JP4483468B2 - Noise reduction circuit, electronic device, noise reduction method - Google Patents

Description

  In particular, the present invention relates to a noise reduction circuit that reduces noise mixed in an audio signal from a microphone, an electronic apparatus including the noise reduction circuit, and a noise reduction method. Specifically, the noise reduction circuit has a low level near the same level as the noise level. The present invention relates to a noise reduction circuit, an electronic device, and a noise reduction method capable of reducing noise while suppressing influence on audio components.

  In an electronic device including a microphone such as a video camera, a digital camera, or an IC (integrated circuit) recorder, noise mixed in an audio signal is roughly classified into external noise and internal noise. External noise includes noise caused by mechanical operations such as mechanical components installed in the vicinity of the microphone, noise generated by noise sources, noise caused by various switch operations, and other external factors. . Internal noise includes so-called sensor components such as microphones, thermal noise generated from resistors and semiconductor elements that make up the pre-amplifier (preamplifier), etc. These are called white noises, and equally include broadband frequency components. Random noise.

  Recent video cameras are equipped with multiple microphones, and stereo sound field processing is performed on the output from each microphone, and white noise mixed in the audio signal of each microphone is emphasized by the stereo sound field processing. There was a problem that.

FIG. 15 shows a stereo sound field processing circuit of a video camera according to Patent Document 1, for example.
Japanese Patent No. 2946638

  This circuit includes delay circuits 5 and 6 in which an audio signal output from one of the two omnidirectional microphones 1 and 2 is delayed for a time corresponding to the distance between the two microphones 1 and 2, and a delay circuit Adders 10 and 9 in which the audio signals delayed by 5 and 6 are superimposed with the polarity reversed through attenuators 7 and 8, and an equalizer 12 that adjusts the frequency characteristics of the output signals from the adders 10 and 9, 11.

  For example, when the microphone 1 for the R channel is closer to the sound source, the time for the sound wave to reach the microphone 2 for the L channel is delayed from the arrival to the microphone 1. Therefore, the output signal of the microphone 1 via the amplifier 3 is delayed by the delay circuit 5 by the delay time, and the delayed signal is added by subtracting it from the output signal of the microphone 2 via the amplifier 4. The output of the device 10 is substantially cancelled. On the contrary, when the sound source is close to the microphone 2, the output of the R channel is canceled. As a result, a stereo feeling (level difference between right and left) can be obtained.

  That is, the stereo sound field processing circuit is a circuit that emphasizes a slight phase difference. When there is no phase difference between the audio signals of the microphones 1 and 2, the arithmetic processing in the circuit is performed as shown in FIG. The later directivity pattern shows a monaural characteristic in which L (solid line) and R (dashed line) substantially match. When there is a phase difference between the two, as shown in FIG. In the directivity pattern, L (solid line) and R (broken line) are separated in the left-right direction to show stereo characteristics.

  White noise generated in each of the L channel and the R channel is generated by a random phase difference. Therefore, the random noise is stereo-processed by the stereo sound field processing circuit, and when the input signals are added in phase by the adders 9 and 10, for example. The level deteriorates, and the equalizers 11 and 12 in the subsequent stages increase the gain of the noise band. In addition, since the stereo characteristics as shown in FIG. 16B are present, noise spreads over the entire sound field, and the generated white noise is emphasized even at a very small level.

  External noise can be reduced by a structural method such as acoustic isolation of the microphone and the noise source, or by an electrical method using an adaptive filter, etc. Since noise includes frequency components of audible sound evenly, it cannot be easily removed by a filter that removes only a specific band, and the audio component is also removed at the same time.

  In addition, the generation of internal noise can be suppressed by using expensive parts other than general-purpose parts, such as dedicated elements and semiconductor parts designed to reduce thermal noise (white noise). It is difficult to adopt it especially for consumer devices because of its scale.

Therefore, it is preferable to reduce noise by electrical signal processing. For example, in the noise reduction circuit disclosed in Patent Document 2, a speech component and noise are distinguished based on the (amplitude) level of the signal, and if the level is equal to or lower than a predetermined level, there is no speech component and only noise. Therefore, the signal is not output or the signal level is reduced.
Japanese Patent Laid-Open No. 7-44996

  Even if the predetermined level is set as a guideline for discriminating between audio components and noise, signals below the predetermined level are not limited to noise, and may include audio components with low levels. Therefore, as in Patent Document 2, when the signal is below a predetermined level, if the signal is not output or the signal level is reduced, even the audio component included below the predetermined level is removed or leveled. As a result, the sound component with a low level becomes inaudible or difficult to hear.

  The present invention has been made in view of the above-described problems, and an object thereof is to provide a noise reduction circuit, an electronic device, and a noise reduction method capable of reducing noise while suppressing an influence on a sound component having a small level close to the noise level. It is in.

  The noise reduction circuit of the present invention extracts N (N is a positive number greater than or equal to 2) audio channels and low-level components that extract low-level components below a predetermined limiter level from the audio signals of the N audio channels. Extraction means and all the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are converted into the audio signals of the selected one audio channel. And a calculating means for adding.

  The electronic apparatus according to the present invention extracts a low-level component that extracts N level (N is a positive number of 2 or more) audio channels and low-level components below a predetermined limiter level from the audio signals of each of the N audio channels. All of the low level components extracted from the audio signals of the N-1 audio channels other than the one audio channel selected from the component extraction means and the N audio channels are audio signals of the selected one audio channel. And a noise reduction circuit having an arithmetic means for adding to.

  The low level component extracted by the low level component extraction means includes most of the noise component and part of the audio component. The noise components included in the audio signal in each audio channel (audio signal transmission path) are random noises that are not correlated with each other as compared to the audio components. Decrease. That is, random noise is added to the audio signal of a certain audio channel by adding the low level component (noise component) of another audio channel different from this audio channel. The level of the component can be reduced.

  Even if the low-level component extracted by the low-level component extracting means includes a low-level audio component in addition to the noise component, the low-level component is removed from each audio signal as in the past. Instead, since the low level components extracted from the audio signals of different audio channels are added, the audio components included in the low level components are not removed. Therefore, only the noise component can be reduced without significantly affecting the reproduction of a small sound.

  In general, when N random noises included in the low-level component are averaged, the level is reduced to 1 / √N.

  In addition, an audio signal level detection unit for detecting a total level of audio signals from a plurality of audio channels is provided, and further, in the low level component extraction unit according to the total level of the audio signal detected by the audio signal level detection unit. If limiter level changing means that changes the limiter level is provided, the noise level is reduced by reducing the limiter level when the level of the audio signal is so large that noise is not a concern and it is not necessary to reduce the noise so much. It is possible to suppress the influence on the sound component of a small level near the noise level by eliminating or reducing the degree of reduction.

  In addition, an audio signal level detecting unit for detecting a total level of a plurality of audio signals is provided, and a low level component extracted by a low level component extracting unit is further selected in accordance with the total level detected by the audio signal level detecting unit. If a low-level component ratio variable means is provided that changes the ratio of multiple low-level components included in the audio signal to which the noise is added, the level of the audio signal is so large that noise is not an issue, and noise reduction is required. In the case where there is no noise, by changing the ratio between the low-level components, noise reduction is not performed or the degree of reduction is reduced, so that the influence on a low-level sound component near the noise level can be suppressed.

  In addition, if the configuration is provided with noise gate means for receiving an output signal of the arithmetic means and outputting an output level lower than the input level for low level components below a predetermined gate level, Therefore, the level can be further reduced with respect to the low level component whose level has been reduced, and the noise reduction effect can be further enhanced.

  Further, it is configured to include audio signal level detection means for detecting the total level of a plurality of audio signals, and gate level variable means for changing the gate level according to the total level detected by the audio signal level detection means. For example, if the level of the audio signal is so high that you don't mind the noise and you don't need to reduce the noise so much, do not reduce the noise by reducing the gate level, or reduce the degree of reduction. It is possible to suppress the influence on the sound component of a small level near the noise level.

  The noise reduction method of the present invention includes a step of extracting low level components below a predetermined limiter level from the audio signals of N (N is a positive number of 2 or more) audio channels, and N audio channels. Adding all of the low level components extracted from the audio signals of the N-1 audio channels other than one audio channel selected from the audio signals of the selected one audio channel. It is said.

  The low level component includes most of the noise component and a part of the audio component. The noise components included in the audio signal in each audio channel (audio signal transmission path) are random noises that are not correlated with each other compared to the audio components, and adding these random noises reduces the level of the audio components. To do. That is, random noise is added to the audio signal of a certain audio channel by adding the low level component (noise component) of another audio channel different from this audio channel. The level of the component can be reduced.

  Even if the low-level component includes a low-level audio component in addition to the noise component, the low-level component is not removed from each audio signal as in the conventional case, but the audio of different audio channels is used. Since the low level components extracted from the signal are added, the audio component included in the low level component is not removed. Therefore, only the noise component can be reduced without significantly affecting the reproduction of a small sound.

  According to the noise reduction circuit of the present invention, it is possible to reduce noise while avoiding that an audio component having a level near the noise level is difficult to hear or not being heard, and to play and listen to a necessary audio signal clearly. be able to. In addition, expensive dedicated parts and circuits designed so as not to generate noise are unnecessary, and an increase in circuit scale can be suppressed and costs can be reduced.

  According to the electronic device of the present invention, it is possible to reduce noise while avoiding that the sound component having a level near the noise level is difficult to hear or not being heard, and to reproduce and listen to the necessary audio signal clearly. Can improve the quality of electronic equipment. In addition, expensive dedicated parts and circuits designed so as not to generate noise are unnecessary, and an increase in the size of the electronic device can be suppressed and cost can be reduced.

  According to the noise reduction method of the present invention, it is possible to reduce noise while avoiding that an audio component having a level near the noise level is difficult to hear or not being heard, and to reproduce and listen to a necessary audio signal clearly. be able to.

[First Embodiment]
FIG. 1 shows a noise reduction circuit according to a first embodiment of the present invention. This noise reduction circuit has three audio channels (audio signal transmission paths) corresponding to the three microphones 51, 52, and 53. In addition, three arithmetic means 61, 62, and 63 are provided corresponding to the three microphones 51, 52, and 53, respectively. Are connected to amplifiers 54, 55 and 56, respectively. Furthermore, low level component extraction means 71 is provided for extracting low level components below a predetermined limiter level from the output signals from the amplifiers 54, 55, 56.

  Each microphone 51, 52, 53 converts an audible sound into an audio signal that is an electrical signal. Each of the microphones 51, 52, and 53 is described as being omnidirectional, but may be directional. The input to the noise reduction circuit is not limited to the microphone, and may be an audio channel output from another audio output device.

  An example of a specific configuration of the low level component extraction means 71 is shown in FIG. The low level component extracting unit 71 includes an absolute value converting unit 104 that converts the signal levels of the output signals from the microphones 51, 52, and 53 through the amplifiers 54, 55, and 56 into absolute values; A level detector 105 that receives the output and detects a peak value or average value of the signal level of each output signal from each of the microphones 51, 52, 53, a peak value or average value detected by the level detector 105, and a predetermined value The comparison means 106 for comparing the limiter level 103 and the switch 107 for switching whether the terminal 107c is connected to the terminal 107a or the terminal 107b according to the output of the comparison means 106 are provided.

  However, the present invention is not limited to this, and a low level component can be extracted from the output signals from the microphones 51, 52, and 53 with the same configuration as that shown in FIG. For example, in an analog circuit, the absolute value converting means 104 can be composed of a rectifier circuit, the level detecting means 105 can be composed of a detection circuit such as a low-pass filter, and the comparing means 106 can be composed of a comparator.

  Returning to FIG. 1, the calculation means 61 adds the low level components extracted from the output signals of the other amplifiers 55 and 56 by the low level component extraction means 71 to the output signal of the amplifier 54, and further adds all the low level components. Performs averaging of level components. Similarly, the calculation means 62 adds the low level components extracted from the output signals of the other amplifiers 54 and 56 by the low level component extraction means 71 to the output signal of the amplifier 55, and further adds all the low level components. Perform averaging. Similarly, the calculation means 63 adds the low level components extracted from the output signals of the other amplifiers 54 and 55 by the low level component extraction means 71 to the output signal of the amplifier 56, and further adds all the low level components. Perform averaging.

  Next, the operation of the noise reduction circuit configured as described above will be described.

  The audio signals output from the microphones 51, 52, and 53 are amplified to optimum levels for subsequent processing by the amplifiers 54, 55, and 56, respectively. Input to the extraction means 71.

  The low level component extraction means 71 receives audio signals from the respective microphones 51, 52, 53 via the amplifiers 54, 55, 56, and extracts low level components below a predetermined limiter level. As shown in FIG. 13A, the limiter level is set in the vicinity of the level of white noise mixed in the audio signal, and the low level component extracting means 71 has a period during which an audio signal exceeding the limiter level exists. As shown in (b), an operation for making a level 0 no signal is performed, and white noise at a minute level with respect to the audio signal level is extracted.

  Specifically, for example, the audio signal of the microphone 51 via the amplifier 54 is input to the terminal 107a of the switch 107 and the absolute value converting means 104 shown in FIG.

  In the absolute value converting means 104, the audio signal level from the microphone 51 is converted into an absolute value, and the output is input to the level detecting means 105. The level detection means 105 detects the peak value or average value of the audio signal level that has been converted to an absolute value, and the comparison means 106 compares the value with the limiter level set from the terminal 103.

  The comparison means 106 connects the terminal 107c of the switch 107 to the terminal 107b at the timing of the input signal having a level larger than the limiter level, and connects the terminal 107c of the switch 107 to the terminal at other timings, that is, the timing of the input signal below the limiter level. 107a is connected.

  Since the terminal 107b of the switch 107 is connected to the ground, the level of the output signal from the terminal 107c is 0 at the timing of the input signal higher than the limiter level, and the limiter is at the timing of the input signal below the limiter level. The signal below the level is output as it is.

  Similarly, for the audio signals from the microphones 52 and 53 via the other amplifiers 55 and 56, the low level component extraction means 71 extracts the low level components included in the respective audio signals.

  The calculation means 61 includes a sound signal from the microphone 51 via the amplifier 54, a noise component n2 included in the sound signal from the microphone 52 extracted by the low level component extraction means 71, and a low level component extraction means 71. The noise component n3 included in the extracted sound signal from the microphone 53 is input, the noise component n1 included in the sound signal from the microphone 51, the noise component n2 included in the sound signal from the microphone 52, and the microphone 53. Is added and averaged with the noise component n3 included in the audio signal.

  Further, the arithmetic means 61 extracts the low level audio component s2 ′ below the limiter level contained in the audio signal from the microphone 52 extracted by the low level component extracting means 71 and the low level component extracting means 71. The low level audio component s 3 ′ below the limiter level included in the audio signal from the microphone 53 is input, and the low level audio component s 1 ′ below the limiter level included in the audio signal from the microphone 51 and the microphone 52. The low-level sound component s2 ′ included in the sound signal and the low-level sound component s3 ′ included in the sound signal from the microphone 53 are added and averaged.

  As a result, the calculation means 61 outputs s1 + {(n1 + n2 + n3) / 3} + {(s1 ′ + s2 ′ + s3 ′) / 3}. Note that s1 in this output represents an audio component having a level higher than that of the above-described level limiter in the audio signal from the microphone 51.

  Similarly, the calculating means 62 extracts the sound signal from the microphone 52 via the amplifier 55, the noise component n1 contained in the sound signal from the microphone 51 extracted by the low level component extracting means 71, and the low level component extraction. The noise component n3 included in the audio signal from the microphone 53 extracted by the means 71 is input, the noise component n1 included in the audio signal from the microphone 51, and the noise component n2 included in the audio signal from the microphone 52 Then, an averaging process with the noise component n3 included in the audio signal from the microphone 53 is performed.

  Further, the arithmetic means 62 extracts the low level audio component s1 ′ below the limiter level included in the audio signal from the microphone 51 extracted by the low level component extracting means 71 and the low level component extracting means 71. The low level audio component s 3 ′ below the limiter level included in the audio signal from the microphone 53 is input, and the low level audio component s 2 ′ below the limiter level included in the audio signal from the microphone 52 and the microphone 51. The averaging process of the low level audio component s1 ′ included in the audio signal and the low level audio component s3 ′ included in the audio signal from the microphone 53 is performed.

  As a result, the calculation means 62 outputs s2 + {(n1 + n2 + n3) / 3} + {(s1 ′ + s2 ′ + s3 ′) / 3}. Note that s2 in this output represents an audio component having a level higher than that of the above-described level limiter in the audio signal from the microphone 52.

  Similarly, the computing unit 63 extracts the noise signal n1 included in the audio signal from the microphone 53 via the amplifier 56, the audio signal from the microphone 51 extracted by the low level component extracting unit 71, and the low level component extraction. The noise component n2 included in the audio signal from the microphone 52 extracted by the means 71 is input, the noise component n1 included in the audio signal from the microphone 51, and the noise component n2 included in the audio signal from the microphone 52 Then, an averaging process with the noise component n3 included in the audio signal from the microphone 53 is performed.

  Further, the arithmetic means 63 extracts the low level audio component s1 ′ below the limiter level included in the audio signal from the microphone 51 extracted by the low level component extracting means 71 and the low level component extracting means 71. The low level audio component s 2 ′ below the limiter level included in the audio signal from the microphone 52 is input, and the low level audio component s 3 ′ below the limiter level included in the audio signal from the microphone 53 and the microphone 51. The low-level audio component s1 ′ included in the audio signal and the low-level audio component s1 ′ included in the audio signal from the microphone 51 are added and averaged.

  As a result, the calculation means 63 outputs s3 + {(n1 + n2 + n3) / 3} + {(s1 '+ s2' + s3 ') / 3}. Note that s3 in this output represents an audio component having a level higher than that of the above-described level limiter in the audio signal from the microphone 53.

  In other words, the noise components n1, n2, and n3 are white noises that are not correlated with each other (the phases are different). In general, the level is reduced to 1 / √N by averaging these N random waveforms. Is done. In this embodiment, it can be reduced to 1 / √3. On the other hand, the low-level sound components s1 ′, s2 ′, and s3 ′ are normally emitted from sound sources that are sufficiently separated from the intervals between the microphones 51, 52, and 53, and thus have a correlation, and are averaged. However, the level does not decrease.

  Note that the number of microphones is not limited to three, and the level can be reduced to 1 / √N by adding and averaging the white noise from each of the N microphones as described above, so the number of microphones is large. The more effective the noise reduction effect is. Particularly in recent video cameras, there is an advantageous background in applying the present invention that the number of mounted microphones tends to increase in order to obtain a multi-surround sound field.

  As described above, in this embodiment, the noise level is reduced by adding and averaging the white noise included in each audio channel (transmission path of the audio signal from each microphone). The low level component is not removed from the audio signal as in the prior art. For this reason, an audio component other than the white noise at a level equal to or lower than the limiter level is not removed, and the influence on a small level sound close to the white noise level can be suppressed.

  Also, as described above, since the white noise components included in each audio channel are the same waveform, that is, the same waveform of the same level and the same phase as {(n1 + n2 + n3) / 3}, that is, a monaural signal, the above-described FIG. The white noise component is not enhanced even if the stereo operation processing circuit for enhancing the signal phase difference between the audio channels shown in FIG.

[Second Embodiment]
FIG. 2 shows a noise reduction circuit according to the second embodiment of the present invention. This is a case where there are two microphones (that is, two audio channels) in the noise reduction circuit shown in the first embodiment. More concrete.

  The noise reduction circuit according to the present embodiment includes two adders 27 and 28 provided corresponding to the two microphones 21 and 22, respectively, and an amplifier 23 and a delay connected between the microphone 21 and the adder 27. And an amplifier 24 and a delay unit 26 connected between the microphone 22 and the adder 28. Further, an adder 30 that receives the output of the amplifier 23 at the-terminal and an output of the amplifier 24 at the + terminal, and a low level component extraction means for extracting a low level component lower than a predetermined limiter level from the output signal of the adder 30 31, and an amplifier 32 that multiplies the output of the low-level component extraction means 31 by a coefficient and outputs the result to the + terminal of the adder 27 and the − terminal of the adder 28.

  In the present embodiment, the adder 30, the amplifier 32, the adder 27, and the adder 28 correspond to the calculation unit of the first embodiment.

  Each of the microphones 21 and 22 converts an audible sound into an audio signal that is an electric signal. The microphones 21 and 22 are described as being non-directional, but may be directional. The input to the noise reduction circuit is not limited to the microphone, and may be an audio channel output from another audio output device.

  The output signal of the microphone 21 is input to one plus terminal of the adder 27 via the amplifier 23 and the delay unit 25, and is also input to the minus terminal of the adder 30 via the amplifier 23. The output signal of the microphone 22 is input to the + terminal of the adder 28 via the amplifier 24 and the delay unit 26, and is also input to the + terminal of the adder 30 via the amplifier 24.

  From the adder 30, a difference signal between the output signal of the microphone 22 and the output signal of the microphone 21 is output. This difference signal includes a difference signal of audio components of the microphones 22 and 21 and a difference signal of white noise. Here, for example, in the case of a video camera, the subject that is the sound source of the sound component is almost always far away from the interval between the two microphones 22 and 21, and is relatively equal to the sound source. Since it is located at a distance, the correlation is high, and when the adder 30 subtracts both, the audio signal is canceled, and the output signal of the adder 30 provides many white noise signal components.

  The output signal of the adder 30 is input to the low level component extraction means 31. The low level component extraction unit 31 has the configuration shown in FIG. 10, for example, as in the first embodiment, and extracts a low level component equal to or lower than a predetermined limiter level included in the input signal. FIG. 13A shows an input signal to the low level component extraction unit 31, and FIG. 13B shows an output signal from the low level component extraction unit 31. As shown in FIG. 13A, the limiter level is set in the vicinity of the level of white noise mixed in the audio signal, and the low level component extraction means 31 is in FIG. 13 during a period in which there is an audio signal exceeding the limiter level. As shown in (b), an operation for making a level 0 no signal is performed to extract white noise at a minute level with respect to the audio signal.

  The output signal from the low level component extraction means 31 is input to the amplifier 32, multiplied by a coefficient 0.5 by the amplifier 32, and input to the other + terminal of the adder 27 and the-terminal of the adder 28, respectively. The

  Here, if the audio component included in the signal output from the microphone 21 is Rs, the noise component is Rn, the audio component included in the signal output from the microphone 22 is Ls, and the noise component is Ln, an adder 27, The outputs Ra and La from 28 are expressed by the following equations (1) and (2), respectively.

Ra = (Rs + Rn) +0.5 (Ln-Rn) = Rs + 0.5 (Ln + Rn) (1)
La = (Ls + Ln) −0.5 (Ln−Rn) = Ls + 0.5 (Ln + Rn) (2)

  That is, in the adder 27, the signal (Rs + Rn) from the microphone 21 via the amplifier 23 and the delay unit 25 is extracted by the low level component extraction means 31 and multiplied by the coefficient 0.5 by the amplifier 32. 5 (Ln−Rn) is added, and the adder 28 extracts the signal (Ls + Ln) from the microphone 22 via the amplifier 24 and the delay unit 26 and the low-level component extraction means 31, and the coefficient in the amplifier 32. {-0.5 (Ln−Rn)} multiplied by 0.5 is added, and as a result, the low level components of Ra and La are both 0.5 (Ln + Rn).

  The low-level components Ln and Rn correspond to white noise having no correlation with each other (the phases are different), and 0.5 (Ln + Rn) is an addition average of Ln and Rn. Therefore, as described in the first embodiment, generally, when N random waveforms are averaged, the level can be reduced to 1 / √N. Therefore, in this embodiment, the level of white noise is reduced to 1 / √N. It can be reduced to √2.

  Further, in the present embodiment, the arithmetic means composed of the adder 30, the amplifier 32, the adder 27, and the adder 28 performs the above-described averaging process only for the noise components Ln and Rn. That is, since the audio components Ls and Rs from the microphones 21 and 22 are correlated with each other (substantially in phase), the noise components Ln and Rn are not correlated with each other. Therefore, the adder 30 performs (Ls + Ln) and (Rs + Rn). ) Is extracted, the noise component is extracted, the audio component is not extracted, and the influence on the audio component can be reduced.

  As described above, in the present embodiment, the noise level is reduced by using the randomness of white noise included in each audio channel (the transmission path of the audio signal from each microphone 21 and 22) and averaging them. Therefore, the low level component is not removed from the audio signal as in the prior art. For this reason, an audio component other than the white noise at a level equal to or lower than the limiter level is not removed, and the influence on a small level sound close to the white noise level can be suppressed.

  Further, as described above, since the white noise components included in the audio signals of the respective audio channels are the same waveform, that is, the same phase, that is, monaural signals such as 0.5 (Ln + Rn), the above-described FIG. Even if the signal of each channel is subjected to stereo calculation processing by the stereo sound field processing circuit that emphasizes the phase difference of the signal between the channels, the white noise component is not enhanced.

  The delay units 25 and 26 are for adjusting the timing of signals to be added by the adders 27 and 28, respectively, and are not necessarily required.

[Third Embodiment]
FIG. 3 shows a noise reduction circuit according to the third embodiment of the present invention. The noise reduction circuit of the present embodiment has a configuration in which an audio signal level detection means 70 and a limiter level variable means 73 are added to the noise reduction circuit of the first embodiment, and is the same as the first embodiment. The same reference numerals are given to the constituent parts, and detailed description thereof will be omitted.

  In each of the above embodiments, the low level component of each audio channel after the averaging in the computing means becomes a monaural signal having the same waveform at the same phase, the same level, and therefore, a slight low level audio component is included below the limiter level. Even if the signal of each audio channel is processed in stereo by the stereo sound field processing circuit that emphasizes the signal phase difference between the audio channels shown in FIG. For the following voice composition, the stereo effect cannot be obtained.

  Therefore, in the present embodiment, by utilizing the masking effect of human hearing, when the sound signal from the microphone is at a relatively large level, noise is not reduced or the degree of noise reduction is lowered. This is because human hearing is not aware of the presence of a small sound (here, white noise) that is present in a relatively loud sound so that, for example, it becomes difficult to hear a human voice in a loud noise. For this reason, in this embodiment, the total level of the audio signal from each microphone is detected, and the limiter level is changed in accordance with the detected level so as not to reduce noise more than necessary. Try to reduce the impact on the signal.

  The total level of the audio signals from the microphones 51, 52, 53 is detected by the audio signal level detection means 70. Specifically, the absolute value converting means 114 shown in FIG. 11 converts the sum level of the audio signals from the microphones 51, 52, and 53 into an absolute value, and the output is input to the level detecting means 115. The level detection means 115 detects the peak value or average value of the sum level of the audio signal converted into an absolute value.

  A coefficient is generated in the limiter level varying means 73 according to the level, and the limiter level in the low level signal extracting means 71 is set according to the coefficient. For example, when the sum level of the audio signal is relatively large and white noise is not so noticeable, the limiter level shown in FIG. 13A is set to 0, that is, the low level component extracted by the low level signal extraction means 71. Is set to 0 so that the output signals from the microphones 51, 52, and 53 via the amplifiers 54, 55, and 56 are output as they are from the arithmetic means 61, 62, and 63, respectively.

  Alternatively, the limiter level is reduced as the sum level of the audio signal increases, and the proportion of low-level audio components that are monophonic with white noise is reduced.

[Fourth Embodiment]
FIG. 4 shows a noise reduction circuit according to the fourth embodiment of the present invention, which is a more specific example in the case of two microphones in the noise reduction circuit shown in the third embodiment.

  The noise reduction circuit according to the present embodiment has a configuration in which an adder 29, a level detection unit 35, a comparison unit 36, and a coefficient generation unit 37 are added to the noise reduction circuit according to the second embodiment. In addition, the same code | symbol is attached | subjected to the same component as the said 2nd Embodiment, and the detailed description is abbreviate | omitted.

  In the present embodiment, the adder 29 and the level detection means 35 function as an audio signal level detection means, and the comparison means 36 and the coefficient generation means 37 function as a low level component ratio variable means.

  The adder 29 adds the audio signal from the microphone 21 via the amplifier 23 and the audio signal from the microphone 22 via the amplifier 24. The level detection means 35 receives the output of the adder 29 and detects the total level of the added audio signal. The operation of the level detector 35 will be described in detail. The sum level of the audio signals from the microphones 21 and 22 is converted into an absolute value by the absolute value converting means 114 shown in FIG. input. The level detection means 115 detects the peak value or average value of the sum level of the audio signal converted into an absolute value.

  For example, in an analog circuit, the absolute value converting means 114 can be constituted by a rectifier circuit, and the level detecting means 115 can be constituted by a detection circuit such as a low-pass filter.

  The peak value or average value is compared with a reference level set from a terminal 38 by a comparison means (for example, a comparator) 36. For example, when the audio signal total level is higher than the reference level, the coefficient generated by the subsequent coefficient generation unit 37 is reduced, and conversely, when the audio signal total level is lower than the reference level, the coefficient generation unit 37. Increase the coefficient generated by.

  The coefficient generated by the coefficient generator 37 sets the gain in the amplifier 32. If the total level of the audio signal is large and white noise is not an issue, the coefficient is set to 0 so that noise reduction by addition averaging is not performed, so that the low-level audio component below the limiter level is monauralized. prevent. When the coefficient is set to 0, the adder 27 outputs the output signal (Rs + Rn) from the microphone 21 (through the amplifier 23) shown in the second embodiment as it is. Similarly, the adder 28 outputs the output signal (Ls + Ln) from the microphone 22 (via the amplifier 24) as it is.

  On the contrary, when the total sum level of the audio signal is small and white noise is relatively conspicuous, the coefficient is set to 0.5, and noise reduction is performed by averaging the low level components equal to or lower than the limiter level as in the second embodiment.

  The coefficient can be set in the range of 0 to 0.5. For example, when the coefficient is set to 0.3, the expression (1) in the second embodiment is the following expression (1a): The expression (2) becomes the following expression (2a).

Ra = (Rs + Rn) +0.3 (Ln-Rn) = Rs + 0.3Ln + 0.7Rn (1a)
La = (Ls + Ln) −0.3 (Ln−Rn) = Ls + 0.7Ln + 0.3Rn (2a)

  That is, the ratio of the low-level components Rn and Ln included in the respective Ra and La is changed as compared with the expressions (1) and (2). When the coefficient is 0.3, the effect of reducing the low level components including white noise is reduced as compared with the case where the coefficient is 0.5, that is, the addition averaging of the low level components Rn and Ln. When the coefficient is 0.5, that is, when the averaging of the low level components Rn and Ln is performed, the maximum noise reduction effect is obtained. When the coefficient is 0, noise reduction is not performed and the signal of each audio channel is output as it is. As the value approaches 0 to 0.5, the noise reduction effect is improved.

  In the above formulas (1a) and (2a), the low level components included in Ra and La are not included in the same manner as (0.3Ln + 0.7Rn) and (0.7Ln + 0.3Rn), and are therefore included in these. The low level component of the audio signal does not become monaural, and stereo processing can be performed at a later stage to obtain a stereo feeling for the low level audio component. This is true for coefficients in the range 0 <coefficient <0.5.

  Further, the comparing means 36 and the coefficient generating means 37 can also function as limiter level varying means, and the limiter in the low level component extracting means 31 is based on the coefficient generated by the coefficient generating means 37 as in the third embodiment. By changing the level, a balance adjustment may be made between the noise reduction level and the effect on the low-level sound component near the same level as the noise.

[Fifth Embodiment]
FIG. 5 shows a noise reduction circuit according to the fifth embodiment of the present invention. The noise reduction circuit according to the present embodiment has a configuration in which noise gate means 64, 65, and 66 are added to the noise reduction circuit according to the first embodiment. Therefore, the same components as those in the first embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  The noise gate means 64, 65, 66 are connected to the calculation means 61, 62, 63, respectively, and the respective audio signals whose low level components below a predetermined limiter level are reduced by the calculation means 61, 62, 63 are as follows. Are respectively input to the noise gate means 64, 65 and 66.

  Each noise gate means 64, 65, 66 outputs a low level component below a predetermined gate level with an output level smaller than the input level. This will be described with reference to FIG. In FIG. 14, the horizontal axis represents the level (input level) of the input signal to each noise gate means 64, 65, 66, and the vertical axis represents the level (output level) of the output signal from each noise gate means 64, 65, 66. Represents. The gate level is set to be equal to or lower than the level of the noise component {(n1 + n2 + n3) / 3} that is monauralized by the addition average in each arithmetic means 61, 62, 63, and an input level smaller than this gate level (see FIG. 14). The input level of the gate region shown) is not output or the output level is set lower than the input level. For a signal with an input level exceeding the gate level, the input is output as it is (assuming input: output = 1: 1).

  As described above, in the present embodiment, the white noise below the predetermined gate level included in the signal is further determined from the signal in which the white noise below the predetermined limiter level is reduced by the averaging process in each of the arithmetic means 61, 62, 63. Since the signal is output or the level is lowered, the noise reduction effect can be further enhanced.

[Sixth Embodiment]
FIG. 6 shows a noise reduction circuit according to a sixth embodiment of the present invention, which is a more specific example in the case of two microphones in the noise reduction circuit shown in the fifth embodiment.

  The noise reduction circuit according to the present embodiment is obtained by adding an adder 39, a low level component extraction unit 43, an amplifier 45, delay units 40 and 41, and adders 46 and 47 to the noise reduction circuit according to the second embodiment. It becomes the composition. The same components as those in the second embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  In the present embodiment, the adder 39, the low level component extraction means 43, the amplifier 45, and the adders 46 and 47 function as noise gate means.

  The output of the adder 27 is input to one + terminal of the adder 39 and the + terminal of the adder 46 via the delay unit 40. The output of the adder 28 is input to the other + terminal of the adder 39 and the + terminal of the adder 47 via the delay device 41. The output of the adder 39 is input to the low level component extraction means 43, where a low level component lower than a predetermined limiter level is extracted and input to the minus terminals of the adders 46 and 47 via the amplifier 45.

  Here, the output signal of the adder 27 is changed from the above equation (1) to Ra, the output signal of the adder 28 is changed from the above equation (2) to La, and the audio component included in the signal output from the microphone 21 is When Rs, the noise component is Rn, the audio component included in the signal output from the microphone 22 is Ls, the noise component is Ln, and the multiplication coefficient in the amplifier 45 is 0.5, the output Rb of the adder 46 is The output Lb of the adder 47 is expressed by the following equations (3) and (4), respectively.

Rb = Ra-0.5 (Ln + Rn) = Rs + 0.5 (Ln + Rn) -0.5 (Ln + Rn) = Rs (3)
Lb = La-0.5 (Ln + Rn) = Ls + 0.5 (Ln + Rn) -0.5 (Ln + Rn) = Ls (4)

  That is, the low level components Ln and Rn including white noise are canceled and only the audio components Rs and Ls are output.

[Seventh Embodiment]
FIG. 7 shows a noise reduction circuit according to the seventh embodiment of the present invention. The noise reduction circuit according to the present embodiment has a configuration in which the gate level variable means 74 is added to the combination of the noise reduction circuit according to the third embodiment and the noise reduction circuit according to the fifth embodiment. ing. Therefore, the same components as those in the third embodiment and the fifth embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  The gate level varying means 74 receives the coefficient generated by the coefficient generating means 73 according to the total level of the audio signals from the microphones 51, 52, 53 detected by the audio signal level detecting means 71, The gate level in each noise gate means 64, 65, 66 is set.

  Thereby, when the total level of the audio signal from each microphone is relatively large and the white noise is not concerned, the gate level for the low level component as described above in each noise gate means 64, 65, 66 is set to 0, or Avoid processing that reduces the level. As a result, it is possible to avoid removing and reducing the level of audio components other than white noise at a level lower than the gate level, and it is possible to reproduce small sounds below the gate level.

  Further, since the noise components included in the output signals from the respective arithmetic means 61, 62, 63 are the same as described above, for example, {(n1 + n2 + n3) / 3}, a common gate level can be set. It is not necessary to provide the gate level varying means 74 for each audio channel, and only one is required. Thereby, cost reduction, a circuit scale, and size reduction of an electronic device can be achieved.

[Eighth Embodiment]
FIG. 8 shows a noise reduction circuit according to the eighth embodiment of the present invention, which is a more specific example in the case of two microphones in the noise reduction circuit shown in the seventh embodiment.

  The noise reduction circuit according to the present embodiment is configured by combining the noise reduction circuit according to the fourth embodiment and the noise reduction circuit according to the sixth embodiment. Therefore, the same components as those in the fourth embodiment and the sixth embodiment are denoted by the same reference numerals, and detailed description thereof is omitted.

  That is, the limiter level in the low level component extraction means 31 and the gate level in the low level extraction means 43 are changed, or the multiplication coefficient in the amplifiers 32 and 45 is changed in accordance with the total level of the audio signals from the microphones 21 and 22. Thus, the balance between the noise reduction level and the effect on the low-level audio signal near the same level as the noise is adjusted.

  Note that the limiter level and the gate level respectively set by the low level component extraction unit 31 and the low level component extraction unit 43 are not limited to the same. For example, the adder 27 and the adder 28 perform level reduction by adding and averaging the low level components, and the audio signal below the level limiter is not removed. Therefore, the limiter level is made relatively large, and the multiplication coefficient in the amplifier 32 is Compared with the multiplication coefficient in the amplifier 45 (for example, 0.5), the adders 46 and 47 remove or reduce low level components below the gate level, so that the gate level is made smaller than the limiter level, If the multiplication coefficient in the amplifier 45 is smaller than the multiplication coefficient in the amplifier 32 (less than 0.5), the optimum noise reduction processing can be performed while suppressing the influence on the sound component having a level near the noise level.

[Ninth Embodiment]
FIG. 9 shows a noise reduction circuit according to the ninth embodiment of the present invention. The noise reduction circuit according to the present embodiment is the same as the noise reduction circuit according to the seventh embodiment described above, except that the microphone, the arithmetic means provided corresponding to the microphone, and the noise gate means are two, and further the audio signal level extraction. As a means, an automatic gain control circuit (AGC) 75 is used. In addition, the same code | symbol is attached | subjected to the same component as the said 7th Embodiment, and the detailed description is abbreviate | omitted.

  In this embodiment, the total level of the audio signals of the microphones 51 and 52 is obtained from the AGC 75. For example, in the case of a video camera, since the sound input to the microphone needs to be recorded with a predetermined dynamic range, the level of the audio signal from the microphone is generally controlled by AGC. In other words, when the audio signal is large, the level is lowered by AGC, and when the audio signal is small, the level is raised by AGC to make it easy for humans to hear. When the level gain in this AGC is small, the level is suppressed and almost cannot be heard, and conversely, when the level gain is large, it can be heard very conspicuously.

  Therefore, in this embodiment, when the level gain in the AGC 75 is small, the limiter level in the low level component extraction means 71 and the gate level in the noise gate means 64 and 65 are reduced to reduce the noise reduction effect, and conversely, the AGC 75 When the level gain at is large, the noise reduction effect can be optimized in accordance with the gain control operation in the AGC 75 by increasing the limiter level and the gate level to increase the noise reduction effect.

  As described above, in the present embodiment, it is not necessary to newly provide an audio signal level detection means, and the AGC 75 having the existing configuration is used. Therefore, an increase in cost is suppressed, and the circuit scale and the electronic device are not hindered. .

  The noise reduction circuit shown in each embodiment described above is applied to an electronic device including a microphone, for example, a video camera, a digital still camera, a tape, a disk as a recording medium, various recorders using a semiconductor memory, and the like. However, the present invention can also be applied to electronic devices having a plurality of channels of input / output, for example, surround multi-channel recording / playback devices.

  As mentioned above, although each embodiment of this invention was described, of course, this invention is not limited to these, A various deformation | transformation is possible based on the technical idea of this invention.

  In addition, each embodiment can be applied to the case where stereo sound field processing or multi-surround sound field processing is not performed in the subsequent stage. Moreover, the microphone may have directivity.

  The low level component extracted by the low level component extraction means may not include a component having a level equal to the set limiter level, and only a component smaller than the limiter level may be extracted. The same is true for the gate level.

  The coefficient generation in the coefficient generation means 37 in FIGS. 4 and 8 is not limited to whether the total level of the audio signal is larger or smaller than the reference level. For example, the inversion means 136 as shown in FIG. A coefficient may be directly generated from the output level of the level detecting means 35 by being interposed between the detecting means 35 and the coefficient generating means 37. That is, when the sum level of the audio signal is large, the coefficient generated by the coefficient generating unit 37 is reduced (so that the noise reduction effect is reduced), and conversely, the output of the level detecting unit 35, that is, the audio signal When the total level is small, the coefficient generated by the coefficient generation unit 37 may be increased (the noise reduction effect is increased).

  Noise to be reduced is not limited to white noise (thermal noise) of internal parts, but even if it is noise that is mixed from the outside, it may be a random waveform that has a low level compared to the audio signal and has no correlation between the audio channels. Can be reduced by the present invention.

1 is a block diagram showing a configuration of a noise reduction circuit according to a first embodiment of the present invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 4th Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 5th Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 6th Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 7th Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 8th Embodiment of this invention. It is a block diagram which shows the structure of the noise reduction circuit which concerns on the 9th Embodiment of this invention. It is a figure which shows one structural example of the low level component extraction means which concerns on this invention. It is a figure which shows one structural example of the audio | voice signal level detection means which concerns on this invention. It is a figure which shows the modification of the structure which produces | generates the coefficient according to an audio | voice signal level. It is a figure explaining the low level component extraction effect | action in a low level component extraction means. It is a figure explaining the noise gate effect | action in a noise gate means. It is a circuit diagram which shows an example of a stereo arithmetic processing circuit. It is a figure which shows a directivity pattern when there is no phase difference in the audio | voice signal of two microphones, and when there exists.

Explanation of symbols

  21, 22, 51 to 53, microphones, 25, 26, 40, 41, delay devices, 27, 28, 29, 30, 39, 46, 47, adders, 31, 71, low level component extraction means, 32, 45 ... Amplifier, 35, 70, 75 ... Audio signal level detection means, 61-63 ... Calculation means, 64-66 ... Noise gate means, 74 ... Gate level variable means.

Claims (12)

N (N is a positive number of 2 or more) audio channels;
Audio signal level detection means for detecting the total level of the audio signals of each of the N audio channels ;
Low level component extraction means for extracting low level components below a predetermined limiter level from the N audio signals ;
Limiter level changing means for changing the limiter level according to the total level detected by the audio signal level detecting means;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Computing means;
A noise reduction circuit comprising: N (N is a positive number of 2 or more) audio channels;
Audio signal level detection means for detecting the total level of the audio signals of each of the N audio channels ;
Low level component extraction means for extracting low level components below a predetermined limiter level from the N audio signals;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Computing means;
N included in the audio signal of the selected one audio channel to which the low level component extracted by the low level component extraction unit is added according to the total level detected by the audio signal level detection unit. Low level component ratio variable means for changing the ratio of the individual low level components;
A noise reduction circuit comprising: N (N is a positive number of 2 or more) audio channels;
Low level component extraction means for extracting low level components below a predetermined limiter level from the N audio signals;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Computing means;
Noise gate means for receiving an output signal of the computing means and outputting an output level lower than the input level for a low level component below a predetermined gate level ;
A noise reduction circuit comprising: Audio signal level detection means for detecting a total level of the N audio signals;
Gate level varying means for changing the gate level according to the total level detected by the audio signal level detecting means;
The noise reduction circuit according to claim 3 , further comprising: N (N is a positive number of 2 or more) audio channels;
Audio signal level detection means for detecting the total level of the audio signals of each of the N audio channels ;
Low level component extraction means for extracting low level components below a predetermined limiter level from the N audio signals ;
Limiter level changing means for changing the limiter level according to the total level detected by the audio signal level detecting means;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Computing means;
An electronic device comprising: N (N is a positive number of 2 or more) audio channels;
Audio signal level detection means for detecting the total level of the audio signals of each of the N audio channels ;
Low level component extraction means for extracting low level components below a predetermined limiter level from the N audio signals;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Computing means;
N included in the audio signal of the selected one audio channel, to which the low level component extracted by the low level component extraction unit is added according to the total level detected by the audio signal level detection unit. Low level component ratio variable means for changing the ratio of the individual low level components;
An electronic device comprising: N (N is a positive number of 2 or more) audio channels;
Low level component extraction means for extracting low level components below a predetermined limiter level from the N audio signals;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Computing means;
Noise gate means for receiving an output signal of the computing means and outputting an output level lower than the input level for a low level component below a predetermined gate level;
An electronic device comprising: Audio signal level detection means for detecting a total level of the N audio signals;
Gate level varying means for changing the gate level according to the total level detected by the audio signal level detecting means;
The electronic device according to claim 7 , further comprising: Detecting the sum level of each of the audio signals of N (N is a positive number of 2 or more) audio channels ;
Extracting low level components below a predetermined limiter level from the N audio signals ;
Changing the limiter level according to the total level detected by the audio signal level detecting means;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Steps,
A noise reduction method comprising : Detecting the sum level of each of the audio signals of N (N is a positive number of 2 or more) audio channels ;
Extracting low level components below a predetermined limiter level from the N audio signals;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Steps,
Changing a ratio of N low-level components included in the audio signal of the selected one audio channel, to which the extracted low-level components are added, according to the total level;
A noise reduction method comprising : Extracting low level components below a predetermined limiter level from the audio signals of N (N is a positive number of 2 or more) audio channels;
All the low level components extracted from the audio signals of N-1 audio channels other than one audio channel selected from the N audio channels are added to the audio signal of the selected one audio channel. Steps,
Receiving an input of the audio signal of the selected one audio channel to which the low level component is added, and outputting a low level component below a predetermined gate level with an output level smaller than the input level ; ,
A noise reduction method comprising : Detecting a total level of the N audio signals;
Changing the gate level according to the sum level;
The noise reduction method according to claim 11 , further comprising :

Images