When the user operates the recording trigger switch of the operation unit 110 in the recording pause state, the control unit 109 controls each unit to start recording of captured images and sounds. That is, first, the moving image data output from the imaging unit 101 and the audio input unit 102 are written in the memory 103 in accordance with a recording instruction signal from the control unit 109.

  The encoding processing unit 106 reads out the moving image data and audio data stored in the memory 103, compresses and encodes them according to a known MPEG method, and outputs the encoded moving image data and audio data to the recording / reproducing unit 107. . The recording / playback unit 107 multiplexes the encoded moving image data and audio data according to the recording format, and records the multiplexed data on the recording medium 108. When the user gives an instruction to stop recording, the recording / reproducing unit 107 stops data recording on the recording medium 108.

  In this embodiment, moving image data and audio data recorded from the start of recording to the stop of recording are managed as one scene.

  Next, the reproduction operation will be described. When the user instructs the playback mode with the mode dial of the operation unit 110, the control unit 109 switches the video camera 100 to the playback mode. The user can select a scene to be reproduced from a plurality of scenes recorded on the recording medium 108 by the operation unit 110, and instructs the reproduction of the selected scene. In response to this instruction, the control unit 109 causes the recording / playback unit 107 to play back the encoded moving image / audio data of the selected scene from the recording medium 108. The recording / reproducing unit 107 sends the reproduced encoded data to the encoding processing unit 106. The encoding processing unit 106 decodes the encoded moving image data and encoded audio data from the recording / reproducing unit 107, and stores the reproduced moving image data and the reproduced audio data in the memory 103.

  The display control unit 104 reads the playback moving image data from the memory 103 and displays the playback image on the display unit 105.

  The audio output unit 111 reads the reproduced audio data from the memory 103 and outputs it to the speaker 112. In the present embodiment, the speaker 112 is a speaker for stereo sound of two channels on the left and right. Therefore, 5.1 channel audio data cannot be output, as will be described later. Therefore, the audio output unit 111 converts the reproduced 5.1-channel audio data into 2-channel audio data and outputs the 2-channel audio data to the speaker 112.

  Reproduced moving image data and reproduced audio data on the memory 103 can be sequentially read out and output from the output unit 113 to an external device. The output unit 113 includes, for example, a USB or IEEE 1394 digital interface.

  FIG. 3 shows a schematic block diagram of the voice input unit 102. The microphone unit 201 includes four omnidirectional microphones 301 to 304 serving as four sound input means arranged in close proximity. FIG. 4 shows the arrangement of the microphones 301 to 304 viewed from the upper surface of the video camera 100. That is, the microphone 301 is relatively located on the front side of the video camera 100, the microphone 304 is located on the rear side, the microphone 302 is located on the right side, and the microphone 303 is located on the left side.

  The AD converter 305 includes A / D converters 305A to 305D corresponding to the microphones 301 to 304. The A / D converters 305A to 305D convert analog audio outputs from the microphones 301 to 304 into digital signals, respectively. Each A / D converter 305A-305D has an amplifier at its input stage. The audio data D1 to D4 output from the A / D converters 305A to 305D are input to the wind noise removing unit 306 and the wind noise detecting unit 309.

  The wind noise detection unit 309 detects wind noise in the digital audio signals D1 to D4 from the digital audio signals D1 to D4 output from the A / D converters 305A to 305D. And the signal L which shows the level (magnitude) of the detected wind noise is output. The wind noise removing unit 306 removes wind noise from the digital audio signals D1 to D4 output from the A / D converters 305A to 305D according to the wind noise level signal L from the wind noise detecting unit 309. The wind noise removing unit 306 outputs the digital audio signals D11 to D41 from which the wind noise has been removed to the sound field converting unit 307. Detailed operations of the wind noise detection unit 309 and the wind noise removal unit 306 will be described later.

  The sound field conversion unit 307 performs arithmetic processing on the 4-channel digital audio signals D11 to D41 from the wind noise removal unit 309 by a known method to generate a 5.1-channel digital audio signal. 5.1 channel audio signal is from front right channel (R), front left channel (L), front center channel (C), rear right channel (RS), rear left channel (LS) and low frequency channel (LF). Become.

  Specifically, the sound field conversion unit 307 generates a center channel (C) audio signal from the audio signal D11 and the audio signal D4. A front left channel (L) audio signal is generated from the audio signal D11 and the audio signal D2. The front right channel (R) audio signal is generated from the audio D11 and the audio signal output D31. A rear left channel (LS) audio signal is generated from the audio signal D21 and the audio signal D41. A rear right channel (RS) audio signal is generated from the audio signal D31 and the audio signal D41. A low frequency band (LF) audio signal is generated using the low frequency band components of the audio signals D11 to D41.

  Note that 5.1 channel audio may be in accordance with specifications such as Dolby Surround (trademark), but the present embodiment is not limited to this method.

  The audio data of channels other than the low frequency channel LF is output to the automatic level control (ALC) unit 308. The audio signal of the low-frequency channel LF is supplied to the ALC unit 308 via the volume adjustment unit 310 as a low frequency component of the input audio signal.

  The volume adjustment unit 310 adjusts the sound level (volume) of the low frequency channel LF according to the wind noise level signal L from the wind noise detection unit 309 and outputs the adjusted sound level (volume) to the ALC unit 308. The detailed operation of the volume adjustment unit 310 will be described later.

  The ALC unit 308 has a constant level for each level of the audio signals of the channels C, L, R, LS, and RS from the sound field converting unit 307 and the audio signal of the low frequency channel LF from the volume adjusting unit 310 as a whole. Adjust so that The audio data whose level has been adjusted by the ALC unit 308 is stored in the memory 103.

  Specifically, the ALC unit 308 determines a level adjustment amount so that the level of any channel having the highest level among the audio signals of each channel from the sound field conversion unit 307 becomes a predetermined level. And according to the determined level adjustment amount, the level of the audio signal of all the channels is adjusted in common. In 5.1ch audio, the balance between the channels is important, and it is necessary to adjust the audio level of each channel so that the balance between the channels is optimal. Therefore, in this embodiment, the ALC unit 308 can adjust the audio level of each channel uniformly, so that the level can be adjusted while maintaining this balance.

  FIG. 6 shows a schematic block diagram of the wind noise detection unit 309. The wind noise detection unit 309 has a system for detecting the wind noise level L1 using the audio signals D1 and D2, and a system for detecting the wind noise level L2 using the audio signals D3 and D4. Then, the wind noise detection unit 309 compares the two wind noise levels L1 and L2, and finally calculates an average wind noise level L. Although the audio signals D1 and D2 are paired and the audio signals D3 and D4 are paired, this combination is convenient and does not depend on the positions of the microphones 301 to 304.

  In the case of normal sound, since directivity is low in the low sound range, signals having the same phase are obtained if a plurality of microphones are close to each other. However, the low frequency range generated by wind on the microphone has no correlation and does not have the same phase. The wind noise detection unit 309 detects the wind noise level L using this characteristic.

  The adder 501 adds the audio signal D1 and the audio signal D2, and outputs a sum signal D1 + D2. The subtracter 502 subtracts the audio signal D2 from the audio signal D1, and outputs a difference signal D1-D2. The absolute value conversion unit 503 converts the output signal D1 + D2 of the adder 501 into its absolute value, and outputs the absolute value signal | D1 + D2 | to an LPF (low-pass filter) 505. The absolute value converter 504 converts the output signal D1-D2 of the subtractor 502 into an absolute value, and outputs the absolute value signal | D1-D2 | to the LPF 506 having substantially the same transfer characteristics as the LPF 505. LPFs 505 and 506 are digital filters that remove high frequency components of the input signal. As far as the low frequency range of the audio signal is concerned, the output signal of the LPF 505 is approximately equal to the input signal | D1 + D2 |, and the output signal of the LPF 506 is approximately equal to the input signal | D1-D2 |.

  The subtracter 507 subtracts the output signal | D1-D2 | of the LPF 506 from the output signal | D1 + D2 | of the LPF 505. The output of the subtracter 507 is approximately equivalent to | D1 + D2 | − | D1−D2 |. The envelope detection unit 508 detects the envelope of the output signal of the subtractor 507 and outputs the detected envelope level L1.

  Of the audio signals input from the microphones 301 to 304, in the case of audio from a subject that is not wind noise, the low frequency components of the audio signals D1 to D4 have the same phase. At this time, the output signal of the LPF 505 is | D1 + D2 | ≈ | 2 × D1 | ≈ | 2 × D2 |, and the output signal of the LPF 506 is | D1−D2 | ≈0. As a result, the output of the subtracter 507 becomes | 2 × D1 | or | 2 × D2 |.

  On the other hand, in the case of wind noise, the low sound range of the audio signals D1 to D4 has no correlation. For this reason, the output | D1-D2 | of the LPF 506 is larger than the output | D1 + D2 | of the LPF 505. In particular, when the phase of the wind noise component included in the audio signals D1 and D2 is 180 ° different, the output of the LPF 505 is | D1 + D2 | ≈0, and the output of the LPF 506 is | D1-D2 | ≈ | 2 × D1 | ≈ | 2 × D2 |.

  As a result, when wind noise is included, the output of the subtracter 507 becomes a negative value. In particular, when the phase of the wind noise component included in the audio signals D1 and D2 differs by 180 °, the output of the subtractor 507 is − | 2 × D1 | or − | 2 × D2 |.

  Thus, when the sign of the output value of the subtractor 507 is negative, the output signal of the subtracter 507, that is, the low frequency component of the difference between the audio signal D1 and the same D2 reflects the level of wind noise.

  When the sign of the output value of the subtracter 507 is negative, the envelope detection unit 508 outputs the envelope level L1 of the output signal of the subtracter 507. When the sign of the output value of the subtracter 507 is positive, the value 0 is output as the output L1.

  The operation for calculating the noise level L2 from the audio signals D3 and D4 is basically the same as that for the audio signals D1 and D2.

  That is, the adder 509 adds the audio signal D3 and the audio signal D4, and outputs a sum signal D3 + D4. The subtracter 510 subtracts the audio signal D4 from the audio signal D3, and outputs a difference signal D3-D4. The absolute value conversion unit 511 converts the output signal D3 + D4 of the adder 509 into its absolute value, and outputs the absolute value signal | D3 + D4 | to the LPF 513. The absolute value converter 512 converts the output signal D3-D4 of the subtractor 510 into its absolute value, and outputs the absolute value signal | D3-D4 | to the LPF 514 having substantially the same transfer characteristics as the LPF 513. As far as the low frequency range of the audio signal is concerned, the output signal of the LPF 513 is approximately equal to the input signal | D3 + D4 |, and the output signal of the LPF 514 is approximately equal to the input signal | D3-D4 |.

  The subtractor 515 subtracts the output signal | D3−D4 | of the LPF 514 from the output signal | D3 + D4 | of the LPF 513. The output of the subtracter 515 corresponds approximately to | D3 + D4 |-| D3-D4 |. The envelope detector 516 detects the envelope of the output signal of the subtractor 515 and outputs the detected envelope level L2.

  In the case of audio from a subject that is not wind noise among the audio signals input from the microphones 301 to 304, the output signal of the LPF 513 is | D3 + D4 | ≈ | 2 × D3 | ≈ | 2 × D4 |, and the output of the LPF 514 The signal is | D3-D4 | ≈0. As a result, the output of the subtracter 515 is | 2 × D3 | or | 2 × D4 |.

  On the other hand, in the case of wind noise, since the low sound range of the audio signals D1 to D4 has no correlation, the output | D3-D4 | of the LPF 514 is larger than the output | D3 + D4 | of the LPF 513. In particular, when the phase of the wind noise component included in the audio signal D3 and the same D4 is 180 ° different, the output of the LPF 513 is | D3 + D4 | ≈0, and the output of the LPF510 is | D3-D4 | ≈ | 2 × D3 | ≈ | 2 × D4 |.

  As a result, when wind noise is included, the output of the subtracter 515 becomes a negative value. In particular, when the phase of the wind noise component included in the audio signal D3 and the same D4 is 180 °, the output of the subtracter 515 is − | 2 × D3 | or − | 2 × D4 |.

  In this way, when the sign of the output value of the subtractor 515 is negative, the output signal of the subtracter 515, that is, the low frequency component of the difference between the audio signal D3 and D4, reflects the level of wind noise. .

  Similar to the envelope detector 508, the envelope detector 516 outputs the envelope level L2 of the output signal of the subtracter 515 when the sign of the output value of the subtracter 515 is negative. When the sign of the output value of the subtracter 515 is positive, the value 0 is output as the output L2.

  The determination unit 517 calculates an average value of the level L1 from the envelope detection unit 508 and the level L2 from the envelope detection unit 516, and outputs the average value L as an overall average level L. Instead of outputting the average value of the wind noise levels L1 and L2, a large value of the wind noise levels L1 and L2 may be output as the detection level L.

  FIG. 7 shows a schematic block diagram of the wind noise removing unit 306. The operation of the wind noise removing unit 306 will be described with reference to FIG.

  The dust removal unit 306 includes a processing system that removes wind noise contained in the audio signals D1 and D2, and a processing system that removes wind noise contained in the audio signals D3 and D4. In each system, in this embodiment, the low frequency component of the difference signal D1-D2 is removed from the audio signals D1, D2, and the low frequency component of the difference signal D3-D4 is removed from the audio signals D3, D4. Reduce noise. At that time, the higher the wind noise level L, the higher the difference signal D1-D2, D3-D4 low-frequency cutoff frequency.

  The adder 601 adds the audio signal D1 and the audio signal D2, and outputs a sum signal D1 + D2. The subtracter 602 subtracts the audio signal D2 from the audio signal D1, and outputs a difference signal D1-D2. An HPF (high pass filter) 603 removes a low frequency component equal to or lower than the cutoff frequency of the difference signal D1-D2, and passes the remaining high frequency component. The cutoff frequency control unit 615 switches the cutoff frequency of the HPF 603 according to the wind noise level L from the wind noise detection unit 309.

  FIG. 8 shows three examples of frequency characteristics of the HPF 603. FIG. 8A shows frequency characteristics when the wind noise level L is smaller than the first threshold value. FIG. 8B shows the frequency characteristics when the wind noise level L is equal to or higher than the first threshold and smaller than the second threshold greater than the first threshold. FIG. 8C shows frequency characteristics when the wind noise level L is greater than or equal to the second threshold value. 8A to 8C, the horizontal axis represents frequency, and the vertical axis represents amplitude (or transmittance).

  When the wind noise level L is smaller than the first threshold, the cutoff frequency control unit 615 causes the HPF 603 to attenuate the signal level in the entire band from low to high as shown in FIG. 8A. The HPF 603 is controlled to output. That is, the HPF 603 is in a through state.

  When the noise level L is equal to or higher than the first threshold value and smaller than the second threshold value that is larger than the first threshold value, the cutoff frequency control unit 615 sets the low HPF 603 as shown in FIG. The band cutoff frequency is set to the frequency f1. Thereby, the band component below the frequency f1 is attenuated. When the wind noise level L is greater than or equal to the threshold 2, the cutoff frequency control unit 615 sets the cutoff frequency of the HPF 603 to f2 higher than the frequency f1, as shown in FIG. 8C.

  By calculating the difference between audio signals from two adjacent microphones, a wind noise component can be extracted. When the wind noise level is high, it is considered that the wind noise extends to a high frequency. By controlling the cutoff frequency of the HPF 603 according to the wind noise level L, the wind noise component can be effectively suppressed from the difference signal.

  The adder 604 adds the output of the HPF 603 to the output of the adder 601. The output of the adder 604 is approximately 2D1 (≈ (D1 + D2) + (D1−D2)) when the affected part of the HPF 603 is ignored. The subtracter 605 subtracts the output of the HPF 603 from the output of the adder 601. The output of the subtractor 605 is approximately 2D2 (≈ (D1 + D2) − (D1−D2)) when the affected part of the HPF 603 is ignored.

  As described above, the wind noise components included in the audio signals D1 and D2 have no correlation. Therefore, the difference signal D1-D2 (the low frequency component thereof) in the case of including wind noise is larger than the difference signal D1-D2 in the case of normal speech. By removing or suppressing this with the HPF 603, wind noise can be reduced.

  The amplifier 606 adjusts the audio level of the output signal of the adder 604 to, for example, ½. Similarly, the amplifier 607 adjusts the audio level of the output signal of the subtractor 605 to, for example, ½. As a result, the amplifier 606 outputs the audio signal D11 from which wind noise has been removed or suppressed, and the amplifier 607 outputs the audio signal D21 from which wind noise has been removed or suppressed.

  The other system for removing wind noise from the audio signals D3 and D4 operates in the same manner. That is, the adder 608 adds the audio signal D3 and the audio signal D4, and outputs a sum signal D3 + D4. The subtractor 609 subtracts the audio signal D4 from the audio signal D3, and outputs a difference signal D3-D4.

  The HPF 610 attenuates the low frequency component below the low frequency cutoff frequency from the difference signal D3-D4 and passes the remaining high frequency component. The cutoff frequency control unit 615 controls the low frequency cutoff frequency of the HPF 610 to the same low frequency cutoff frequency as that of the HPF 603 according to the wind noise level L.

  The adder 611 adds the output of the adder 608 and the output of the HPF 610. The output of the adder 611 is approximately 2D3 (≈ (D3 + D4) + (D3−D4)) when the affected part of the HPF 610 is ignored. The subtractor 612 subtracts the output of the HPF 610 from the output of the adder 608. The output of the subtractor 612 is approximately 2D4 (≈ (D3 + D4) − (D3−D4)) when the affected part of the HPF 610 is ignored.

  The amplifier 613 adjusts the audio level of the output signal of the adder 611 to 1/2, for example. Similarly, the amplifier 614 adjusts the audio level of the output signal of the subtractor 612 to, for example, 1/2. As a result, the amplifier 613 outputs the audio signal D31 from which the wind noise has been removed, and the amplifier 614 outputs the audio signal D41 from which the wind noise has been removed.

  As described above, when the wind noise is not detected, that is, when the wind noise level L is very low, the HPFs 603 and 610 output the input signals of the entire band from the low range to the high range without being attenuated. Further, when the wind noise level L increases, the cutoff frequency of the HPFs 603 and 610 increases, and low frequency components up to higher frequency components are removed.

  In the present embodiment, the cutoff frequency control unit 615 illustrated FIGS. 8A to 8C as examples of the frequency characteristics of the HPFs 603 and 610. Of course, the cutoff frequency control unit 615 includes the low frequency range of the HPFs 603 and 610. The cutoff frequency can be controlled continuously or discontinuously.

  In the unit configuration shown in FIG. 7, the same combination as the pair of audio signals D1 and D2 and the pair of audio signals D3 and D4 used for wind noise detection in the wind noise detection unit 309 shown in FIG. 6 is used. In this way, by using the same combination as the wind noise detection, it is possible to suppress the influence of variations in the characteristics of the microphones 301 to 304. Of course, the difference between the audio signals D1 and D3 may be calculated, the difference between the audio signals D1 and D4 may be calculated, and the low frequency components of these difference signals may be removed.

  The volume adjustment unit 310 is composed of a variable gain attenuator in which the amount of attenuation increases as the wind noise level L increases. The volume adjustment unit 310 adjusts the amplitude of the low frequency channel LF according to the wind noise level L from the wind noise detection unit 309 and outputs the adjusted signal to the ALC unit 308.

  5A to 5C show examples of gain characteristics with respect to the wind noise level L of the volume adjustment unit 310. FIG. The horizontal axis indicates the wind noise level L, and the vertical axis indicates the gain of the volume adjusting unit 310. In FIG. 5A, the gain is constant when the wind noise level L is in the range of 0 to the predetermined value La, and the gain is decreased as the level L increases above La. In FIG. 5B, the gain is simply reduced as the wind noise level L increases. In FIG. 5C, the gain is constant in the range where the wind noise level L is from 0 to the first threshold value La, and in the range of Lb higher than La to La, the gain is decreased as the level L increases. Above Lb, the gain is made constant again.

  The volume adjustment unit 310 adjusts the level of the low-frequency channel LF according to the characteristics shown in any of FIGS. In this way, by adjusting the level of the low-frequency channel LF according to the wind noise level L, the ALC unit 308 can collectively adjust the level of the low-frequency channel LF in the same manner as the levels of other channels. Even with the adjustment of the ALC unit 308, the wind noise is not enhanced.

  FIG. 9 shows another configuration example of the voice input unit 102. The sound field conversion unit 307a is equipped with the functions of the sound field conversion unit 307 and the volume adjustment unit 310. The same components as those in FIG. 3 are denoted by the same reference numerals.

  The configuration shown in FIG. 9 differs from FIG. 3 in that the low-frequency channel LF generation process in the sound field conversion unit 307a is controlled according to the wind noise level L from the wind noise detection unit 309.

  FIG. 10 shows a schematic block diagram of the sound field converter 307a. The audio signals D11 to D41 output from the wind noise removal unit 306 are input to the calculation unit 901 and the low frequency channel generation unit 902. The wind noise level signal L output from the wind noise detection unit 309 is supplied to the low frequency channel generation unit 902.

  The calculation unit 901 generates audio data of channels C, L, R, LS, and RS from input audio signals D11 to D41 by a known calculation. On the other hand, the low frequency channel generation unit 902 extracts audio data of a determined band from the input audio signals D11 to D41, and generates audio data of the low frequency channel LF.

  FIG. 11 shows a schematic block diagram of the low-frequency channel generation unit 902. Input audio signals D11 to D41 are supplied to band pass filters (BPF) 1001 to 1004, respectively. Each of the BPFs 1001 to 1004 extracts a predetermined frequency band, for example, a component between 100 kHz and 200 kHz, from the input audio signals D11 to D41, and outputs the extracted component to the synthesis unit 1005. The combining unit 1005 combines the outputs of the BPFs 1001 to 1004 and outputs the combined outputs to the level adjusting unit 1006.

  Based on the wind noise level signal L from the wind noise detection unit 309, the level adjustment unit 1006 adjusts the level of the low-frequency channel audio data output from the synthesis unit 1005. Specifically, the level adjustment unit 1006 adjusts the level of the output signal from the synthesis unit 1005 according to the wind noise level L, for example, with the characteristics shown in FIG. 5 (A), (B), or (C). To do. The output signal of the level adjustment unit 1006 becomes the audio signal of the low frequency channel LF.

  9 to 11, the sound field conversion unit 307 a adjusts the level of the low frequency channel LF according to the wind noise level L. Accordingly, as in the case of the configuration shown in FIG. 3, the ALC unit 308 can collectively adjust the level of the low-frequency channel LF in the same manner as the levels of other channels. Even with the adjustment of the ALC unit 308, the wind noise is not enhanced.

  In the above description, a 5.1ch audio signal is generated from the audio signals captured by the four microphones 301 to 304. However, the present invention is not limited to 5.1ch, and converts the audio signal to a larger number of channels. It is also applicable to cases. Further, the number of microphones is not limited to four, and may be any number other than this.

1 is a block diagram of a schematic configuration of a video camera according to an embodiment of the present invention. It is an external appearance perspective view of the video camera of a present Example. 3 is a block diagram of a schematic configuration of a voice input unit 102. FIG. FIG. 4 is a layout diagram of four microphones constituting the microphone unit 201. It is an example of the characteristic of the volume adjustment part 310. FIG. It is a schematic block diagram of a wind noise detection part. It is a schematic block diagram of a wind noise removal part. It is a figure which shows the example of three frequency characteristics in a wind noise removal part. It is a schematic block diagram of another structural example of an audio | voice input part. It is a schematic block diagram of the sound field converter shown in FIG. It is a schematic block diagram of a low-frequency channel generation unit shown in FIG.

Explanation of symbols

100: video camera 101: imaging unit 102: audio input unit 103: memory 104: display control unit 105: display unit 106: encoding processing unit 107: recording / playback unit 108: recording medium 109: control unit 110: operation unit 111: Audio output unit 112: speaker 113: output unit 201: microphone unit 202: photographing lens 203: display panels 301 to 304: omnidirectional microphone 305: AD converters 305A to 305D: A / D converter 306: wind noise removing unit 307 307a: Sound field conversion unit 308: Automatic level control (ALC) unit 309: Wind noise detection unit 310: Volume adjustment unit 501: Adder 502: Subtractor 503: Absolute value conversion unit 504: Absolute value conversion unit 505: LPF
506: LPF
507: Subtractor 508: Envelope detector 509: Adder 510: Subtractor 511: Absolute value converter 512: Absolute value converter 513: LPF
514: LPF
515: Subtractor 516: Envelope detection unit 517: Determination unit 601: Adder 602: Subtractor 603: HPF (High Pass Filter)
604: Adder 605: Subtractor 606: Amplifier 607: Amplifier 608: Adder 609: Subtractor 610: HPF
611: Adder 612: Subtractor 613: Amplifier 614: Amplifier 615: Cutoff frequency controller 901: Arithmetic unit 902: Low-frequency channel generators 1001 to 1004: Band pass filter (BPF)
1005: Composition unit 1006: Level adjustment unit

Claims (12)

A plurality of voice input means;
Noise detection means for detecting the magnitude of noise included in a low frequency band of a plurality of voice signals output from the plurality of voice input means;
Noise removing means for removing the noise of the plurality of audio signals output from the plurality of audio input means based on the output of the noise detecting means;
Converting means for converting the plurality of audio signals output from the noise removing means into audio data of a plurality of channels including a low frequency channel and other channels;
Adjusting means for controlling the level of the audio data of the low frequency channel according to the magnitude of noise detected by the noise detecting means;
An audio processing apparatus comprising: level control means for adjusting the level of the audio data of the other channel output from the conversion means and the level of audio data of the low frequency channel output from the adjustment means.   The audio processing apparatus according to claim 1, wherein the adjustment unit increases the attenuation amount of audio data of the low frequency channel as the magnitude of the noise increases.   The noise removing unit is configured such that a difference between two audio signals of a plurality of audio signals output from the plurality of audio input units is input, and the noise detected by the noise detection unit increases. The speech processing apparatus according to claim 1, further comprising: a cutoff frequency control device that increases a cutoff frequency of the high-pass filter.   The converting unit converts the plurality of audio signals output from the noise removing unit into audio data of a plurality of channels having different directivities, and converting the audio data of the plurality of channels having different directivities to the other Output as audio data of the channel, extract low frequency components of the plurality of audio signals output from the noise removing unit, and synthesize the audio signals of the extracted low frequency components, thereby The voice processing apparatus according to claim 1, wherein voice data is generated.   2. The noise detection unit according to claim 1, wherein the noise detection unit detects the magnitude of the noise using a difference between audio signals from any two of the plurality of audio input units. Audio processing device.   The level control means is configured to control the audio data of the other channel and the audio data of the low frequency channel according to the level of any one of the audio data of the other channel and the audio data of the low frequency channel. 2. The speech processing apparatus according to claim 1, wherein the levels of the audio signals are controlled in common. A plurality of voice input means;
Noise detection means for detecting the magnitude of noise included in a low frequency band of a plurality of voice signals output from the plurality of voice input means;
Noise removing means for removing the noise of the plurality of audio signals output from the plurality of audio input means based on the output of the noise detecting means;
Conversion means for generating sound data of a plurality of channels having different directivities from a plurality of sound signals output from the noise removing means, by calculating a plurality of sound signals output from the noise removing means; Detected by an arithmetic unit that generates audio data of a plurality of channels with different directivities, a synthesis unit that extracts and synthesizes low frequency components of a plurality of audio signals output from the noise removal unit, and the noise detection unit A conversion unit having an adjustment unit that adjusts the level of the output signal of the synthesis unit according to the magnitude of the generated noise and outputs the low-frequency channel audio data;
An audio processing apparatus comprising: level control means for adjusting levels of the audio data of the plurality of channels output from the conversion means and the audio data of the low frequency channel.   The audio processing apparatus according to claim 7, wherein the adjustment unit increases an attenuation amount of audio data of the channel of the low frequency component as the magnitude of the noise increases.   The noise removing unit is configured such that a difference between two audio signals of a plurality of audio signals output from the plurality of audio input units is input, and the noise detected by the noise detection unit increases. The voice processing device according to claim 7, further comprising a cutoff frequency control device that increases a cutoff frequency of the high-pass filter.   8. The noise detection unit according to claim 7, wherein the noise detection unit detects the magnitude of the noise using a difference between audio signals from any two of the plurality of audio input units. Audio processing device.   The level control means is configured to control the audio data of the other channel and the audio data of the low frequency component according to the level of any one of the audio data of the other channel and the audio data of the low frequency component. The audio processing apparatus according to claim 7, wherein the levels of the audio signal are controlled in common.   The voice processing apparatus according to claim 1, wherein each of the plurality of voice input units is a non-directional microphone.

Images