JP2011171968A - Apparatus and method for noise elimination - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To generate a prediction signal on the basis of an accurate noise signal, and to properly negate noise from the outside. <P>SOLUTION: A noise elimination apparatus includes: a microphone 110 collecting a sound in a box body and generating a sound-collecting signal; a loudspeaker 132 for output of a regenerative signal; and a corrector 116 correcting the regenerative signal by the same transfer characteristics as those of a transmission line. The noise eliminator further includes: a regeneration eliminator 118 eliminating the corrected regenerative signal from the sound-collecting signal; a filter 124 functioning as a plurality of filters having the different number of taps eliminating a high-frequency component higher than or equal to a specified frequency from the sound-collecting signal eliminating the regenerative signal; and a prediction-signal generator 122 generating the prediction signal on the basis of the sound-collecting signal eliminating the high-frequency component. The noise eliminator further includes: a prediction-signal inverter 126 inverting the prediction signal; and an adder 128 adding the inverted prediction signal to a fundamental-sound signal input from the outside and generating the regenerative signal. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a noise removal apparatus and a noise removal method capable of preventing the quality of a desired reproduced sound from being deteriorated by external noise.

  Sound (reproduced sound) output from the headphones is transmitted to the ears of the person wearing the headphones through a transmission path such as space. Therefore, so-called noise other than the reproduced sound is also transmitted to the human ear through the transmission path, and the sound quality of the reproduced sound is impaired.

  Therefore, there is a headphone that installs a monitoring microphone inside or outside the headphones, adds the inverted signal of the microphone as a noise cancellation signal, adds it to the original sound signal, and outputs it from the speaker, thereby canceling noise from the outside. Are known. Such a noise cancellation signal can be generated using not only an analog circuit but also a digital circuit.

  However, when a noise cancellation signal is generated using a digital circuit, a phase difference due to delay associated with digital signal processing occurs between the actual noise and the calculated noise cancellation signal, and the noise cannot be sufficiently canceled out. . Therefore, a technique is disclosed in which a predetermined number of sampling data ahead is predicted from a predetermined number of sampling data from the past to the present of the noise signal, and the predicted signals are added in phase (for example, Patent Document 1). .

JP 2007-189530 A

  When a noise cancellation signal is generated using a digital circuit, the collected sound signal converted into digital contains a high frequency component that causes a prediction error, and therefore needs to be removed by a low-pass filter. As such a low-pass filter, an IIR (Infinite Impulse Response) filter or an FIR (Finite Impulse Response) filter is used.

  However, when an IIR filter is used as a low-pass filter, phase rotation occurs and the phase difference differs for each frequency. If the phase difference is different for each frequency, when trying to cancel a noise signal in a certain band, the noise component can be reduced by the anti-phase component, but in other bands, the phases overlap and the noise is reversed. The component will increase.

  Also, if a sufficient attenuation characteristic is obtained using the FIR filter, the tap length constituting the filter must be increased, and if the tap length is increased uniformly, a new sound pickup signal is obtained at the sampling point. The generated filter causes a delay and increases the predicted distance for compensating for the phase difference. As the prediction distance increases, the prediction error increases and the noise removal accuracy decreases.

  In view of such problems, the present invention has an object to provide a noise removal apparatus and a noise removal method capable of generating a prediction signal based on an accurate noise signal and appropriately canceling external noise. Yes.

  In order to solve the above-described problems, a noise removal apparatus according to the present invention includes a microphone that collects sound inside a housing and generates a sound collection signal, a speaker that outputs a reproduction signal, a transmission path for the reproduction signal. A correction unit that performs correction with a filter having a transfer characteristic substantially equal to the transfer characteristic, a reproduction removal unit that removes the corrected reproduction signal from the collected sound signal, and a predetermined frequency or more from the collected sound signal from which the reproduction signal has been removed A filter unit that functions as a plurality of filters with different number of taps, a prediction signal generation unit that generates a prediction signal based on the collected sound signal from which the high frequency component has been removed, and an inverted prediction signal And a prediction signal inverting unit for adding the inverted prediction signal to the original sound signal input from the outside to generate a reproduction signal.

  In order to solve the above-described problem, another noise removal apparatus of the present invention includes a microphone that collects sound outside a housing and generates a collected signal, and a high-frequency component that is equal to or higher than a predetermined frequency from the collected signal. A filter unit that functions as a plurality of filters having different numbers of taps to be removed, a prediction signal generation unit that generates a prediction signal based on a collected sound signal from which high-frequency components have been removed, and a prediction signal inversion unit that inverts the prediction signal And an adder that adds the inverted prediction signal to the original sound signal input from the outside to generate a reproduction signal, and a speaker that outputs the reproduction signal.

  The plurality of filters in the filter unit respectively generate a collected sound signal from which high frequency components at a plurality of past sample times are removed, and the older the sample time corresponding to the collected sound signal from which the high frequency components have been removed, You may refer to more sample values than the sample time.

  The filter unit may be configured by a filter without phase rotation.

  In order to solve the above-described problem, the noise removal method of the present invention collects sound inside a housing to generate a sound collection signal, removes a reproduction signal from the sound collection signal, and includes a plurality of taps having different numbers of taps. Functions as a filter, removes high frequency components above a specified frequency from the collected sound signal from which the playback signal has been removed, generates a prediction signal based on the collected sound signal from which the high frequency component has been removed, and inverts the prediction signal The reproduction signal is generated by adding the inverted prediction signal to the original sound signal input from the outside, and the reproduction signal is output.

  In order to solve the above-described problem, another noise removal method of the present invention collects sound outside the housing to generate a collected sound signal, functions as a plurality of filters having different tap numbers, and collects the collected sound signal. A high-frequency component above a predetermined frequency is removed from the sound signal, a prediction signal is generated based on the collected sound signal from which the high-frequency component has been removed, the prediction signal is inverted, and the prediction signal is inverted to the original sound signal input from the outside Is added to generate a reproduction signal and output the reproduction signal.

  The noise removal apparatus of the present invention can generate a prediction signal based on an accurate noise signal, and can appropriately cancel noise from the outside.

It is a functional block diagram showing an electrical configuration of the headphones according to the first embodiment. It is explanatory drawing for demonstrating the process of a prediction signal production | generation part. It is the functional block diagram which showed the electrical structure of the filter part. It is explanatory drawing for linking and explaining processing of each of a filter part and a prediction signal generation part. It is explanatory drawing for linking and explaining processing of each of a filter part and a prediction signal generation part. It is the flowchart explaining the flow of the whole process of the noise removal method using the headphones concerning 1st Embodiment. FIG. 6 is a functional block diagram showing an electrical configuration of headphones according to a second embodiment. It is the flowchart explaining the flow of the whole process of the noise removal method using the headphones concerning 2nd Embodiment.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The dimensions, materials, and other specific numerical values shown in the embodiment are merely examples for facilitating understanding of the invention, and do not limit the present invention unless otherwise specified. In the present specification and drawings, elements having substantially the same function and configuration are denoted by the same reference numerals, and redundant description is omitted, and elements not directly related to the present invention are not illustrated. To do.

  In the following embodiment, a headphone (noise canceling headphone) is cited as an example of a noise removal device. However, the present invention is not limited to such a case. For example, an earphone, a mobile phone, a PHS (Personal Handyphone System), a transceiver, etc. It is possible to use an electric device that allows a user to hear a sound (sound) by wearing or pushing the ear.

  Noise canceling headphones, for example, install a microphone outside (external) the headphones, invert the sound collected signal picked up by the microphone, and output as a noise canceling signal from a speaker provided inside the headphones in addition to the original sound signal . The noise cancellation sound output from the speaker cancels out (cancels) the noise from the outside, and the user can hear only the original sound from which the noise from the outside has been eliminated. In this embodiment, the original sound signal is an audio signal input from an external device such as a CD (Compact Disc) player, a DVD (Digital Versatile Disk) player, or a mobile phone.

  When such a noise cancellation signal is generated using a digital circuit, a phase difference due to a delay associated with digital signal processing occurs between the actual noise signal and the calculated noise cancellation signal. Cannot be countered.

  Therefore, a headphone using a digital circuit predicts a future sound signal from a sound signal picked up by a microphone and generates a prediction signal. By using such a prediction signal, the phase of the noise can be matched when the noise is canceled with the noise cancellation sound.

  However, when generating a prediction signal, if high frequency components that cause a prediction error are not sufficiently excluded from the original collected sound signal, noise cannot be canceled out and sound quality may be impaired. In the following embodiments, digital headphones that are excellent in high-frequency component removal performance and can generate a prediction signal based on an accurate noise signal will be described in detail. In the first embodiment, a feedback-type structure having a microphone (sound collecting unit) inside a headphone housing (housing), and in the second embodiment, a feedforward having a microphone outside the headphone housing. Take the structure of the system as an example.

(First embodiment: headphones 100)
FIG. 1 is a functional block diagram showing an electrical configuration of a headphone 100 according to the first embodiment. The headphone 100 includes a microphone 110, a microphone amplifier 112, an AD (Analog Digital) conversion unit 114, a correction unit 116, a reproduction removal unit 118, a buffer unit 120, a prediction signal generation unit 122, and a filter unit 124. The prediction signal inversion unit 126, the addition unit 128, the DA (Digital Analog) conversion unit 130, and the speaker 132 are included.

  The microphone 110 is a device that converts physical vibrations into an electrical signal, and the sound inside the casing, that is, noise generated from the noise source 150 around the headphone 100 is approximately 2 kHz or less that is not blocked by the casing. A noise collection signal is generated by collecting the noise represented by the frequency component and the reproduction sound emitted from the speaker 132. The microphone 110 applicable to the present embodiment is sufficient if it can convert vibration of an arbitrary transmission medium into a sound signal. For example, a capacitor microphone, a dynamic microphone, a ribbon microphone, a piezoelectric microphone, a carbon microphone, or the like can be used.

  The microphone amplifier 112 amplifies the amplitude of the collected sound signal converted by the microphone 110 so that the AD converter 114 can acquire the amplitude.

  The AD converter 114 removes a frequency component higher than the Nyquist frequency in order to suppress aliasing distortion from the collected sound signal amplified by the microphone amplifier 112 (low-pass filter), and then converts the analog signal into a digital signal at a predetermined sampling frequency. . It is sufficient that the AD conversion unit 114 can acquire a signal of approximately 2 kHz or less, which is a frequency band necessary for canceling (cancelling) noise from the outside, but the AD conversion unit 114 has a higher sampling frequency. By using a circuit capable of realizing the above, the configuration of the low-pass filter can be simplified and the delay can be suppressed. For example, when the sampling frequency is set to 96 kHz, the delay accompanying AD conversion processing, which greatly affects the predicted distance, can be shortened from 20 samples to about 5 samples.

  The correction unit 116 corrects the reproduction signal with a filter having a transmission characteristic substantially equal to the transmission characteristic of the transmission path in consideration of the electric circuit and transmission space from the DA conversion unit 130 to the AD conversion unit 114, and the reproduction removal unit 118 Then, the corrected reproduction signal is removed from the collected sound signal to obtain a noise signal. In the present embodiment, the reproduction signal is an audio signal that is a basis for generating a reproduction sound output from the speaker 132.

  As described above, the collected sound signal includes a reproduced signal in addition to the noise signal. In order to generate the noise cancel signal, the reproduced signal must be removed from the collected sound signal. The reproduced sound vibrates the headphone 100 itself or the air inside the headphone 100 through the speaker 132, so that when it reaches the position of the microphone 110, the housing of the headphone 100, the contour of the user's face, It is influenced by predetermined transfer characteristics based on the shape of the ear, the manner of wearing the headphones 100, and the like. Among these, at least the influence of the casing of the headphone 100 whose shape and properties are known can be measured, and its transfer characteristics can be calculated in advance by assuming an average facial contour and ear shape. . The correction unit 116 generates a reproduction signal substantially equal to the reproduction signal in the collected sound signal by correcting the reproduction signal through the filter having the transfer characteristic calculated in advance.

  The buffer unit 120 includes a RAM (Random Access Memory), an EEPROM (Electrically Erasable and Programmable Read Only Memory), a nonvolatile RAM, a flash memory, and the like, and holds a collected sound signal from which a reproduction signal is removed according to a sampling period. However, the sample value is shifted by a new sound pickup signal. Therefore, the buffer unit 120 can refer to past sample values of the collected sound signal.

  The prediction signal generation unit 122 generates a prediction signal that compensates for the delay in the system, based on the collected sound signal that is held in the buffer unit 120 and from which high-frequency components are removed through the filter unit 124 described later. The main causes of the delay in the system shown in FIG. 1 are the time spent for conversion by the AD conversion unit 114 and the DA conversion unit 130, the time spent for other digital signal processing, and the spatial delay from the speaker 132 to the microphone 110.

  The prediction signal generation method of the prediction signal generation unit 122 uses a method of matching the phase with a real-time noise signal using an adaptive filter using autocorrelation, and a future signal using noise signals from the past to the present. The method is roughly classified into a method of predicting a noise signal and adjusting the phase. Specifically, for example, a forward linear prediction method combining higher-order functions, a statistical method such as the Burg method, and an LMS (Least Mean Square) method for calculating a future signal based on the correlation of itself. The method can be used.

  Any of the above-described methods can be applied to the headphones 100 of the present embodiment, but here, a case where the forward linear prediction method is applied will be described as an example. The prediction signal generation unit 122 uses, for example, an FIR filter structure having a finite number of taps, and calculates a product sum of values derived based on one or a plurality of current and past sample values and a filter coefficient (prediction coefficient). Thus, the sound pickup signal at a future sample time is predicted, and a prediction signal is generated.

FIG. 2 is an explanatory diagram for describing processing of the prediction signal generation unit 122. Here, a case where fourth-order linear prediction is performed will be described as an example. As shown in FIG. 2, the prediction signal generator 122, if the sample time of the prediction signal is four samples a few minutes in the future from the sample point of the most recent collected sound signal, the sample value X ₀ of the latest collected sound signal The past three sample values X _{1 to} X ₃ are obtained _every four sample time points from the sample time point of the latest collected sound signal.

Then, the prediction signal generation unit 122 multiplies the acquired sample values X _{1 to} X ₃ by the corresponding prediction coefficients (A, B, C, and D), adds the products, and adds the products at the fourth sampling time point. A prediction signal Y obtained by predicting the prediction signal is generated.
Y = A × X ₀ + B × X ₁ + C × X ₂ + D × X ₃ (Formula 1)

  However, as described above, the correction unit 116 assumes an average facial contour and ear shape, so that the facial contour, the ear shape, the wearing mode of the headphones 100, and the like vary depending on the user. The transmission characteristics of the transmission path from the speaker 132 to the microphone 110 cannot be completely reproduced. For this reason, the collected sound signal output from the reproduction removing unit 118 is not completely removed, and a high frequency component remains in the reproduced signal. Hereinafter, the remaining high frequency component is referred to as a residual component.

  As in the present embodiment, in the feedback type headphone 100, the high frequency component of approximately 2 kHz or more of the external noise is largely blocked by the housing, so that the influence of the noise is suppressed during viewing, and the noise There is no need to cancel with a cancel signal. Therefore, since noise from the outside does not reach the microphone 110 provided in the housing, it can be considered that the noise signal is not included in the collected sound signal. For this reason, the residual component is the main factor in the high-frequency component of the collected sound signal. This residual component causes a prediction error and, if it is left as it is, performs a prediction process even for a frequency band that is originally unnecessary, resulting in an increase in processing load. Therefore, it is necessary to remove residual components using a low-pass filter.

  However, since the signal component in the vicinity of 2 kHz has a high sound pressure level in the reproduced signal, the attenuation characteristic of the low-pass filter must be steep in order to completely remove the residual component. An FIR filter is used as such a low-pass filter.

  If a steep attenuation characteristic is to be obtained with an FIR filter, the tap length constituting the filter must be increased. If the tap length is increased uniformly, the sampling time point is caused by a filter that generates a new collected sound signal. This causes a delay and increases the predicted distance for compensating for the phase difference. As the prediction distance increases, the prediction error increases and the noise removal accuracy decreases.

  Therefore, in the headphones 100 of the present embodiment, a reduction in noise removal accuracy is avoided by devising the configuration of the filter unit 124.

  FIG. 3 is a functional block diagram showing an electrical configuration of the filter unit 124. FIGS. 4 and 5 are explanatory diagrams for explaining the processing of the filter unit 124 and the prediction signal generation unit 122 in association with each other. is there.

  The filter unit 124 functions as a plurality of low-pass filters with different numbers of taps that further remove high frequency components of a predetermined frequency or higher from the collected sound signal from which the reproduction signal has been removed. As shown in FIG. 3, the filter unit 124 may actually include a plurality of low-pass filters (filters 124a to 124c), or a plurality of low-pass filters by sequentially processing one low-pass filter corresponding to the variable tap length. It may function as a filter.

As described above, the prediction signal generation unit 122 generates the prediction signal together with the latest sample value (sample value of the sound pickup signal) X ₀ and the four sample points from the sample point of the latest sample value in order to generate the prediction signal. In other words, the past three sample values X _{1 to} X ₃ are acquired and used. Here prepares past 3 sample values _X 1 to X ₃ 3 filter 124a~124c with different number of taps corresponding to from 3 filters 124a~124c instead of 3 sample values _X 1 to X ₃ The three filtered signals X ₁ ′ to X ₃ ′, which are the collected sound signals from which the high-frequency components have been removed, are input to the prediction signal generation unit 122.

Here, the filter signals X ₁ ′ to X ₃ ′ have different sample time points in the corresponding buffer unit 120, and the number of samples between the sample time point and the sample time point of the latest sample value is also different. Therefore, the filter 124a~124c respectively generate a filter signal _{X 1} 'to X _3', depending on the sample time corresponding to the filter signal _{X 1} 'to X _3', obtainable number of samples, in particular, the filter signal X _The number of samples is different between the sample time corresponding to ₁ ′ to X ₃ ′ and the sample time of the latest sample value.

For example, the filter 124 a will be described with reference to FIG. 4. The sample value at the sample time corresponding to the filter signal X ₁ ′ output from the filter 124 a has a time corresponding to the allowable delay amount W ₁ after being input to the buffer unit 120. It has elapsed, and four new sample values have been acquired after the sample time corresponding to the filter signal X ₁ ′. Filter 124a, the a new sample value four, the sample value one sample time corresponding to sample values X _1, in the sample value plurality (4 in the past, among the sample values of the range indicated by brackets 160 , for example, a predetermined number of sample values is continuous from the sample value of the sample point corresponding to the sample value X ₁₎ and using (in FIG. 4, using a sample value indicated by the braces 162), operation of the low pass filter To obtain a filter signal X ₁ ′ from which high frequency components have been removed.

In addition, the sample value at the sample time corresponding to the filter signal X ₂ ′ output from the filter 124 _b has elapsed by an allowable delay amount W ₂ after being input to the buffer unit 120, and the filter signal X ₁ ′ Eight sample values that are newer than the corresponding sample time have already been acquired. Similarly, the filter 124c, only the allowable delay amount W ₃ minutes after the acquired samples of the sample time corresponding to the filter signal X _{3 'time} has elapsed, a new sample value is 12 acquired. Therefore, the filters 124b and 124c can perform calculations as a low-pass filter using a larger number of sample values because there are more new sample values. By increasing the tap length in this way, the frequency cutoff performance is improved, and a more reliable signal can be obtained. Here, the filter 124a~124c has a function of reducing the residual high-frequency components of the respective sample points corresponding to the filtered signal X _{1 '~X} 3', the number of samples to the latest signals available from the temporal course Is different.

  As described above, the plurality of low-pass filters (filters 124a to 124c) in the filter unit 124 are older than the sample time point as the sample time point corresponding to the filter-processed signal from which high frequency components are removed is older. Can refer to many new sample values and take many taps.

In order to generate the filter signals X ₁ ′ to X ₃ ′, the above-described filters 124 a to 124 c require at least the sample value at the oldest sample time corresponding to the filter signal X ₃ ′. Therefore, the buffer unit 120 has to prepare buffers for the number of samples corresponding to a predetermined tap length when obtaining the filter signal X ₃ ′. The filter unit 124 outputs filter signals X ₁ ′ to X ₃ ′ by effectively using taps between sample values at the sample time corresponding to the filter signal X ₃ ′ from the latest sample values. That is, the older the sample time corresponding to the filter signal to be generated, the greater the number of taps of sample values to be acquired at the new sample time. With this configuration, it is possible to improve the accuracy of high frequency component removal by effectively using past samples provided in the buffer unit 120.

  Further, as shown in FIG. 5, the filters 124 a to 124 c each have a tap number for generating a new sample value from the sample time corresponding to the filter signal to be generated and an old sample around the sample time corresponding to the filter signal to be generated. The number of taps for generating values may be equal, and the filter coefficients in each of the filters 124a to 124c may be arranged to be even-symmetric between the new sample point and the old sample point.

For example, in FIG. 5, the filter 124a, the new sample value four and sample values four and the same number of past sample value sample value of the sample value of the sample point one total 9 corresponding to X ₁ (in FIG. 5, Is used as a low pass filter to derive a filter signal X ₁ ′ from which high frequency components have been removed. Similarly, each of the filters 124b and 124c uses 17 samples (indicated by curly brackets 164b in FIG. 5) and 25 sample values (indicated by curly brackets 164c in FIG. 5) as low-pass filters. Perform the operation.

  When the number of samples referred to centering on the sample time point corresponding to the filter signal generated in this way is made equal to old and new, the filters 124a to 124c are filters having linear phase characteristics with the same delay for each frequency band. Become. When the delay for each frequency band is equal, delay compensation by the prediction signal is facilitated.

  In the filter unit 124 of the headphone 100 of the present embodiment, a plurality of filters 124 a to 124 having different tap numbers, each adjusted to an appropriate number of taps according to a plurality of sample points corresponding to the filter signal used by the prediction signal generation unit 122. Since the high frequency component can be removed by 124c without increasing the predicted distance, the high frequency component can be accurately removed. Therefore, the prediction signal generation unit 122 can generate a prediction signal based on an accurate noise signal, and the addition unit 128 can appropriately cancel noise from the outside. In addition, since the frequency band of the prediction signal can be narrowed, the prediction signal generation unit 122 can reduce the calculation load and improve the accuracy of the prediction signal.

  Further, the filter unit 124 of the present embodiment is configured by a filter without phase rotation. Thus, the prediction signal generation unit 122 does not need to compensate for phase rotation, and can improve the prediction accuracy of the prediction signal.

At this time, the filter unit 124 may be used sample value X ₀ of the latest collected sound signal may be not to use dare. When the latest collected sound signal is used, it is necessary to wait until the latest collected sound signal is sampled, but the accuracy of the high frequency component removal processing is improved. If the filter is configured only with sample values at other sample times without using the latest collected sound signal, the accuracy is inferior, but the prediction signal can be obtained quickly. Which configuration is adopted can be appropriately selected according to circuit characteristics.

  In the present embodiment, the sample value of the latest collected sound signal is output directly to the prediction signal generation unit 122 without passing through the filter unit 124. However, the present invention is not limited to this case. The high frequency component can also be removed from the collected sound signal.

  In addition, when the LMS method is used instead of the forward linear prediction method for generating the prediction signal, the prediction signal is configured by L-1 filters obtained by removing the tap of the latest sample value from L which is the number of LMS taps. In order to reduce the number of low-pass filters, a common low-pass filter may be used. Even when the LMS method is used, a corrected noise signal is acquired at once by parallel processing using a plurality of filters 124a to 124c, or sequentially processed using one filter with a variable number of taps. May be.

  The prediction signal inversion unit 126 inverts the prediction signal to generate a noise cancellation signal, and the addition unit 128 adds the inverted prediction signal (noise cancellation signal) to the original sound signal input from the outside to generate a reproduction signal. Generate. DA converter 130 converts the reproduction signal from digital to analog, and speaker 132 outputs the converted reproduction signal as reproduction sound.

  Further, in the present embodiment, the case where the original sound signal is digital has been described as an example. However, when the original sound signal is analog, the adder as the adder 128 is arranged at the subsequent stage of the DA converter 130 and is converted to DA. The reproduction signal may be generated by adding the noise cancellation signal converted to analog by the unit 130 and the original sound signal.

  As described above, the prediction signal generation unit 122 according to the present embodiment can generate a prediction signal based on an accurate noise signal, and the addition unit 128 can appropriately cancel noise from the outside. It is possible to enjoy music and the like without being disturbed by noise from the outside even in the jet sound, the traveling sound in the train, and the air conditioning sound in the room.

(Noise removal method)
Next, a noise removal method using the headphones 100 described above will be described using a flowchart.

  FIG. 6 is a flowchart for explaining the overall processing flow of the noise removal method using the headphones 100 according to the first embodiment.

  When the sampling period arrives (YES in S200), the microphone 110 collects the sound inside the housing and generates a sound collection signal (S202). The AD conversion unit 114 removes a frequency component equal to or higher than the Nyquist frequency in order to suppress aliasing distortion from the collected sound signal, and then converts the collected sound signal from analog to digital at a predetermined sampling frequency (S204).

  Then, the correction unit 116 corrects the reproduction signal generated inside the headphones 100 with a filter having a transmission characteristic substantially equal to the transmission characteristic of the transmission path from the DA conversion unit 130 to the AD conversion unit 114 (S206). The removal unit 118 removes the corrected reproduction signal from the collected sound signal to obtain a noise signal (S208).

  The filter unit 124 functions as a plurality of filters 124a to 124c having different numbers of taps, removes high frequency components of a predetermined frequency or higher from the collected sound signal from which the reproduction signal has been removed (S210), and the predicted signal generation unit 122 Based on a noise signal, which is a collected sound signal from which high-frequency components have been removed through the filter unit 124, a delay in the system (a delay in the AD conversion unit 114 and the DA conversion unit 130 and a delay in digital signal processing), and a space A prediction signal that compensates for the delay is generated (S212).

  The prediction signal inversion unit 126 inverts the prediction signal to generate a noise cancellation signal (S214), and the addition unit 128 adds the noise cancellation signal to the original sound signal input from the outside to generate a reproduction signal (S216). ). Then, the DA converter 130 converts the reproduction signal from digital to analog (S218), and the speaker 132 outputs the reproduction signal as reproduction sound (S220).

  Such processing is repeatedly executed, for example, every predetermined sampling period while the headphones 100 are activated.

  As described above, according to the noise removal method of the present embodiment, a predicted signal can be generated based on an accurate noise signal, and external noise can be appropriately canceled out.

(Second embodiment: headphones 300)
In the first embodiment, the feedback type headphone 100 that collects sound inside the casing has been described. Subsequently, in the second embodiment, a feedforward type headphone 300 that collects sound outside the housing and performs noise removal processing will be described.

  FIG. 7 is a functional block diagram showing an electrical configuration of the headphones 300 according to the second embodiment. The headphone 300 includes a microphone 310, a microphone amplifier 112, an AD conversion unit 114, a correction unit 316, a buffer unit 120, a prediction signal generation unit 122, a filter unit 124, a prediction signal inversion unit 126, and an addition unit. 128, a DA converter 130, and a speaker 132. Note that components that are substantially the same as those in the first embodiment described above are denoted by the same reference numerals and description thereof is omitted. Here, the microphone 310 and the correction unit 316 having different configurations will be mainly described.

  The microphone 310 collects sound outside the housing and generates a sound collection signal. Unlike the microphone 110 in the first embodiment, the microphone 310 directly picks up noise outside the housing, and thus the influence of the reproduction signal is extremely small and may be ignored. Therefore, unlike the headphone 100, the headphones 300 do not need to include the correction unit 116 that corrects the reproduction signal and the reproduction removal unit 118 that removes the corrected reproduction signal.

  However, the noise that actually reaches the user's ear has a signal component of approximately 2 kHz or more removed due to the sound insulation effect of the housing, compared to the sound outside the housing that the microphone 310 collects, and is further 2 kHz or less. The component is also attenuated. Therefore, it is necessary to perform a correction corresponding to the transfer characteristic until the noise signal directly picked up reaches the ear. Therefore, the headphones 300 are newly provided with a correction unit 316.

  The correction unit 316 is a filter having a transmission characteristic substantially equal to the transmission characteristic of the transmission path through the housing from the microphone 310 to the position of the user's ear, and corrects the collected sound signal collected by the microphone 310.

  As described above, a signal component of approximately 2 kHz or more is removed from the transmission path through the casing. However, when the correction unit 316 is caused to function as a filter having desired characteristics, as described above, phase rotation may occur depending on the configuration of the filter (IIR), or the prediction signal generation unit 122 generates a prediction with a delay. The predicted distance of the signal increases unnecessarily (FIR). Therefore, the correction unit 316 does not completely reproduce the transfer characteristic of the casing that removes signal components of approximately 2 kHz or higher, and the filter unit 124 removes high frequency components of approximately 2 kHz or higher.

  The filter unit 124 of the headphone 300 according to the present embodiment also has a plurality of filters with different tap numbers that are matched to a plurality of sample points corresponding to the filter signal used by the prediction signal generation unit 122, as in the first embodiment. High frequency components are removed at 124a to 124c. For this reason, it is possible to appropriately cancel external noise.

(Noise removal method)
Next, a noise removal method using the above-described headphones 300 will be described using a flowchart. FIG. 8 is a flowchart for explaining the overall processing flow of the noise removal method using the headphones 300 according to the second embodiment.

  When the sampling period arrives (YES in S200), the microphone 110 collects a sound outside the housing and generates a sound collection signal (S402). The AD conversion unit 114 removes a frequency component equal to or higher than the Nyquist frequency in order to suppress aliasing distortion from the collected sound signal, and then converts the collected sound signal from analog to digital at a predetermined sampling frequency (S204).

  Then, the correction unit 316 corrects the collected sound signal generated by the microphone with a filter having a transfer characteristic substantially equal to the transfer characteristic of the transmission path of the housing (S406), and the filter unit 124 includes a plurality of filters having different tap numbers. Functions as 124a to 124c, and removes a high frequency component of a predetermined frequency (approximately 2 kHz) or more from the collected sound signal from which the reproduction signal has been removed to generate a filter signal (S210).

  Hereinafter, the processing from the prediction signal generation step (S212) to the output step (S220) is substantially the same as the processing described in FIG.

  As described above, according to the noise removal method of the present embodiment, the plurality of filters 124a to 124c having different tap numbers in accordance with a plurality of sampling points of the digitized sound pickup signal can be sound-insulated by the housing. The high frequency component which did not exist is removed from the collected sound signal. Therefore, there is an effect of improving the prediction performance, and it is possible to appropriately cancel noise from the outside.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to this embodiment. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

  For example, each component from the filter unit 124 to the prediction signal inversion unit 126 constituting the headphones 100 and 300 described above may be hardware such as a semiconductor chip having the function, or the function is programmed. Or software. In addition, a configuration in which a digital sound pickup signal is directly acquired from the microphone 110 instead of the AD conversion unit 114 may be employed, or a digital noise cancellation signal or a reproduction signal may be directly input to the speaker 132 instead of the DA conversion unit 130. An output configuration can also be adopted.

  Note that each step in the noise removal method of the present specification does not necessarily have to be processed in time series in the order described in the flowchart, and may include processing in parallel or by a subroutine.

  INDUSTRIAL APPLICABILITY The present invention can be used in a noise removal device and a noise removal method that can prevent the sound quality of a desired reproduced sound from being impaired by external noise.

100, 300 ... headphones (noise removal device)
110, 310 ... Microphones 116, 316 ... Correction unit 118 ... Reproduction removal unit 122 ... Prediction signal generation unit 124 ... Filter unit 126 ... Prediction signal inversion unit 128 ... Addition unit 132 ... Speaker

Claims (6)

A microphone that picks up the sound inside the housing and generates a sound pickup signal;
A speaker for outputting a reproduction signal;
A correction unit that corrects the reproduction signal with a filter having a transmission characteristic substantially equal to the transmission characteristic of the transmission path;
A reproduction removing unit that removes the corrected reproduction signal from the collected sound signal;
A filter unit that functions as a plurality of filters with different numbers of taps for removing high frequency components of a predetermined frequency or higher from the collected sound signal from which the reproduction signal has been removed;
A prediction signal generation unit that generates a prediction signal based on the collected sound signal from which the high-frequency component has been removed;
A prediction signal inversion unit for inverting the prediction signal;
An adder for adding the inverted prediction signal to the original sound signal input from the outside to generate the reproduction signal;
A noise removal apparatus comprising: A microphone that picks up the sound outside the housing and generates a sound pickup signal;
A filter unit that removes high frequency components of a predetermined frequency or higher from the collected sound signal and functions as a plurality of filters having different tap numbers;
A prediction signal generation unit that generates a prediction signal based on the collected sound signal from which the high-frequency component has been removed;
A prediction signal inversion unit for inverting the prediction signal;
An addition unit that adds the inverted prediction signal to the original sound signal input from the outside to generate a reproduction signal;
A speaker for outputting the reproduction signal;
A noise removal apparatus comprising:   The plurality of filters in the filter unit respectively generate a sound pickup signal from which the high frequency component has been removed at a plurality of past sample points, and the sample time point corresponding to the sound pickup signal from which the high frequency component has been removed is old The noise removal apparatus according to claim 1 or 2, wherein a larger number of sample values than the sample time point are referred to.   The noise removing apparatus according to any one of claims 1 to 3, wherein the filter unit includes a filter having no phase rotation. Collects the sound inside the housing to generate a sound collection signal,
Removing the reproduction signal from the collected sound signal;
Functions as a plurality of filters with different tap numbers, removes high frequency components of a predetermined frequency or more from the collected sound signal from which the reproduction signal has been removed,
Generating a prediction signal based on the collected sound signal from which the high-frequency component has been removed;
Inverting the prediction signal;
Add the inverted prediction signal to the original sound signal input from the outside to generate the reproduction signal,
A method for removing noise, comprising outputting the reproduction signal. Pick up the sound outside the housing to generate a sound pickup signal,
Functions as a plurality of filters with different number of taps, removes high frequency components above a predetermined frequency from the collected sound signal,
Generating a prediction signal based on the collected sound signal from which the high-frequency component has been removed;
Inverting the prediction signal;
Add the inverted prediction signal to the original sound signal input from the outside to generate a playback signal,
A method for removing noise, comprising outputting the reproduction signal.

Images