JP2005039509A - Stereo microphone device and stereo operation method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a stereo microphone device capable of obtaining a stereo characteristic by which there is no deterioration of window noise in a window noise band, obtaining a flat gain characteristic over the whole stereo operation processing band and simplifying an equalizer arranged on the post stage. <P>SOLUTION: The stereo microphone device comprises band division means 55-58 for dividing a voice band from at least one of nondirectional microphones 51, 52 into a plurality of bands and adders 63-66 for adding 1st band outputs obtained through 1st delay means 60, 61 for delaying the time of 1st band signals from the band division means 56, 57 to outputs from the other microphones and subtracting 2nd band outputs through 2nd delay means 59, 62 for delaying the time of 2nd band signals from the band division means 55, 58 from outputs outputted from the other microphones. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to, for example, a stereo microphone device that records stereo sound and a stereo calculation method.

  A technique for generating a stereo sound field with a built-in microphone (hereinafter abbreviated as a microphone) of a video camera is generally a built-in stereo microphone described in Patent Document 1.

  FIG. 8 shows a conventional stereo arithmetic circuit according to Patent Document 1. In FIG. First, Rch and Lch audio signals are input from the microphones 101 and 102, respectively, and the signal level is optimized by amplifiers (AMP) 103 and 104. The Rch signal is added to the + terminal of the adder 109 and generally LPF (Low Pass Filter). Is input to a delay unit DL (Delay Line) 105. Similarly, the Lch signal is input to the + terminal of the adder 110 and the delay unit DL106, and the low frequency components of the voice band delayed by the DLs 105 and 106 are set to appropriate levels by the attenuators ATT (Attenuator) 107 and 108, respectively. And input to the minus terminals of the adders 110 and 109.

  In the adders 109 and 110, matrix processing is performed by subtracting low-frequency component signals from the channels, that is, signals below the cutoff frequency of the LPF set by the DLs 105 and 106, and stereo processing is performed in the low-frequency band. The sound field is reproduced. Further, the outputs of the adders 109 and 110 are adjusted in frequency characteristics by equalizers (Equalizers) 111 and 112 constituted by filter circuits, and output from terminals 113 and 114 as Rch and Lch signals.

Here, FIG. 9 shows and describes the stereo directivity characteristic generated by the stereo arithmetic circuit of FIG.
First, in the stereo arithmetic circuit of FIG. 8, omnidirectional microphones having directivity in all directions are used for the microphones 101 and 102, and the delay units DL105 and 106 have a time delay depending on the distance between the microphones. The signals are inputted with Lch and Rch signals with a phase difference corresponding to this time delay, and are subtracted from each other channel and subjected to stereo operation processing. The directivity characteristics of Lch (solid line) and Rch (broken line) are substantially the same as indicated by 121 and show monaural characteristics. However, after stereo operation processing, the directivity characteristics of Lch (solid line) and Rch are shown in FIG. 9B. The directivity characteristics (broken line) are separated in the left-right direction as indicated by 122 and 123, respectively, and exhibit stereo characteristics.

  However, since the wind noise signal is input to the Lch and Rch signals with a random phase difference, the level is deteriorated by 6 dB when the input signals are added in the same phase by the adders 109 and 110, for example. In addition, in order to increase the gain of the wind noise band in the subsequent EQs 111 and 112, there is a problem that the wind noise is greatly emphasized as compared with the audio signal. In particular, due to the recent miniaturization of products, the distance between the microphones is narrowed, and the gain increase amount in the subsequent EQ is further increased, and this tendency becomes more prominent.

  Next, FIG. 10 shows a frequency characteristic example of a wind noise signal in a general video camera. The frequency characteristic 131 of the wind noise increases in level with a 1 / F characteristic (F is a frequency) as the frequency decreases from about 1 kHz. In practice, however, the level of the extremely low frequency decreases due to the characteristics of the microphone unit used and the influence of the coupling capacitor of the analog circuit, so that it often has a peak in the vicinity of about 200 Hz.

Japanese Patent No. 2946638

  However, since the stereo calculation circuit of the prior application emphasizes the uncorrelated component that is the phase difference and level difference of the L and R signals, wind noise that is mostly uncorrelated is emphasized when passing through this calculation circuit. There is. In particular, due to the recent miniaturization of equipment, the L and R microphone intervals are narrowed, and the structural stereo effect cannot be obtained, so that the circuit correction by the stereo arithmetic circuit is emphasized, and this problem becomes more prominent. There was an inconvenience.

  The present invention is made in view of this point, and by changing the method of stereo calculation processing in the low-frequency side wind noise band and the other band on the high-frequency side, respectively, especially in the wind noise band. The present invention proposes a stereo microphone device and a stereo calculation method that can obtain a stereo characteristic that does not deteriorate the wind noise, and that can obtain a flat gain characteristic over the entire stereo calculation processing band, thereby simplifying the equalizer in the subsequent stage.

  In order to solve the above-described problems and achieve the object of the present invention, the stereo microphone device of the present invention is configured to divide a sound band from at least one omnidirectional microphone of a plurality of omnidirectional microphones into a plurality of bands. And at least one non-directional microphone other than the omnidirectional microphone divided by the band dividing unit, the first band output via the first delay unit that delays the first band signal from the band dividing unit. The second band output via the adding means for adding to the output of the directional microphone and the second delay means for delaying the second band signal from the band dividing means is time-divided by the band dividing means. Subtracting means for subtracting from the output of at least one omnidirectional microphone other than the directional microphone is provided.

  Accordingly, it is possible to provide a stereo microphone device that can suppress wind noise in a low frequency range and obtain flat stereo characteristics in the entire frequency band.

  The stereo calculation method of the present invention includes a step of dividing a sound band from at least one omnidirectional microphone of a plurality of omnidirectional microphones into a plurality of bands by a band dividing unit, and a first from the band dividing unit. The first band output via the first delay means for delaying the band signal of time is added by the adding means with the output of at least one omnidirectional microphone other than the omnidirectional microphone divided by the band dividing means. And a second band output via the second delay means for time delaying the second band signal from the band dividing means, and at least one non-directional microphone other than the omnidirectional microphone divided by the band dividing means. Subtracting means from the output of the directional microphone by the subtracting means.

  Thereby, it is possible to perform a stereo calculation that suppresses wind noise in a low frequency range and obtains flat stereo characteristics in the entire frequency range.

  According to the invention of claim 1, by dividing the audio band into a plurality of bands, and stereoizing each band by separate stereo processing, the frequency characteristics can be flattened over the entire audio band, Since no gain correction is required, an equalizer circuit constituted by a subsequent filter or the like can be eliminated.

  In addition, according to the invention of claim 2, a stereo effect can be obtained while suppressing wind noise in a low frequency range that is particularly susceptible to wind noise, and there is no level dip (null point) in the high frequency range. Therefore, a flat stereo effect is obtained over the entire audio band.

  According to the invention of claim 3, by dividing the audio band into a plurality of bands and stereoizing each band by separate stereo operation processing, the frequency characteristics are flattened over the entire audio band. In addition, since no gain correction is required, it is possible to eliminate the need for an equalizer process constituted by a subsequent filter process.

  According to the invention of claim 4, a stereo effect can be obtained while suppressing wind noise particularly in a low range that is easily affected by wind noise, and there is no level dip (null point) in the high range. Therefore, a flat stereo effect is obtained over the entire audio band.

Therefore, the present invention has an object of performing a stereo calculation process without enhancing the wind noise in these wind noise bands, which is emphasized by a conventional stereo calculation circuit.
Here, FIG. 1 shows a stereo operation principle diagram of a stereo microphone device according to an embodiment applied to the present invention.

  First, Rch and Lch audio signals are input from the microphones 1 and 2, respectively, and the signal level is optimized by the AMPs 3 and 4, and the Rch signal is input to one plus terminal of the adder 7 and the delay DL5. Similarly, the Lch signal is input to one + terminal of the adder 8 and the delay unit DL6, and the audio signal delayed by the DLs 5 and 6 is input to the other + terminal of the adders 8 and 7. In the adders 7 and 8, matrix processing is performed by subtracting the signals delayed in DL5 and DL6 from the channels, and a stereo sound field is reproduced. Further, the outputs of the adders 7 and 8 are output from the terminals 9 and 10 as Rch and Lch signals.

  Further, the difference between the conventional stereo calculation block diagram shown in FIG. 8 and the stereo calculation processing of FIG. 1 will be described. First, in the conventional processing of FIG. 8, a delay characteristic DL 105, 106 delays T [sec] and the adders 109 and 110 subtract from each other's channels will be described with reference to FIG. The horizontal axis represents a frequency F [Hz] having a period of T, and the vertical axis represents GAIN [dB]. Here, the solid line 21 is a frequency characteristic when there is no phase difference between the inputs of the microphone 101 and the microphone 102, for example, when sound is input from the front center direction, and this is expressed by Equation (1). (However, the input signal was a sine wave with an amplitude of 1, and ATTs 107 and 108 were through.)

(Equation 1)
sin ωt−sin ω (t−T) = 2 sin (πfT) · cos (ωt−πfT)

  According to Equation 1, the amplitude term 2sin (πfT) is such that GAIN becomes the minimum value (zero) when f = 0 [Hz] and f = 1 / T [Hz], and GAIN when f = 1 / 2T [Hz]. Becomes the maximum value (+6 dB), and the outputs of the adders 109 and 110 are the same.

  On the other hand, the broken line 22 and the alternate long and short dash line 23 are when there is a phase difference between the inputs of the microphone 101 and the microphone 102, for example, when sound is input from any direction that is not equidistant from the microphone 101 and the microphone 102. The frequency characteristic is expressed by the following equation (2). (However, the input signal was a sine wave with an amplitude of 1, and ATTs 107 and 108 were through.)

(Equation 2)
sinωt−sinω (t− (T ± Ts)) = 2sin (πf (T ± Ts)) · cos (ωt−πf (T ± Ts))

  In this case, a signal delay Ts (where T> Ts) due to a difference in distance from the sound source is added to T, resulting in a relative delay (T ± Ts). The sign ± changes depending on whether Lch or Rch is used as a reference, depending on the left-right direction of the sound source. Therefore, according to Equation 2, the amplitude term 2sin (πf (T ± Ts)) is f = 0 [Hz] and f = 1 / (T ± Ts) [Hz], and GAIN becomes the minimum value (zero). When f = 1/2 (T ± Ts) [Hz], GAIN becomes the maximum value (+6 dB), and the outputs of the adders 109 and 110 correspond to the input phase difference as shown by the broken line 22 and the alternate long and short dash line 23. There is a difference and this becomes stereo separation.

  In general, since f = 1 / 2T [Hz] or less is a stereo calculation processing band, it has a frequency characteristic in which the low band is attenuated, and the frequency is obtained by level-correcting the low band with EQs 111 and 112 in the subsequent stages. The characteristics are flattened.

  Next, FIG. 3 shows an example of frequency characteristics in the case where the delay operation DL5, 6 delays T [sec] in the stereo operation principle diagram of the present invention in FIG. explain. The horizontal axis represents a frequency F [Hz] having a period of T, and the vertical axis represents GAIN [dB]. Further, the solid line 31 is a frequency characteristic when there is no phase difference between the inputs of the microphone 1 and the microphone 2, for example, when sound is input from the front center direction, which is expressed by the following equation (3). (However, the input signal was a sine wave with an amplitude of 1.)

(Equation 3)
sinωt + sinω (t−T) = 2 cos (πfT) · sin (ωt−πfT)

  According to Equation 3, the amplitude term 2cos (πfT) is such that GAIN becomes the maximum value (+6 dB) when f = 0 [Hz] and f = 1 / T [Hz], and GAIN when f = 1 / 2T [Hz]. Becomes the minimum value (zero), and the outputs of the adders 7 and 8 are the same.

  On the other hand, the broken line 32 and the alternate long and short dash line 33 are when there is a phase difference between the inputs of the microphone 1 and the microphone 2, for example, when sound is input from an arbitrary direction that is not equidistant from the microphone 1 and the microphone 2. This is a frequency characteristic expressed by the following equation (4). (However, the input signal was a sine wave with an amplitude of 1.)

(Equation 4)
sinωt + sinω (t− (T ± Ts)) = 2 cos (πf (T ± Ts)) · sin (ωt−πf (T ± Ts))

  In this case, a signal delay Ts (where T> Ts) due to a difference in distance from the sound source is added to T, resulting in a relative delay (T ± Ts). The sign ± changes depending on whether Lch or Rch is used as a reference, depending on the left-right direction of the sound source. Therefore, according to the equation 4, the amplitude term 2cos (πf (T ± Ts)) is such that GAIN becomes the maximum value (+6 dB) at f = 0 [Hz] and f = 1 / (T ± Ts) [Hz]. When f = 1/2 (T ± Ts) [Hz], the GAIN becomes the minimum value (zero), and the outputs of the adders 7 and 8 correspond to the input phase difference as shown by the broken line 32 and the alternate long and short dash line 33. There is a difference and this becomes stereo separation.

  If the stereo calculation processing band is set to f = 1 / 2T [Hz] or less, the high frequency band has a attenuated frequency characteristic.

  2 and 3, for example, if the delay T is set to 50 [μS], the frequency 1 / 2T becomes 10 [kHz], and the above-described wind noise band is sufficiently lower than the stereo calculation processing band. 2 and the frequency characteristics in the case of subtraction applied to the conventional example of FIG. 2 need to increase the gain to near 0 dB with the EQ in the subsequent stage, but the wind noise is deteriorated. In the frequency characteristics when adding, the low frequency does not need to be further increased by the subsequent EQ, and conversely, if necessary, the gain is reduced to near 0 dB. In either case, wind noise deteriorates. There is an advantage to be improved without.

  Therefore, in the stereo calculation processing of the stereo microphone device according to the embodiment applied to the present invention, it is possible to perform the stereo calculation processing while suppressing wind noise. However, the problem in FIG. 3 is that the null frequency at which GAIN is −∞. The signal delay Ts is within the stereo processing band and the signal delay Ts varies depending on the sound source direction. In other words, if the delay T = 50 [μS] is set as described above, the frequency 1 / 2T becomes 10 [kHz], a null point is generated in the stereo calculation processing band, and the frequency 1 / 2T point is further increased by Ts. , 1/2 (T + Ts) and 1/2 (T-Ts). Further, if the delay T is further reduced, the null point is shifted to a high frequency, and this inconvenience can be avoided. However, the stereo separation cannot be increased and the purpose of the stereo operation processing cannot be achieved.

  Therefore, in the stereo calculation processing of the stereo microphone device according to the embodiment applied to the present invention, the stereo calculation processing band is divided, and the wind noise band has the frequency characteristics in the case of addition applied to the present invention in FIG. The above-mentioned problem is solved by using the frequency characteristics in the case of subtraction applied to the conventional example of FIG.

First, FIG. 5 shows a stereo calculation block example 1 of a stereo microphone device according to an embodiment applied to the present invention.
Rch and Lch audio signals are input from the microphones 51 and 52, respectively, and the signal levels are optimized by the AMPs 53 and 54. The Rch signal is added to the + terminal of the adder 65, the band dividing means (2) 55 and the band dividing means (1). 56. Similarly, the Lch signal from the AMP 54 is input to the + terminal of the adder 66, the band dividing unit (1) 57 and the band dividing unit (2) 58. Here, the band dividing means (1) 56 and the band dividing means (1) 57 extract the same band 1 signal, but this band 1 is set to the wind noise band shown in FIG. Similarly, the band dividing means (2) 55 and the band dividing means (2) 58 extract the same band 2 signal, and this is set to the second band other than the band 1.

  Then, the Rch band 2 signal divided from the band dividing means (2) 55 is delayed via the delay unit DL (2) 59, and the Rch band 1 signal divided from the band dividing means (1) 56 is delayed to the delay unit DL ( 1) Delayed through 60. Further, the Lch band 1 signal divided from the band dividing means (1) 57 is delayed via the delay unit DL (1) 61, and the Lch band 2 signal divided from the band dividing means (2) 58 is delayed by the delay unit DL. (2) Delayed via 62.

  The delayed Rch band 1 signal is input to the addition input terminal of the adder 63, the Rch band 2 signal is input to the subtraction input terminal of the adder 63, the Rch band 1 signal is added by the adder 63, and the Rch band The two signals are subtracted. The adder 66 adds the Lch signal and the Rch band 1 signal, subtracts the Rch band 2 signal, and outputs the result from the terminal 68 as an Lch signal.

  Similarly, the delayed Lch band 1 signal is input to the addition input terminal of the adder 64, the Lch band 2 signal is input to the subtraction input terminal of the adder 64, and the Lch band 1 signal is added by the adder 64. The Lch band 2 signal is subtracted. The adder 65 adds the Rch signal and the Lch band 1 signal, subtracts the Lch band 2 signal, and outputs the result from the terminal 67 as the Rch signal.

Here, the band dividing means in FIG. 5 will be further described with reference to FIG.
FIG. 4A shows an example of band division by BPF (Band Pass Filter). Band (1) 41 is extracted by BPF1, and band (2) 42 is extracted by BPF2. Here, the band 1 which is the wind noise band and the other band 2 in the stereo calculation block example 1 of FIG. 5 correspond to the band (1) 41 and the band (2) 42. FIG. 4B shows an example of band division by LPF (Low Pass Filter). Band (1) 44 is extracted by LPF (1) 43, and the output of LPF (1) 43 is subtracted from the output of LPF (2) 45. By doing so, the band (2) 46 is extracted. As described above, the band (1) 44 is set to the band 1 which is the wind noise band in the stereo computation block example 1 of FIG. 5 and the band (2) 46 is set to the other band 2.

Next, FIG. 6 shows an example of frequency characteristics in the stereo calculation processing of the stereo microphone device of the embodiment applied to the present invention shown in FIG.
First, in FIG. 5, the delay frequency DL (1) 60, DL (1) 61 delays T1 [sec], and the output frequency is calculated by the adder 63, the adder 64, the adder 65, and the adder 66. A characteristic example is shown, and T2 [sec] is obtained by using a band (1) 77 of a band determined by a frequency 1 / 2T1 [Hz] or less having a period of T1 and further delay elements DL (2) 59 and DL (2) 62. In the example of the output frequency characteristic when the adder 63, the adder 64, the adder 65, and the adder 66 are operated, the frequency 1 / 2T2 [Hz] having a period of T2 and the 1 / 2T1 [Hz] The band is divided into the band (2) 78 of the band determined in (1).

  Further, solid lines 71 and 74 indicate a case where there is no phase difference between the inputs of the microphone 51 and the microphone 52, and broken lines 72 and 75 and alternate long and short dash lines 73 and 76 indicate a case where there is a phase difference between the inputs of the microphone 51 and the microphone 52, respectively. Therefore, as shown in FIG. 6, if the delays T1 and T2 are determined so that the frequency characteristics become flat when the band (1) 77 and the band (2) 78 are added, it is necessary to increase the gain in the wind noise band. Therefore, the stereo characteristics with reduced wind noise can be obtained, the conventional stereo characteristics can be obtained outside the wind noise band, and the frequency characteristics are flat, so that the subsequent EQ can be simplified.

Further, FIG. 7 shows and describes a stereo calculation block example 2 of the stereo microphone device according to the embodiment applied to the present invention.
The microphones 81, 82, and 83 are omnidirectional microphones arranged in an equilateral triangle, for example. Rch, Lch, and Cch audio signals are input, and the signal levels are optimized by the AMPs 84, 86, and 85, and the Rch signals are added. Similarly, the Lch signal is input to the addition input terminal of the adder 93, the Cch signal is input to the band dividing means (2) 87 and the band dividing means (1) 88, and the band is input. In the dividing means (1) 88, the band 1 is extracted as in FIG. 5 and set to the wind noise band shown in FIG. In the band dividing means (2) 87, band 2 is extracted and set to band 2 excluding band 1.

  Then, the Cch band 2 signal divided from the band dividing means (2) 87 is delayed through a delay unit DL (2) 89, and the Cch band 1 signal divided from the band dividing means (1) 88 is sent to a delay unit DL ( 1) Delayed via 90.

  The delayed Cch band 1 signal is input to the addition input terminal of the adder 91, the Cch band 2 signal is input to the subtraction input terminal of the adder 91, the Cch band 1 signal is added by the adder 91, and the Cch band The two signals are subtracted. The adder 92 adds the Rch signal and the Cch band 1 signal, subtracts the Cch band 2 signal, and outputs the result from the terminal 94 as the Rch signal. The adder 93 adds the Lch signal and the Cch band 1 signal, subtracts the Cch band 2 signal, and outputs the result from the terminal 95 as the Lch signal.

  The stereo effect can be obtained while obtaining the same effect as in FIG. 5 even when there are two or more microphones as in the stereo operation block example 2 of the stereo microphone device of the embodiment applied to the present invention in FIG. . In the present invention, ATT and EQ are not necessarily required.

  The present invention can be applied to a case where a stereo sound field is generated by a built-in microphone of a video camera.

It is a stereo calculation principle figure of the stereo microphone apparatus of embodiment applied to this invention. It is a figure which shows the frequency characteristic example in subtraction. It is a figure which shows the frequency characteristic example in addition. FIGS. 4A and 4B are diagrams showing band division. FIG. 4A shows band division by BPF, and FIG. 4B shows band division by LPF. It is a figure which shows the stereo calculation block example 1 of the stereo microphone apparatus of embodiment applied to this invention. It is a figure which shows the example of a frequency characteristic in the stereo calculation process of the stereo microphone apparatus of embodiment applied to this invention. It is a figure which shows the stereo calculation block example 2 of the stereo microphone apparatus of embodiment applied to this invention. It is a figure which shows the conventional stereo calculation block. It is a figure which shows the stereo directional characteristic in a video camera, Fig.9 (a) is before a stereo calculation process, FIG.9 (b) is after a stereo calculation process. It is a figure which shows the frequency characteristic example of the wind noise signal in a video camera.

Explanation of symbols

  1, 2, 3, 3, AMP, 5, 6, DL, 7, 8, adder, 9, 10, terminal, 21 to 23, subtraction frequency characteristic, 31 to 33, addition frequency characteristic, 41, band DESCRIPTION OF SYMBOLS 1, 42 ... Band 2, 43 ... LPF1, 44 ... Band 1, 45 ... LPF2, 46 ... Band 2, 51, 52 ... Microphone, 53, 54 ... AMP, 56, 57 ... Band division means (1), 55, 58 ... Band division means (2), 60, 61 ... Delayer DL (1), 59, 62 ... Delayer DL (2), 63 to 66 ... Adder, 67, 68 ... Terminal, 71 to 73 ... Subtraction frequency Characteristic, 74 to 76 ... Addition frequency characteristic, 77 ... Band 1, 78 ... Band 2, 81, 82, 83 ... Microphone, 84, 85, 86 ... AMP, 88 ... Band division means (1), 87 ... Band division means (2), 90 ... delay device DL (1), 89 ... delay device DL ( ), 91 to 93 ... adder, 94, 95 ... terminal

Claims (4)

In a stereo microphone device that outputs a directional stereo sound field signal from a plurality of omnidirectional microphones,
Band dividing means for dividing an audio band from at least one omnidirectional microphone of the plurality of omnidirectional microphones into a plurality of bands;
Further, the first band output via the first delay means for time delaying the first band signal from the band dividing means is converted into at least one non-directional microphone other than the omnidirectional microphone divided by the band dividing means. Adding means for adding to the output of the directional microphone;
Further, the second band output via the second delay means for delaying the second band signal from the band dividing means is converted into at least one non-directional microphone other than the omnidirectional microphone divided by the band dividing means. A stereo microphone device comprising subtracting means for subtracting from the output of a directional microphone. The stereo microphone device according to claim 1, wherein
The stereo microphone device according to claim 1, wherein the band dividing unit divides the first band including wind noise measured in advance and the second band other than the first band. In a stereo calculation method in a stereo microphone device that outputs a directional stereo sound field signal from a plurality of omnidirectional microphones,
Dividing a voice band from at least one omnidirectional microphone of the plurality of omnidirectional microphones into a plurality of bands by a band dividing unit;
Further, the first band output via the first delay means for time delaying the first band signal from the band dividing means is converted into at least one non-directional microphone other than the omnidirectional microphone divided by the band dividing means. Adding the output of the directional microphone and the adding means;
Further, the second band output via the second delay means for delaying the second band signal from the band dividing means is converted into at least one non-directional microphone other than the omnidirectional microphone divided by the band dividing means. A stereo calculation method comprising a step of subtracting from an output of a directional microphone by a subtracting means. The stereo calculation method according to claim 3,
The band dividing means divides into a first band including wind noise measured in advance and a second band other than the first band.

Images