KR101789781B1 - Apparatus and method for attenuating noise at sound signal inputted from low impedance single microphone - Google Patents

Abstract

An apparatus for attenuating noise from a voice signal by a low impedance microphone according to the present invention includes a microphone for generating a digital converted signal Xvar (n) by digitally converting a sound pressure signal received by one microphone, (N)) and the period inverse signal (X'ns (n)) by using the least square mean (Least (N)), which is a difference signal between the output signal and the periodic mantel (Xns (n)), and generates an error signal Xerr (n) If the magnitude is greater than the desired magnitude, the output signal is modified and then combined with the periodic mantel (Xns (n)) and if the generated error signal Xerr (n) is attenuated to below the desired degree And outputs the output signal as the periodic noise canceling signal (Xnps (n)) in which the periodic noise is attenuated Including the ability to.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to an apparatus and a method for attenuating noise from a voice signal by a low-impedance single microphone,

The present invention relates to an apparatus for attenuating noise from a voice signal by a low impedance single microphone, and more particularly to an apparatus for attenuating an increased noise portion by amplifying a voice signal inputted through a low impedance microphone with a very high amplification degree, And a method of processing the attenuation.

In general, various methods for attenuating the noise from the surroundings except for the voice are amplified by amplifying the voice of the human being from the various sounds introduced into the microphone to a greater extent.

Japanese Patent Application No. 10-1395329 (entitled "Method for Removing Noise Using Two Microphones and Mobile Communication Terminal)" (hereinafter referred to as "prior art"), a method of using two or more microphones to attenuate electrical noises The method comprising: Particularly, in a mobile communication terminal equipped with two microphones, a voice including a user's voice and surrounding noises is received and a noise is separated and removed. In accordance with the signal strength of the signals from the microphones, One of the microphones is set as the main microphone according to the position of the microphone, and the signals from the microphones are compared with each other to discriminate the ambient noise.

However, this method has a disadvantage in that noise analysis is performed depending on the position where the microphone is disposed, because noise is analyzed using the difference in sound absorption by the microphones. One microphone can not be implemented, and two or more microphones . In addition, when noise above a certain level is simultaneously introduced into each microphone, noise may not be analyzed as noise. In addition, as the number of required microphones increases, the desired voice signal may be distorted, and additional noise may occur.

Patent Document 1: Registration No. 10-1395329

The present invention is to improve a method of attenuating noise from a voice that is received by using two or more microphones. The present invention can efficiently attenuate ambient noise and electrical noise from a voice signal while using only one low- Device.

According to an aspect of the present invention, there is provided an apparatus for attenuating noise from a voice signal by a low impedance microphone, the apparatus comprising: a variable impedance amplifier for amplifying a volume of a sound pressure signal (Xmic (t)) generated by a low impedance microphone; And generates spectral information Xspec (n) by analyzing a frequency component from the generated digital signal to generate spectral information Xspec (n) (Nv) to perform a frequency analysis for analyzing the level and instantaneous loudness change between the frequency bands from the frequency transformed signal Xvar (n) (Nth), the range of the noise region is selected from the threshold processing signal Xthl (n), and the level of the selected noise region is adjusted to adjust the noise level Generates a period noise reduction signal Xnps (n) by removing the periodic noise from the noise level adjustment signal Xnad (n), generates a period noise reduction signal Xnps (n) And generates a level displacement adaptive signal Xnvc (n) by controlling the level of the periodic noise signal Xnad (n) and the period of time during which the threshold processing signal Xthl (n) and the noise level adjustment signal Xnad The threshold processing signal Xthl (n) and the noise level adjusting signal Xnad (n) are converted into the time matching signal Xtm (n) in order to compensate for the difference in time at which the attenuation signal Xnps And outputs the processed speech signal Xnc (n) by selectively mixing the periodic noise reduction signal Xnps (n) with the level displacement adaptive signal Xnvc (n) and the time matching signal Xtm (n) ).

In particular, an apparatus for attenuating noise from a voice signal by a low-impedance microphone according to an embodiment of the present invention may include a microphone for generating a digital converted signal Xvar (n) by digitally converting the sound pressure signal generated by the microphone A variable amplification and signal conversion unit; (Nn) and the period inverse signal X'ns (n) by using the digital converted signal Xvar (n), and outputs the period inverse signal X'ns (n) Generates an error signal Xerr (n), which is a difference signal between the output signal and the periodic mantel (Xns (n)), and outputs the error signal Xerr If the magnitude of the error signal Xerr (n) is greater than a desired degree, the output signal is adjusted and then combined with the periodic mantel Xns (n), and the generated error signal Xerr (n) And outputs the final output signal as the periodic noise attenuated periodic noise reduction signal (Xnps (n)) if the periodic noise is attenuated to a predetermined level or less. A noise displacement controller for controlling a level of the period noise reduction signal Xnps (n) to generate a level displacement adaptive signal Xnvc (n) adapted to an arbitrary level displacement curve; And outputs the resultant speech signal Xnc (n) in which the periodic noise is attenuated by giving an arbitrary weight to the periodic noise reduction signal Xnps (n) and the level displacement adaptive signal Xnvc (n) A noise adjusting / synthesizing unit for outputting the noise through the speaker; And a system operation unit for controlling operations of the variable amplification and signal conversion unit, the periodic noise processing unit, the noise displacement adjustment unit, and the noise adjustment combination unit.

The noise attenuator further includes a tone color and threshold level processing unit 130 for receiving the digital converted signal Xvar (n) to control the tone color and to generate the threshold processed signal Xthl (n) by limiting the threshold level; And a noise region selection unit 140 for selecting a noise region in the critical processing signal and adjusting a level of the selected noise region to generate a noise level adjustment signal Xnad (n).

The time delays? T1 and? T2 of the threshold processing signal Xthl (n) and the noise level adjusting signal Xnad (n) with respect to the period noise reduction signal Xnps (n) And a signal processing matching unit 170 for generating a matching signal Xtm (n), wherein the noise adjusting and synthesizing unit further receives the time matching signal Xtm (n) And outputs the resultant speech signal Xnc (n) in combination with the periodic noise reduction signal Xnps (n) and the level displacement adaptive signal Xnvc (n).

The periodic noise processing unit includes a noise signal sum process block 151 and a noise signal sum process block 151 for processing the input signal by the first transfer function H (n) to generate the period mantle Xns (n) A period and a noise signal sum reverse processing block 152 for processing the input signal by a second transfer function H '(n) to generate the period inverse signal X'ns (n) An adaptive Fir Filter LMS term block 153 for applying the Fir Fiter to the inverse signal X'ns (n) and performing the least square mean (LMS) processing on the signal generated by applying the Fir Filter to generate the output signal .

Further comprising a signal analysis unit (120) for analyzing a frequency component of the digital converted signal (Xvar (n)) to generate spectral information (Xspec (n)), wherein the system operation unit (190) And can control the operation of each part based on the information Xspec (n).

According to the above-described configuration, the present invention can efficiently attenuate ambient noise and electrical noise while using only one low-impedance microphone.

In particular, by using a low-impedance microphone, incoming ambient noise can be primarily controlled.

Even if a voice signal generated by a microphone with a low impedance is amplified with a high degree of amplification, noise can be effectively removed, so that only a desired voice signal can be clearly heard.

In addition, by performing the tone color control and the preprocessing of an unnecessary sound source on the voice signal by the variable amplified signal conversion unit and the tone color and threshold level processing unit, the tone of the listened voice can be realized with the optimum sound quality.

In addition, since the signal processing matching unit performs the time compensation processing on the signal delay occurring during the speech processing, it is possible to improve the image separation phenomenon between the direct listening speech and the processed speech output. Thereby, the listener can listen to the voice with high clarity.

In addition, since only one microphone is required, there is a high degree of freedom in designing a microphone in a headset or earset.

In particular, since the present invention provides an attenuation function only for periodic noise, it does not cause deterioration of sound quality for non-periodic signals such as voice signals.

FIG. 1 is a block diagram schematically illustrating a configuration of a device for attenuating noise from a voice signal by a low-impedance microphone according to an exemplary embodiment of the present invention. Referring to FIG.
FIG. 2 is a diagram for explaining the variable amplification and signal conversion unit of the noise attenuator shown in FIG. 1 in more detail.
3 is a diagram for explaining the signal analyzer in the noise attenuator in more detail.
FIG. 4 is a diagram for explaining the tone color and threshold level processing unit of the noise attenuator in more detail.
5 is a diagram for explaining the noise region selection unit of the noise attenuator in more detail.
6 is a diagram for explaining the periodic noise processing unit of the noise attenuator in more detail.
7 is a diagram for explaining the adaptive FIR filter LMS stage constituting the periodic noise processing unit shown in FIG. 6 in more detail.
8 is a diagram showing a specific configuration of the noise displacement adjusting unit in the noise attenuator.
9 is a diagram showing the relationship between the input value and the output value in the structure of the noise displacement gradient control stage and the noise displacement gradient control stage of the noise displacement adjustment unit.
10 is a diagram for explaining the signal processing matching unit of the noise attenuator in more detail.
11 is a diagram for explaining the noise control combining unit of the noise attenuator in more detail.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, preferred embodiments of an apparatus and method for attenuating noise from a voice signal by a low-impedance microphone according to the present invention will be described with reference to the accompanying drawings. For the sake of reference, terms referring to the respective elements of the present invention have been exemplarily named in consideration of their functions, and therefore, the terms of the present invention should not be predicted and limited in terms of the term itself.

A system capable of recognizing a periodic signal as noise from two or more voice signals generated by two or more microphones disposed around the listener and attenuating the recognized periodic signal to clearly listen only to a desired voice signal is known .

In the case of using two or more microphones, at least one microphone is arranged so that the voice desired by the listener is mainly introduced, and the other microphones are arranged such that ambient noise to be filtered by the noise is mainly introduced, And removes noise by removing or attenuating a signal including ambient noise from a signal containing the desired voice.

However, if the intensity of the noise from the outside is swallowed, the desired voice signal can be damaged to remove the noise, so that noise cancellation is limited.

Further, when two or more microphones are used, the sensitivity characteristics of each microphone itself must be adjusted, and the arrangement of the microphones also needs to be adjusted. There is a need for a process of detecting a periodic pattern from signals generated by receiving and determining noise, and generating an attenuation pattern for attenuating noise.

In addition, each of two or more microphones may generate a time difference when a signal is generated by itself. If signals having time differences are directly combined with each other, signal distortion may occur, and even another noise .

On the other hand, in the noise attenuator of the present invention, a low impedance microphone is used, so that a voice signal is generated by the sound pressure of voice. Therefore, ambient noise with low sound pressure can be limited in the inflow.

And since only one microphone is used, it is not necessary to consider the time difference from the signal by another microphone. That is, it is possible to eliminate the necessity of adjusting the characteristics of each microphone, which may occur when a plurality of microphones are used, and it is possible that the time difference of the signals depending on the placement positions of the microphones may distort the voice signal or generate additional noise The problem does not occur fundamentally.

The apparatus for attenuating noise from a voice signal input by a single low impedance microphone according to the present invention can be applied to a headset, an earphone, a telephone, a radio, and the like for receiving voice of a person in a place where a lot of ambient noise is generated.

First, with reference to the block diagram of FIG. 1, a schematic description of a configuration of a device for attenuating noise from a voice signal by a low-impedance microphone according to an embodiment of the present invention (referred to as a noise attenuator in this specification) do.

Referring to FIG. 1, the noise attenuator includes a low-impedance single microphone 90 for receiving a voice and generating a sound pressure signal Xmic (t), a matching signal Xmatch () for matching a sound pressure signal to an impedance matching system a variable amplification and signal conversion unit 110 for variably amplifying a matching signal and converting the amplified signal into a digital conversion signal Xvar (n); A tone analyzer 120 for generating spectral information Xspec (n) based on the tone signal Xspec (n), a tone color and threshold level processor (tone generator) 120 for controlling the tone color of the digital signal and limiting the threshold level to generate the threshold processing signal Xthl A noise region selection unit 140 for selecting a noise region in the critical processing signal and generating a noise level adjustment signal Xnad (n) by adjusting the level of the selected noise region; Selective attenuation of noise to reduce periodic noise A noise suppression unit 150 for generating a level displacement adaptive signal Xnvc (n) by controlling an adaptation level range according to noise displacement in the periodic noise attenuation signal, A signal processing matching unit 170 for converting a threshold processing signal and a noise level adjusting signal into a time matching signal Xtm (n) in order to compensate a time delay for the periodic noise reduction signal, The adaptive environment (noisy or low ambient environment, or music source (the music source may be non-periodic but may have some periodicity and can be treated as noise) by adjusting the gain of each of them and selectively mixing them (N)) corresponding to various conditions by appropriately compensating the tones according to the input conditions (for example, the environment receiving the input voice signal Xnc (n)) and outputting the resultant voice signals Xnc action But the control is achieved by operating the system in particular comprises a section 190 for controlling the signal processing operations of the parts based on the spectrum information. The apparatus may further include an amplifier 210 for amplifying the resultant speech signal Xnc (n) to provide the resulting speech signal to a speaker for audibly outputting a speech signal to the listener as a result of the noise being attenuated.

The noise attenuator disclosed in the present invention may include only one microphone. Alternatively, only one microphone signal input means may be provided. In particular, the microphone of the present invention preferably has a low impedance of 50 Ω or less, preferably 150 Ω or less.

Generally, a microphone can distinguish between a condenser microphone and a dynamic micro. The condenser microphone has an impedance of 150? To several k ?, and the dynamic microphone has an impedance of 5? To 600?. The low impedance in the present invention means 150 Ω or less.

The microphone having such a low impedance operates in such a manner as to sense the sound pressure by the sound to generate the analog sound pressure signal Xmic (t). The sound pressure signal includes a non-periodic voice signal and a periodic noise signal.

On the other hand, such a microphone detects sound pressure and is not sensitive to ambient sound. Therefore, there is an effect that the ambient noise can be suppressed by the microphone itself. These microphones may be used in devices that speak close to the mouth of a speaker such as a headset, an earset, a telephone handset, a radio, and the like.

However, since the analog sound pressure signals generated by these low impedance microphones are low in sensitivity, they must be amplified with high amplification. However, in this amplification process, weakly inputted ambient noise may be amplified together, or the electrical noise caused by the signal amplifying device itself may be amplified together, and the desired voice may be buried in noise. The noise attenuator according to the present invention can effectively attenuate ambient noise and electrical noise even in the amplifying operation.

The impedance matching unit 100 performs signal matching with the following sections including the low impedance microphone and the variable amplification signal conversion unit 110. The signal matching unit 100 may convert the sound pressure signal Xmic (t) into a matching signal Xmatch (t) using a matching transformer (not shown).

The variable amplification and signal conversion unit 110 variably amplifies the low volume of the matching signal X (match (t)) transmitted by the impedance matching unit by an arbitrary degree of amplification, digitally converts the variable amplified signal, (Xvar (n)). That is, since the present invention uses a sound pressure signal by a microphone of low impedance, the sound volume is extremely low. Therefore, the volume of the sound pressure signal should be amplified with a sufficiently high amplification degree.

On the other hand, by amplifying the sound pressure signal received from the microphone in an analog signal according to the mutual ground separation method through impedance matching, the ground induced noise between the low impedance input signal and the amplified signal can be reduced. Also, by performing the volume amplification before performing the noise elimination processing on the signal, it is possible to detect and reliably remove the minute noise inherent in the signal or the late noise that can be added by the signal amplification in advance.

2 is a block diagram showing the structure of the variable amplification and signal conversion unit. The matching signal Xmatch (t), which is the input analog signal, is variably amplified to an arbitrary amplification degree over the entire frequency band through the variable amplification block 111, and the variable amplified signal is amplified by the A / D conversion block 112 And converted into a digital signal Xvar (n).

The variable amplification process refers to a process in which if the input signal is larger than the reference level, it is regarded that the signal is a sound pressure signal from a microphone whose impedance is close to 150 O and the amplification degree is appropriately set. If the input signal is smaller than the reference level, And the amplification degree is increased.

The generated digital conversion signal Xvar (n) is provided to the tone color and threshold level processing unit 130 and the signal analysis unit 120, respectively.

The signal analyzer 120 analyzes the frequency component from the digital converted signal Xvar (n) to generate spectral information Xspec (n). Spectrum information can be used to analyze frequencies including levels and instantaneous volume changes between frequency bands. That is, the spectrum information can be utilized to obtain the level information for each frequency and the instantaneous level change of the frequency band for each frequency band. Thus, it is possible to find the tone color configuration information, the instantaneous sound peak value detection, and the feedback oscillation sound signal.

Spectrum information consists of the following form. Xspec (n) = [gN FilterN] _T x Xvar (n). In [gN FilterN] _T , each filter stage is arranged in parallel, and spectral information is generated by analyzing and digitizing each frequency component.

The weight information for the variable amplification processing performed by the variable amplification signal conversion unit 110 and the signal processing for the tone color and the threshold level processing unit 130 can be obtained from the spectrum information.

Here, the weight may be a weight for adjusting the degree of filtering of the tone color and the multi-stage serial filters constituting the threshold level processing unit 130.

3 is a block diagram showing the structure of the signal analysis unit. The digital conversion signal Xvar (n) has different frequency passbands and can be simultaneously input to the filters connected in parallel. Each filter analyzes signals according to frequency bands, (Xspec (n)).

Particularly, the signal analysis unit 120 performs spectral analysis using a signal before the full-scale speech processing, that is, the digital conversion signal Xvar (n). By using the digital converted signal Xvar (n) before speech processing in this manner, the original noise and the instantaneous sound pressure change amount included in the sound pressure signal can be accurately analyzed and the frequency can be specified. In addition, the weights necessary for the filtering and level control used in the voice processing can be accurately determined by the analysis result.

The tone color and threshold level processing unit 130 generates a tone color and a threshold level based on the spectral information generated by the signal analysis unit (by using the weight value and the noise displacement control variable generated by the system operation unit based on the spectrum information, (Xvar (n)) for each frequency band to control the tone color of the desired frequency, remove unnecessary sound sources, tone match, and limit the level of the beep signal to within the threshold level, n).

By limiting the level of the beep sound to be within the threshold level, it is possible to prevent the discomfort and the hearing impairment to the listener due to the momentary loud noise when the amount of ambient noise increases momentarily.

4 is a block diagram showing the structure of a tone color and threshold level processing unit. The amplified digital converted signal Xvar (n) passes through the plurality of filter blocks 131 to perform tone color matching, and the level of the signal sound is limited to within the threshold level through the limiter block 132.

That is, by controlling the gain of each of the cascade-connected multistage filters (Filter1, Filter2, ... FilterN) constituting the filter block 131, The tone color is controlled, and a pre-processing is performed in which unnecessary sound sources are removed. This allows tuning the timbre of the desired frequency and controlling the frequency at which the feedback oscillation occurs.

In addition, to detect the negative peak value by the moment limiter block 132, and processing the damping moment -ΔG_ _Limit.

Thus, the processed signal to the threshold (Xthl (n)) = -ΔG_ Limit × ([gN FilterN] × Xvar (n)).

The noise region selection unit 140 samples the signal to check the noise distribution from the threshold processing signal Xthl (n) on which the tone color and the threshold level processing has been performed and processes the signal as a minyoku block. Signal is inversely transformed by a signal corresponding to a component of the input signal, processed by the sub-signal block buffer to select a noise region, and the noise level adjustment signal Xnad (n) is generated by adjusting the level of the selected noise region.

The role of the noise region selection unit is to improve the clarity of speech by preliminarily processing noise regions of maximum and minimum values in consideration of the gain between the frequency bands of the signals as the noise amount of the signal increases.

5 is a diagram showing a structure of a noise region selection unit. The threshold processing signal Xthl (n) is input to the minihow block and the sub-signal block, respectively.

The minuscule block 141 processes the transfer function Hp (n) = Fs sum cycle [Xthl (n)] to the threshold processing signal Xthl (n) to generate a mnemonic.

The sub-signal block 142 generates a sub-signal by subjecting the critical processing signal Xthl (n) to a transfer function Hn (n) = -Fs sums [ΔXthl (n)], The instantaneous fluctuation noise is improved by the sum of? Gg [Hn) n)) to which the gain of the displacement level adjusting block 143 is applied and the sum Hp (n). Thus, the periodic noise processor 150 processes only the periodic noise region to remove noise.

Subsequently, the sub-signals having passed through the subtractor and the block sum variable displacement level adjusting block 143 are added together to generate the noise level adjusting signal Xnad (n).

The periodic noise processing unit 150 generates a periodic noise reducing signal Xnps (n) by attenuating the periodic noise in the input noise level adjusting signal Xnad (n).

6 is a diagram for explaining the structure of the periodic noise processing unit. The cyclic noise processing unit 150 includes a cyclic and noise signal sum processing block 151, a cyclic and noise signal sum inverse processing block 152, an adaptive Fir Filter LMS trunk block 153, And a block for summing and feeding back the signals. LMS means Least Mean Square (LMS) processing.

The noise level adjustment signal Xnad (n) is input to the period and noise signal sum processing block 151 and the period and noise signal sum reverse processing block 152, respectively.

The period and noise signal sum process block 151 generates a period period Xns (n) by performing a function H (n) on the noise level adjustment signal Xnad (n).

The period and noise signal summation processing block 152 processes the noise level adjustment signal Xnad (n) as a function H '(n) to generate a period signal X'ns (n).

The adaptive Fir Filter LMS short block 153 attenuates the periodic noise using the period mime (Xns (n)), the period signal (X'ns (n)) and the error signal (Xerr (n)).

Fir Filter (Finite Impulse Response Filter) is a type of digital filter that performs filtering on a desired frequency component with only certain values of input signals, and has a linear phase because there is no regression component. The Fir filter acts like an analog filter in terms of phase control, thereby making it a fundamental filter that is essential for performing phase signal processing.

On the other hand, IIR Filter (Infinite Impulse Response Filter) has a problem that the phase is nonlinear due to the regression component, which is problematic for use in periodic noise cancellation.

By using the Fir filter, it is possible to effectively attenuate the noise because the phase is maintained even if the inverse signal is synthesized in the repetitive periodic signal processing.

That is, at the FIR Filter LMS stage 153, the active coefficient, which is a result weight according to the LMS process, is obtained and applied to an arbitrary Fir Filter coefficient to perform signal processing. The Fir Filter coefficient is controlled by the LMS result weight, which is the active coefficient generated according to the LMS processing. The T _period or G _weight required for calculation of the _weight may be generated in the FIR Filter LMS stage 153 or may be provided from the system operation unit 190. Thereby, the cycle of the periodic noise processing unit 150 and the cyclic signal processing of the noise signal sum processing block 151, the periodic noise signal sum inverse processing block 152, and the adaptive Fir Filter LMS unit 153 are performed. The adaptive Fir Filter LMS stage 153 receives the differences Xnps (n) from the adaptive Fir Filter LMS stage 153 and the output Xns (n) from the period and noise signal sum process stage block 151, Signal is input as the error signal Xerr (n), which means that if the error signal Xerr (n) approaches 0, the frequency component of the periodic noise is attenuated.

This process is performed by inputting the period signal (X'ns (n)) and outputting the output signal reflecting the adaptive Fir Filter coefficient, that is, the error signal generated when the value Xnps (n) (Xnr (n)) is repeatedly adjusted until the value Xerr (n) becomes close to zero.

The error signal Xerr (n) represents the periodic noise included in the signal (for example, the noise level adjustment signal Xnad (n)) input to the period noise processing unit 150. [

On the other hand, Fir Filter the tap (tap) has a delay, each tap weight _{(W 0 (n), W} 1 (n), W 2 (n), ... Wm (n)) desired according to the application of You can implement filtering to control the frequency. After performing this processing, the periodic noise (or the error signal Xerr (n)) is determined in the LMS result. The periodic noise is attenuated by repeating the adjustment of the active coefficient until the error signal approaches zero.

If the error signal Xerr (n) is zero or down to any suitable level, the periodic noise attenuation signal Xnps (n), which is the output of the Fir Filter LMS stage 153, indicates that the frequency component corresponding to the periodic noise has attenuated And is entered into the next component.

7 is a view for explaining the structure of an adaptive Fir Filter LMS short block. The adaptive Fir Filter LMS term block 153 processes the period inverse signal X'ns (n) to recognize the periodic noise contained in the noise level adjustment signal Xnad (n). Further, it performs signal processing reflecting the LMS activity coefficient according to the error signal Xerr (n).

The error signal Xerr (n) may be defined as a difference generated by summing the periodic noise Xns (n) and the periodic noise reduction signal Xnps (n) output from the periodic noise processing unit, . Thus, if the magnitude of the error signal Xerr (n) is greater than the amount set for noise attenuation, the LMS active coefficient of the adaptive Fir Filter LMS stage block 153 is further adjusted to further attenuate the periodic noise, The periodic noise cancellation signal Xnps (n) may be again summed with the periodic mantissa Xns (n).

By repeating this feedback procedure, it can be determined that the periodic noise is attenuated in the periodic noise cancellation signal Xnps (n) when the error signal Xerr (n) approaches zero.

The reason for using the FIR filter in the present invention is to maintain the linear phase of the signal to be subjected to noise reduction. That is, since the Fir filter has linearity, the periodic signal can be removed linearly when mixing the opposite phase of the periodic signal (i.e., noise).

Periodic noise reduction signal (Xnps (n)) is a passing through the Fir Filter that _{X'ns (n) × W m (} n) signal, Xnps (n) = Σ W m (n) × X'ns (nm ).

Periodic noise processing unit 150, by applying a signal (X'ns (n)) Fir adaptive Filter coefficient (W _m (n)) on using the active algorithm, to minimize the signal is recognized as an error of the mean square have.

6, Xerr (n) = Xns (n) + Xnps (n). If the transfer function of the periodic and noise signal sum processing stage 151 is H (n) and the transfer function of the period and noise signal sum inverse processing stage 152 is H '(n), Xns (n) (N) = Xnad (n) x H (n), and similarly, X'ns (n) = Xnad (n) x H '(n).

Finally, Xerr (n) = Xns (n) + Xnps (n), and taking the mean square root of the active algorithm, is summarized as follows.

Xerr (n) = 2 Xerr (n) x (dXerr (n) / dW (n))

= 2 Xerr (n) × d {Xns (n) + (X'ns (n) × [W m (n)] T)} / dW (n)

That is, Xns (n) and X'ns (n) from each other is because the signals in reverse phase relations, [W _m (n)] according to the active factor, such as a vector table _T [W _m (n)] of the cyclic mutual offset _T When processing is performed, Xerr (n) will be close to 0, and finally Xnps (n) will contain only an aperiodic signal.

On the other hand, it is usually possible to remove periodic signals with only the LMS active algorithm. However, it is necessary to judge the extent to which it will be regarded as a periodic noise and to judge to what extent the determined periodic noise will be canceled.

This determination is made by the system operation unit 190 and the periodic and noise signal sum process stage 151, the periodic noise and noise signal sum process stage 152 and the adaptive Fir Filter LMS stage 153, Can be adjusted by controlling G (weight).

The noise displacement controller 160 controls the level of the period noise reduction signal Xnps (n) to generate a level displacement adaptive signal Xnvc (n) adapted to the level displacement curve.

If the degree of canceling the periodic signal from the signal containing speech is increased, the induced noise causing the afterimage effect may be rather increased. In order to solve such a side effect, the noise displacement controller 160 is applied.

8 is a diagram for explaining the structure of the noise displacement adjusting unit.

The signal including the speech is output through the periodic noise processing unit 150 and the periodic noise canceling signal Xnps (n) whose periodic component is attenuated is outputted. The noise displacement adjusting unit 160 adjusts the periodic noise reducing signal Xnps )) Of the induced noise portion still remains.

The noise displacement controller 160 includes a pre-processing gain stage Pre G, a noise displacement gradient control stage 161, and a post-processing gain stage (Post G). The pre-processing gain stage and post-processing gain stage are applied to level the signal. The noise displacement gradient control stage 161 is applied to process a portion of the induced noise portion.

9 is a diagram showing the relationship between the input value and the output value in the structure of the noise displacement gradient control stage and the noise displacement gradient control stage of the noise displacement adjustment unit.

The noise displacement gradient control stage 161 as shown in FIG. 9 (a) is used to attenuate the induced noise portion that is unintentionally added by the processing in the period noise processing unit 150, (Filter1, Filter2 ... FilterN) and a gain slope control (GSC).

The level noise suppression signal Xnps (n) is a level displacement adaptive signal Xnvc (n) by controlling the level of the frequency band in which the volume control, the residual image noise control, and the induced noise appear. The level displacement adaptive signal Xnvc (n) is generated as follows. Xnvc (n) = Pre G x [gN FilterN] x GSC x Post G.

Here, [gN FilterN] is a filter for controlling the level of the frequency band in which the induced noise appears, and may be a notch filter, a chef filter, a peak filter, or the like.

In the gain gradient control stage (GSC), the residual-image noise control process is performed by controlling the distortion of the sound wave according to Attack, Release, Sustain, Decay Time and the gain slope according to the volume.

9 (b) shows an example of controlling the level in the gain gradient control stage GSC. In the figure, the horizontal axis represents the input value of the signal, and the vertical axis represents the output value of the signal. In the drawing, a straight line having a constant slope represents a reference line in which a ratio of input to output of the signal linearly increases. At this time, in the system operating unit 190, as shown by a displacement line, arbitrary input values can be controlled in such a manner that the gain is set higher than the reference line and the other part is lowered in level than the reference. Thus, the residual image noise control process can be performed.

The signal processing matching unit 170 compares the threshold processing signal Xthl (n) output from the tone color and threshold level processing unit 130 with the noise level adjustment signal Xnad (n) output from the noise region selection unit 140, (T1 and t2) between the input signals and the periodic noise cancellation signal (Xnps (n)), and performs a process of compensating for the time delay A time-frequency matching signal Xtm (n) is generated by applying a predetermined frequency band pass filter to improve the generated noise.

10 is a diagram for explaining the structure of the signal processing matching unit. The threshold process signal Xthl (n) and the noise level adjustment signal Xnad (n) may be input to the time delay stage 171. [ Since the threshold processing signal Xthl (n) is ahead of the period noise reduction signal Xnps (n) by the time? T1, a delay of the time? T1 is required and the noise level adjustment signal Xnad n) is ahead of the periodic noise reduction signal Xnps (n) by the time? t2, a delay of the time? t2 is required.

Therefore, the time delay stage 171 is processed so as to have the same time delay by applying appropriate delay processing to each of the threshold processing signal Xthl (n) and the noise level adjusting signal Xnad (n).

The signal to which the time delay is applied passes through the band filter block 172 and passes only the frequency of the desired band to become the time matching signal Xtm (n). The time matching signal Xtm (n) is provided to the noise adjustment combining section 180. [

The time matching signal Xtm (n) may be a time delay processed threshold processing signal Xthl (n) or a time delay processed noise level adjusting signal Xnad (n) Lt; / RTI > That is, in the signal processing matching unit The threshold processing signal Xthl (n) and the noise level adjusting signal Xnad (n) are input together and the respective delay times are matched and the output can be output as one Xtm (n).

An appropriate delay process may also be applied to the periodic noise reduction signal Xnps (n) of the period noise processing unit 150 and the level displacement adaptive signal Xnvc (n) of the noise displacement adjustment unit 160, When the time matching signal Xtm (n), the periodic noise reducing signal Xnps (n), and the level displacement adaptive signal Xnvc (n) input to the noise adjustment combining unit 180 are synthesized with each other, A ringing phenomenon caused by a delay can be prevented.

The noise-adjusted synthesis unit 180 receives the processed signals output from the respective blocks, selectively processes the received signals, compensates for the tones according to the adaptive environment, and removes the periodic noise, thereby obtaining a resultant speech signal Xnc (n) ). That is, the resultant speech signal Xnc (n) is generated by mixing any one or two or more of the periodic noise reducing signal Xnps (n), the level displacement adaptive signal Xnvc (n) and the time matching signal Xtm (n) ).

11 is a view for explaining the structure of the noise control section.

The noise adjustment combining unit 180 includes an amplifying stage G1 for receiving the level displacement adaptive signal Xnvc (n) from the noise displacement adjusting unit 160 and adjusting the gain thereof, An amplification stage G2 for receiving and adjusting the gain Xnps (n) and an amplification stage G3 for receiving and adjusting the time matching signal Xtm (n) from the signal processing matching unit 170 Respectively. And a signal synthesis stage 181 for selecting and outputting any one of the signals adjusted by the amplification stage or synthesizing two or more signals and outputting the synthesized signal as a resultant speech signal Xnc (n).

The resultant speech signal Xnc (n) selected and synthesized by the amplification stages G1, G2 and G3 and the signal synthesis stage 181 can be expressed as follows. Xnc (n) = (G1 x Xnvc (n)) + (G2 x Xps (n)) + (G3 x Xm (n)).

Here, the gains G1, G2 and G3 of the amplification stage can be adjusted according to the degree of noise attenuation desired by the listener.

Conventional noise attenuators do not synthesize and process a plurality of input signals, but merely turn on / off the processed and processed signals connected to one signal path to the output terminal (for example, for noise attenuation, It is difficult to adapt to the change of surrounding environment (that is, change of the adaptive environment) and it is difficult to respond to the request of the listener.

However, the noise attenuator of the present invention diversifies the degree of noise attenuation by variously selecting signals using the amplification stages G1, G2, G3 and the signal synthesis stage 181 of the noise control synthesis unit 180, So that it is possible to effectively adapt to the surrounding environmental factors and the demands of the listener.

As an example of the control operation, when it is desired to reduce the degree of noise attenuation, only the time matching signal Xtm (n) may be output by setting G1 = 0, G2 = 0 and G3 = G2 + G3 = 1). In this output, tones, feedback oscillations, and threshold level controlled audio can be heard.

In the case where the degree of noise attenuation is to be maximized, for example, values of G1 = 0.4, G2 = 0.6, and G3 = 0 can be applied. Thus, the level displacement adaptive signal Xnvc (n) and the periodic noise cancellation signal Xnps (n) can be synthesized and output. In this output, the timbre, the feedback oscillation and the threshold level can be controlled, the periodic noise can be attenuated, and the level can be adjusted.

On the other hand, when listening to a voiced sound by receiving a certain amount of original sound, it is only necessary to set a small weight without setting G3 to zero.

This control can be expressed as the following equation.

In the case of Xnc (n) = (G3 x Xtm (n)), only the time matching signal Xtm (n) is outputted as G3 varies. According to such a signal, it is possible to realize a voice in which tone color, feedback oscillation and threshold level are controlled, and the instantaneous variation noise is improved by selection of a noise region.

The level displacement adaptive signal Xnvc (n) and the periodic noise attenuation signal Xnps (n) are varied in accordance with the variation of G1 and G2 in the case of Xnc (n) = (G1 x Xnvc (n)) + (G2 x Xps n) are synthesized and output. By this signal, the timbre, the feedback oscillation and the threshold level are controlled, the instantaneous fluctuation noise improvement and the periodic noise are attenuated, and the level-adjusted voice can be realized.

In the case of Xnc (n) = (G1 x Xnvc (n) + G2 x Xps (n) + G3 x Xm (n), various signal synthesis is performed according to the variable G1, G2, It can be implemented according to the adaptation environment.

That is, G1, G2, and G3 are variable values that are not fixed values, and can be applied to vary depending on the adaptive environment.

The noise attenuator according to an embodiment of the present invention according to the present invention selectively attenuates only the noise components generated by the amplification of the introduced ambient noise and the high amplification degree from the sound pressure signal received by only one low impedance microphone By providing one or more of the various output signals (Xthl (n), Xnad (n), Xnps (n), Xnvc (n)) generated in each sound processing step to the listener by weight- It is possible to listen to the sound that is well-liked.

It is also possible to provide a high level of clarity by matching the time delays generated between the various output signals Xthl (n), Xnad (n), Xnps (n), and Xnvc (n) and adjusting the frequency band.

Claims (9)

An apparatus for attenuating periodic noise from a sound pressure signal received by a single microphone, comprising:
A variable amplification and signal conversion unit for digitally converting the sound pressure signal generated by the microphone to generate a digital conversion signal Xvar (n);
A tone color and threshold level processing unit for receiving the digital converted signal Xvar (n) to control the tone color and to generate the threshold processed signal Xthl (n) by limiting the threshold level;
A noise region selection unit for selecting a noise region from the critical processing signal and adjusting a level of the selected noise region to generate a noise level adjustment signal Xnad (n);
(Nn) and the period inverse signal X'ns (n) using the noise level adjustment signal Xnad (n), and outputs the period inverse signal X'ns (n) Generates an error signal Xerr (n), which is a difference signal between the output signal and the periodic mantel (Xns (n)), and outputs the error signal Xerr If the magnitude of the signal Xerr (n) is greater than a desired degree, the output signal is adjusted and then combined with the periodic mantel Xns (n), and the generated error signal Xerr (n) And outputting the final output signal as the periodic noise attenuated signal Xnps (n) when the periodic noise is attenuated to a level lower than the predetermined periodic noise suppression signal Xnps (n).
A noise displacement controller for controlling a level of the period noise reduction signal Xnps (n) to generate a level displacement adaptive signal Xnvc (n) adapted to an arbitrary level displacement curve;
And outputs the resultant speech signal Xnc (n) in which the periodic noise is attenuated by giving an arbitrary weight to the periodic noise reduction signal Xnps (n) and the level displacement adaptive signal Xnvc (n) A noise adjusting / synthesizing unit for outputting the noise through the speaker; And
And a system operation unit for controlling operations of the variable amplification and signal conversion unit, the periodic noise processing unit, the noise displacement adjustment unit, and the noise adjustment combining unit. delete The method according to claim 1,
(T1 and t2) of the threshold processing signal Xthl (n) and the noise level adjusting signal Xnad (n) with respect to the period noise reduction signal Xnps (n) (Nth) and the noise level adjusting signal Xnad (n), so that the noise level adjusting signal Xthl (n) and the noise level adjusting signal Xnad (n) Further comprising a signal processing matching unit for generating a time matching signal Xtm (n) including at least one,
The noise adjustment combining unit further receives the time matching signal Xtm and gives a weight to the period noise reducing signal Xnps (n) and the level displacement adaptive signal Xnvc (n) And outputs the resultant speech signal Xnc (n). The method of claim 3,
Further comprising delaying each of the periodic noise reduction signal (Xnps (n)) and the level displacement adaptive signal (Xnvc (n)) so as to have the same phase as the time matching signal (Xtm (n) Noise attenuator. 5. The method of claim 4,
The noise-
(N) and the time-matched signal Xtm (n) by assigning an arbitrary weight to each of the periodic noise reduction signal Xnps (n), the level displacement adaptive signal Xnvc (Xnc (n)). ≪ / RTI > The method according to claim 1,
Wherein the periodic noise processing unit comprises:
A noise signal sum process block and a period for processing the input signal by the first transfer function H (n) to generate the period mantle Xns (n)
And a period for processing the input signal by a second transfer function H '(n) to generate the period inverse signal X'ns (n)
An adaptive Fir Filter LMS stage block (LMS block) which applies Fir Fiter to the period inverse signal (X'ns (n)) and LMS-processes the signal generated by applying the error signal (Xerr Wherein the noise attenuator is a noise attenuator. The method according to claim 1,
Further comprising a signal analyzing section for analyzing a frequency component of the digital converted signal Xvar (n) to generate spectral information (Xspec (n)),
Wherein the system operating unit changes weighting values of filtering or level control used in the respective units based on the spectrum information (Xspec (n)). CLAIMS What is claimed is: 1. A method of attenuating periodic noise from a sound pressure signal received by a single microphone, comprising:
Converting the sound pressure signal generated by the microphone into a digital converted signal Xvar (n);
Receiving the digital converted signal Xvar (n) to control the tone color and limit the threshold level to generate a threshold processed signal Xthl (n);
Selecting a noise region in the critical processing signal and adjusting a level of the selected noise region to generate a noise level adjusting signal Xnad (n);
(Nn) and the period inverse signal X'ns (n) using the noise level adjustment signal Xnad (n), and outputs the period inverse signal X'ns (n) Generates an error signal Xerr (n), which is a difference signal between the output signal and the periodic mantel (Xns (n)), and outputs the error signal Xerr If the magnitude of the signal Xerr (n) is greater than a desired degree, the output signal is adjusted and then combined with the periodic mantel Xns (n), and the generated error signal Xerr (n) And outputting the final output signal as the periodic noise-attenuated periodic noise cancellation signal (Xnps (n)) if the periodic noise is attenuated to a level below
Generating a level displacement adaptive signal Xnvc (n) adapted to an arbitrary level displacement curve by controlling the level of the periodic noise attenuation signal Xnps (n); And
And outputs the resultant speech signal Xnc (n) in which the periodic noise is attenuated by giving an arbitrary weight to the periodic noise reduction signal Xnps (n) and the level displacement adaptive signal Xnvc (n) So as to be output through a speaker. 9. The method of claim 8,
(? T1 and? T2) of the threshold processing signal Xthl (n) and the noise level adjusting signal Xnad (n) with respect to the period noise reduction signal Xnps (n) (Nth) and the noise level adjusting signal Xnad (n) so that the signal Xthl (n) and the noise level adjusting signal Xnad (n) Generating a time matching signal (Xtm (n)) including at least one of:
(N) and the time-matched signal Xtm (n) by assigning an arbitrary weight to each of the periodic noise reduction signal Xnps (n), the level displacement adaptive signal Xnvc (Xnc (n)). ≪ / RTI >

Images