KR102163373B1 - Apparatus and method for auditory management, and apparatus for attenuating noise at sound signal inputted from one microphone - Google Patents

Abstract

The hearing management apparatus disclosed in the present invention includes: an acoustic range determination unit for determining an ambient sound within a predetermined range from the reference sound when a predetermined reference sound specialized for a subject is set, and outputting the reference sound and each of the ambient sounds An output time determining unit for determining an output time for each sound to be performed, an sound time matching unit for generating sound output control information by matching the sound output time to each of the reference sound and the ambient sound, and the sound output control information It includes an output sound generator for generating the auditory management sound to be output to the speaker based on.

Description

Hearing management device and method, and a device that attenuates the noise of a single microphone voice signal with a hearing management function {APPARATUS AND METHOD FOR AUDITORY MANAGEMENT, AND APPARATUS FOR ATTENUATING NOISE AT SOUND SIGNAL INPUTTED FROM ONE MICROPHONE}

The present invention relates to a hearing management apparatus and method, and more particularly, to an apparatus and method capable of improving or protecting an auditory function by outputting a sound of a frequency specialized to a subject. In addition, the present invention relates to an apparatus for attenuating noise of a single microphone voice signal capable of implementing a hearing management method by having such a hearing management device.

BACKGROUND ART A method and apparatus for stimulating auditory cells using an acoustic signal capable of improving hearing or improving ringing phenomena such as tinnitus and hearing loss disorders are disclosed. Korean Patent Registration No. 10-0963888 (2010.06.08.) (hereinafter referred to as prior art).

In the prior art, the user's hearing threshold is determined by providing a beep sound with a resolution of 1/k octave, and based on the determined hearing threshold, the frequency of the damaged auditory cell region is determined, and a predetermined intensity of the corresponding target frequency band is determined. By outputting an acoustic signal, the user's auditory cells are stimulated.

However, in the prior art, the object is to intensively stimulate a specific auditory cell region by intensively outputting one sound in a specific frequency band. However, if only a specific auditory cell region is intensively stimulated in this way, a deviation of the sensing ability between the intensively stimulated auditory cell region and the surrounding auditory cell region may occur.

Moreover, in practice, if one sound in a specific frequency band for intensive stimulation is continuously output, the user only adapts to the output sound, and there is a problem in the overall stimulation effect.

Further, in the prior art, it is described that the acoustic signal is composed of one or a combination of amplitude modulated sound, frequency modulated sound, pulse sound, continuous sound, and amplitude modulated narrowband noise, but no specific method is described.

In the present invention, further from the prior art as described above, which stimulates only a very narrow band part in the audible region of the auditory structure or gives a wide stimulus to all regions, the sound of the adjacent part of a specific region is sequentially sweeping. It is intended to increase the effect of hearing management by applying changes in a random manner.

The hearing management apparatus according to an embodiment of the present invention includes: a sound range determination unit configured to determine an ambient sound within a predetermined range from the reference sound when a predetermined reference sound specialized for a subject is set; An output time determination unit determining an output time for each sound for outputting the reference sound and the ambient sound; A sound time matching unit for generating sound output control information by matching the sound output time to each of the reference sound and the ambient sound; And an output sound generator configured to generate an auditory management sound to be output to a speaker based on the sound output control information.

Here, the frequency of the reference sound and the ambient sound is calculated by a formula of {(reference frequency)×2^(M/L)}. In this case, M represents the Mth scale from the reference frequency, and L represents the number of scales to divide one octave from the reference frequency.

In addition, the reference sound and the ambient sound may be selected based on the scale of the piano.

In addition, the ambient sound may be selected from a scale in the range of 1/K octave from the reference sound (here, K may be one of 1, 2, 1/2, and 1/4).

On the other hand, the sound range determination unit divides the audible sound frequency range into a plurality of bands having a predetermined 1 octave unit and partially overlapping each other, and when a specific band including the set reference sound is specified, the specified All or some scales included in the band may be selected as the ambient sound.

In addition, the output time determination unit, in a sweep method for sequentially outputting the reference sound and the ambient sound according to a frequency order, a sweep output time for each sound to be outputted, and a frequency order of the reference sound and the ambient sound It is possible to determine a random output time for each note in which each note is to be output in a non-sequentially output random method.

In this case, the sweep output time for each sound is at least a minimum time for recognizing the existence of each sound being output in the sweep method, or less than a maximum time at a time when not recognizing that each sound is cut off, Can be determined.

Further, the auditory management sound is configured to be output according to a randomly mixed method of a sweep method in which the reference sound and the ambient sound are sequentially output according to a frequency order and a random method that is output in a non-sequential order that does not follow the frequency order. Can be.

The hearing management method according to another embodiment of the present invention includes the steps of determining an ambient sound within a predetermined range from the reference sound when a predetermined reference sound specialized for a subject is set; Determining an output time for each sound for outputting the reference sound and the ambient sound, respectively; Generating sound output control information by matching the sound output time to each of the reference sound and the ambient sound; And generating an auditory management sound to be output through a speaker based on the sound output control information and applying it to the subject.

Here, the reference sound and the ambient sound may be selected based on the scale of the piano.

On the other hand, the step of determining the output time for each sound includes determining a sweep output time for each sound to output each sound in a sweep method sequentially outputting the reference sound and the ambient sound according to a frequency order, and the reference sound and The method may further include determining a random output time for each sound in which each sound is to be output in a random manner in which the ambient sound is output in a non-sequential manner that does not follow a frequency order.

In addition, the generating of the auditory management sound includes a method in which a sweep method in which the reference sound and the ambient sound are sequentially output according to a frequency order and a random method that is output in a non-sequential order not in a frequency order are arbitrarily mixed. The auditory management sound may be generated to be output.

According to another embodiment of the present invention, an apparatus having an auditory management function and attenuating periodic noise from a voice signal by a single microphone includes: detecting a periodic signal from a signal digitally converted from a voice introduced into a single microphone, A periodic sound attenuation control unit configured to output a signal from which the periodic signal is attenuated by mixing the digitally converted signal with the reverse phase of the periodic signal; An adaptation level range controller configured to output a level-adapted signal by controlling the level of the attenuated signal; An attenuation adjustment selector configured to mix a signal from which the periodic signal output from the periodic sound attenuation control unit is attenuated and one or both of the level-adapted signal output from the adaptive level range control unit and output as a processed audio signal; When a predetermined reference sound specialized for the subject is set, an ambient sound within a predetermined range from the reference sound is determined, an output time for each sound for outputting the reference sound and the ambient sound is determined, and the reference sound and the surrounding sound An auditory management unit for generating sound output control information by matching the sound output time to each sound, and generating an auditory management sound to be output to a speaker based on the sound output control information; And a signal selection unit for providing at least one of the processed voice signal or the auditory management sound to a speaker.

Here, the periodic sound attenuation control unit comprises: a signal block sum processing unit processing the digitally converted signal as a mental signal; A signal block sum inverse processing unit that processes the mental signal into an inverse signal having an inverse phase; By calculating the result signal with the mental code to determine the periodic signal, generating random adaptive Fir Filter coefficients divided for each frequency to attenuate the determined periodic signal, and applying the coefficients to the inverse signal by weighting It may further include an adaptive Fir Filter LMS stage which repeats processing the signal from which the periodic signal is attenuated as the result signal.

In addition, the apparatus having an auditory management function and attenuating periodic noise from a voice signal by a single microphone, in order to detect a frequency at which a feedback oscillation sound exists from the processed voice signal output from the attenuation control selection unit, A signal analysis unit for generating spectrum information by analyzing a frequency component of the processed speech signal; And a sound matching unit configured to analyze a frequency including the feedback oscillation from the spectrum information, filter the analyzed frequency from the digitally converted signal to output a matched signal, and the attenuation control selector further includes the periodic signal Any one of the attenuated signal, the level-adapted signal, and the matched signal, or a mixed signal of two or more may be output.

Further, the auditory management unit does not follow the sweep output time for each sound to output each sound in a sweep method sequentially outputting the reference sound and the ambient sound according to the frequency order, and the reference sound and the ambient sound frequency order. In a non-sequentially output random method, a random output time for each sound to be output is determined, and the reference sound and the ambient sound are output in the sweep method according to the sweep output time for each sound, and the random output for each sound The auditory management sound may be generated by randomly combining outputs in the random manner over time.

The present invention including the configuration as described above, by variously applying the auditory management sound including not only a specific frequency for which auditory stimulation is required to the subject, but also a proximity frequency region around the frequency, in a sweep method and a random method, Maximize the effect.

In addition, since the ambient noise can be attenuated from the voice signal input through one microphone, the voice signal is distorted as the voice signals from two or more microphones are mutually processed, and additional noise is caused when processing the voice signal. It is possible to provide a noise attenuating device capable of solving the problem. In particular, a device capable of simultaneously or selectively providing the noise reduction function and the hearing management function may be provided.

1 is a block diagram showing the configuration of a hearing management apparatus according to the present invention.
2 is a diagram for describing in more detail an acoustic range determination unit.
3 is a diagram for describing in more detail an output time determination unit.
4 is a diagram for describing in more detail an acoustic time matching unit.
5 is a diagram for describing in more detail an output sound generator.
6 is a flowchart illustrating a hearing management method according to the present invention.
7 is a block diagram showing the configuration of a device for attenuating noise of a single microphone voice signal having an auditory management unit implementing the hearing management function as described above.

Hereinafter, preferred embodiments of a hearing management apparatus and method according to the present invention and a device for attenuating noise of a single microphone voice signal having a hearing management function according to the present invention will be described with reference to the accompanying drawings. For reference, since terms referring to each component of the present invention are exemplarily named in consideration of their functions, the technical content of the present invention should not be predicted and limited by the term itself.

First, a configuration of a hearing management apparatus according to an embodiment of the present invention will be described with reference to the block diagram of FIG. 1. Referring to the drawing, the hearing management apparatus 200 includes an acoustic range determination unit 210 that determines an ambient sound within a predetermined range from the set reference sound when a predetermined reference sound specialized for a subject is set, and the set reference sound and the determined surrounding sound. An output time determination unit 220 that determines an output time for each sound for outputting each sound, and a sound time matching unit 230 that generates sound output control information by matching the sound output time to each of the reference sound and the ambient sound And, based on the sound output control information may include an output sound generator 240 for generating an auditory management sound to be output to the speaker.

In the present invention, the frequency of the ambient sound that can be determined from the reference sound can be calculated by the following equation.

(Frequency of each ambient sound) = {(reference frequency)×2^(M/L)}

Here, M may be an integer indicating that it is an Mth scale from the reference frequency. If M is a positive number, it may be a scale in a high direction from the reference frequency (treble direction), and if M is a negative number, it may be a scale in a low direction (bass direction) from the reference frequency. The L represents the number of scales for dividing a frequency in the range of 1 octave from the reference frequency, and may be a natural number.

In particular, the present invention can use a sound based on the scale of a piano. Being based on a piano scale can be interpreted as meaning that when determining the frequency of each note, the frequency defined for each scale of the piano defined based on A4=440Hz is used.

Thus, when an arbitrary reference frequency is input, a piano scale that is most relevant to the reference frequency (for example, the closest frequency value) can be selected and determined as the reference tone, and several scales in the treble direction around the reference tone. And/or by selecting several scales in the bass direction, it is possible to simply select the ambient sound in a desired range.

If the reference frequency exceeds the frequency of the existing piano scale (for example, when the frequency is equal to or higher than the frequency of the 88th note of the 88 scale), a virtual piano scale is constructed using the above formula to generate the reference sound and /Or you can decide the ambient sound.

It is also possible to use a more subdivided scale (even less subdivided into six scales per octave, etc.), such as a piano scale as the basis, but divided into 24 scales per octave.

The selection of the reference frequency can be determined by the subject who wants to manage the hearing designating a desired frequency value by himself or by searching for a frequency that needs to be managed using a professional measuring equipment. In the present invention, even if a specific reference frequency is selected, it is possible to manage even various frequency parts around it, so that the subject may incorrectly designate the reference frequency without using specialized equipment.

The reference frequency can be selected by designating a specific frequency value, a frequency range, or a specific piano scale.

Alternatively, selecting the reference frequency may include designating a band containing the reference frequency, and designating a section most relevant to the reference frequency within the designated band.

For example, the audible frequency band may be divided into 12 bands, as shown in [Table 1]. Each band consists of a start frequency and a section up to twice its frequency (ie, 1 octave).

Band number (BN) Frequency range B1 200 to 400 Hz B2 300 to 600 Hz B3 500 to 1k Hz B4 750 ~ 1.5k Hz B5 1k to 2k Hz B6 1.5k to 3k Hz B7 2k to 4k Hz B8 3k to 6k Hz B9 4k to 8k Hz B10 5k to 10k Hz B11 6k ~ 12k Hz B12 7k to 14k Hz

And, each band may be divided into a predetermined number of sections by the following equation.

(Frequency of each section) = {(Start frequency)×2^(M/L)}

Here, M denotes the M-th section from the start frequency, and L denotes the number of sections to divide one octave constituting the band. Sections can usually be defined as 12.

And, for example, if the subject has designated band 3 and section 3, the reference frequency may be determined as {(500)×2^(3/12)}=594.6Hz.

On the other hand, as a method that can be applied in this designation method, when the subject designates only a band, the upper 1/2 octave-lower 1/2 octave range (total 1) based on the entire frequency range of the band or the start frequency of the band. Octave range), and when the subject specifies a band and section, it is in the upper 1/4 octave-lower 1/4 octave range (total 1/2 octave range) based on the frequency of the section. It may be implemented to generate and output auditory management sound.

Or, similarly, when only a band is specified, the frequencies of the 12 sections constituting the band are output as ambient sound, and when a section is specified in addition to the band, a total of 1/2 octave range based on the section You can also control to output the frequencies of them as ambient sounds.

Alternatively, when only a band is designated, frequencies of all 12 sections constituting the band may be selected as ambient sound, and the selected ambient sound may be output in a sweep method. On the other hand, if both a band and a section are specified, the section is used as the reference sound, and all ambient sounds of the band are output in a sweep method, and alternately, the ambient sound up to the upper 1/4 octave and the lower 1/4 It can be configured to output the ambient sound up to an octave in a random manner. The sweep method and the random method will be described later.

Regarding the method of designating the band and section, in another example of selecting the reference frequency, it is assumed that the subject has designated 1.5 kHz and section 5. Here, it is assumed that there are 12 sections per 1 octave, and when a section is designated as well as a band, it is set to select an ambient sound within a total range of 1/2 octave based on the section. If so, B4, B5, and B6 in Table 1 may be selected as candidate bands. Meanwhile, since section 5 is designated, the reference frequency is 2002 Hz, so B6 can be selected. Subsequently, all ambient sounds of the selected band B6 may be output in a sweep method. In turn, a reference sound using section 5 as a reference frequency, and notes up to an upper 1/4 octave and notes up to a lower 1/4 octave therefrom may be output in a random manner.

On the other hand, the present invention is a sweep method that sequentially outputs sounds (reference sound and ambient sound) determined as auditory management sounds within a predetermined octave in an ascending or descending order of frequency, and a random order that does not follow the ascending or descending order of frequencies. Alternatively, it may be output in a random manner in which high and low sounds are reciprocated in a random non-sequential order.

In addition, each scale output in the sweep method and/or the random method may be output for a predetermined time. For example, the time at which each scale is output may be determined to be about 20 to 500 ms. The output time of each scale may be variously determined according to the frequency of the corresponding scale, according to an output method of whether to be output in a sweep method or a random method, and according to an order to be output within each method.

At this time, the output time, for example, in the sweep method, is more than a minimum time to recognize that each sound sequentially output has been output and/or a degree that is not recognized as if each sound sequentially output is cut off. It can be determined to be less than the maximum time of. In addition, in the random method, it may be determined that it is more than a minimum time for recognizing that each sound output has been output and/or less than a maximum time that gives discomfort to the subject or makes the auditory management effect meaningless.

The hearing management apparatus 200 may provide an interface (not shown) so that the subject may designate a band and/or section, a reference frequency, a frequency range, and the like. In addition, the subject can set and adjust various parameters of the hearing management device 200 through the interface, such as division of bands and sections, output method of hearing management sound, and octave range for selecting ambient sounds from each output method. will be.

2 is a diagram for describing in more detail an acoustic range determination unit. When the subject inputs at least the reference frequency, the sound range determination unit 210 may determine the reference frequency or the reference sound, and determine the ambient sound within a predetermined 1/K octave range around the reference frequency or the reference sound. Here, K may be an arbitrary number, and the octave range may be set to 4, 2, 1, 1/2, or the like.

In addition, the sound range determination unit 210 may determine an output order (sequence) of the reference sound and the selected ambient sound in a sweep method and/or a random method.

3 is a diagram for describing in more detail an output time determination unit. The output time determination unit 220 may determine an output time for each output method of the set reference sound and the ambient sound. For example, the output time (sweep output time for each sound) may be determined so that each sound of the sweep method changes sequentially in the range of 50 to 500 ms or randomly, and each sound of the random method is sequentially in the range of 20 to 160 ms. The output time (random output time for each sound) may be determined to vary or randomly.

4 is a diagram for describing in more detail an acoustic time matching unit. The sound time matching unit 230 configures a series of information (sound output control information) by matching information related to each sound for which a specific order is set and output time information determined for each sound.

5 is a diagram for describing in more detail an output sound generator. The output sound generator 240 may configure an analog signal (ie, auditory management sound) by modulating an arbitrary sample sound according to sound output control information. Alternatively, a certain sine wave may be used as a sample sound and the auditory management sound may be constructed by modulating the sound. The auditory management sound may be applied to earphones, headphones, sound pressure output devices, bone conduction speakers, and the like, and may be output to a subject. In addition, the hearing management sound may be generated as a sound for the left ear or the right ear, respectively.

The output sound generator 240 may generate a sound to be provided in a sweep method and a sound to be provided in a random method, respectively, and output them alternately or in an arbitrary order. In addition, when each method is alternated, a blank time can be inserted or output continuously without a blank.

6 is a flowchart illustrating a hearing management method performed by the hearing management apparatus according to the present invention.

The hearing management apparatus 200 may receive a reference frequency value through an interface (S10). The reference frequency may be designated in a manner of receiving a band and/or a section. Even, instead of receiving the reference frequency value, the reference sound may be directly selected.

When the reference frequency is input, the reference sound may be determined (S20). The reference sound may be the reference frequency itself, or may be determined from a piano scale that is a frequency closest to the reference frequency. In addition, a 1/K octave upper and lower range of the reference sound is set, and an arbitrary number of ambient sounds may be determined within the set range. The ambient sound may be determined as up to N high tones and N'low tones from the reference sound. N and N'may be the same number or different numbers. In this case, sounds for output in a sweep method and sounds for output in a random manner may be determined to be the same or different from each other.

Subsequently, the determined reference sound and/or the ambient sound is determined to be output time for each sound to be provided as an actual sound to the subject (S30). As for the output time for each sound, the output time for each sound (sweep output time for each sound) to be output in the sweep method and the output time for each sound (random output time for each sound) to be output in the random method may be equal to or different from each other.

Sound output for each output method by matching the sequence of the reference sound and ambient sound to be output in a sweep method and the sweep output time for each sound, and by matching the sequence of the reference sound and ambient sound to be output in a random method and the random output time for each sound Control information is generated (S40).

Based on the sound output control information for each output method, the auditory management sound for each output method is generated, and the auditory management sound is output through the speaker according to the order of the output method (S50).

In addition, the present invention provides an apparatus for attenuating the noise of a single microphone voice signal having a hearing management function implemented by the hearing management apparatus as described above. The configuration of such a noise reduction device is shown in FIG. 7.

The apparatus for attenuating the noise of a single microphone voice signal with an auditory management function in this embodiment includes a microphone input impedance matching and signal conversion unit 100, a sound matching unit 110, and a periodic sound analysis and attenuation control unit. 120, an adaptation level range control unit 130, an attenuation control selection unit 140, a spatial voice position control unit 150, a signal analysis unit 160, a main gain amplification unit 170, and a system It includes an operation unit 190. In addition, the hearing management device as described above has been added as a component named hearing management unit 200.

In particular, the processed voice from the spatial voice position control unit 150 and the hearing management sound from the hearing management unit 200 may be selectively provided to the main gain amplification unit 170 by the signal selection unit 195.

Detailed configuration of the constituent parts will be described later.

The microphone input impedance matching and signal conversion unit 100 matches the input from a single microphone with the impedance of the noise reduction device. In addition, an incoming signal (ie, a signal in which an audio signal and an ambient signal are mixed), which is an analog signal generated by one microphone, is converted into a digital signal. The microphone input impedance matching and signal conversion unit 100 includes an input impedance matching function and an A/D conversion function. The incoming signal input from the microphone is an analog signal and contains not only voice but also surrounding noise. The input impedance matching function realizes the maximum power transfer of the incoming signal.

The sound matching unit 110 creates a mixed signal by selecting one of a digitally converted signal generated from a microphone and a frequency signal that causes sound in a specific frequency band. The mixed signal is output as a matched signal by adjusting the tone for each frequency band.

In addition, a separate external signal that is not introduced from the microphone may be input to the sound matching unit 110. This external signal can be mixed with the digital signal. The external signal may be a signal transmitted from another headset or earset.

The sound matching unit 110 includes one or more of a digital signal output from the microphone input impedance matching and signal conversion unit 100, an external signal, and a signal generated through a signal generation function of the sound matching unit 110. It has a signal selection mix and mux function to select and mix. In addition, the signal output from the signal selection mix and mux functions is converted into a matched signal through a tone and feedback oscillation control function.

The noise attenuating device receives an external signal and processes the noise attenuation together with a microphone input signal to listen together. In this way, even when a signal that is not noise-attenuated is received from another headset or an earphone together, noise is attenuated so that only a clear voice can be heard. In particular, this function is useful when a plurality of headsets or earphones communicate with each other.

The signal generation function controls signal conversion characteristics and tone according to the microphone type to pre-process unnecessary sound sources, adjusts the system's feedback induction, and modifies the voice to be heard to a desired tone. Occurs. In other words, by generating and outputting a sound having a specific frequency in the signal generation function, and allowing the listener to manually adjust the volume of the sound of that frequency as desired, it is possible to adjust the tone/volume for each frequency suitable for the listener's auditory characteristics. Is done.

The signal output from the signal selection mix and mux functions may be any one of a single output of a microphone input signal, a mixed output of a microphone input signal and an external signal, and a single output of a signal output from the signal generation function.

The single output of the microphone input signal is for noise reduction processing only for the digitally converted signal introduced through the microphone. Mixing and outputting a microphone input signal and an external signal is for mixing and processing a digitally converted signal introduced through a microphone and a voice signal transmitted from another headset or earphone. And, the single output of the signal by the signal generation function is to tune the tone of the desired frequency from the voice signal, and to adjust the volume of each frequency band of the voice that the listener hears through the speaker according to his/her auditory characteristics. It is chosen when you do.

The tone and feedback oscillation control function can be composed of multi-stage filters (Filter1, Filter2, Filter3 ... FilterN). The multi-stage filters are arranged for the purpose of tuning the tone tone of the desired frequency and controlling the frequency at which the feedback oscillation occurs, for pre-processing of controlling the tone tone of a desired frequency and removing an unnecessary sound source. Matched signal = {gN FilterN} × Xm (n).

The control amount of each filter can be adjusted by the gains g1, g2, g3 ... gN. The gain of each filter may be adjusted by a control command (a control command commanding to control a feedback oscillation sound of a specific frequency) provided by the system operation unit 190.

Each filter may include a notch filter, a shelf filter, a peak filter, and the like.

The periodic sound analysis and attenuation control unit 120 detects and attenuates only a periodic signal having a periodic waveform pattern from the matched signal output from the sound matching unit 110. In the present invention, the periodic signal is determined as noise.

The periodic sound analysis and attenuation control unit 120 receives a matched signal in which the tone is controlled through the sound matching unit 110, preprocessing is performed to remove unnecessary sound sources, and the generation of feedback oscillation is adjusted as an input. Then, a noise attenuated signal obtained by attenuating only the periodic signal from among the signal components constituting the matched signal is output.

The periodic sound analysis and attenuation control unit 120 includes a signal block sum processing function, a signal block sum inverse processing function, and an adaptive Fir Filter LMS function. LMS stands for Least Mean Square processing.

The Fir Filter LMS function obtains active coefficients, which are weights resulting from LMS processing, and applies them to the Fir Filter coefficients to be adapted. The adaptive Fir Filter coefficients are controlled by active coefficients, which are weights resulting from LMS processing, and the T _period and G _weights may be provided from the system operation unit 190 or generated by the Fir Filter LMS function. Accordingly, the periodic sound analysis and attenuation control unit 120 performs cyclic signal processing of the signal block sum processing function, the signal block sum inverse processing function, and the adaptive Fir Filter LMS function. The adaptive Fir Filter LMS function generates an error signal, and the adaptive Fir filter coefficient is modified and repeated until the error signal becomes zero.

The Fir Filter generates a tap delay. After applying a weight W ₀ (n) to W _m (n) to each tap, a periodic signal (or an error signal) is determined from the LMS result. The noise is attenuated by repeatedly applying the active coefficient until the periodic signal becomes zero.

When the periodic signal becomes 0, it means that the periodic surrounding signal, that is, the surrounding noise, is mostly attenuated.

In processing the matched signal output from the sound matching unit 110, the periodic sound analysis and attenuation control unit 120 processes a signal consisting of a signal block sum processing function, a signal block sum inverse processing function, and an adaptive Fir Filter LMS function. By combining various paths, a signal with attenuated noise can be generated and output.

The signal block sum processing function obtains an error signal by comparing a block-by-block mental code for periodic signal recognition with a periodic noise reduction signal.

The signal block sum inverse processing function reflects the LMS active coefficient according to the error signal to the block-based inverse signal for periodic signal recognition.

As a result, when the error signal approaches zero by actively processing the mental signal and the inverse signal, a noise attenuated signal including only aperiodic signal is generated.

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 processing, and since the Fir Filter has linearity, when the reverse phase of the periodic signal is mixed, the periodic signal can be linearly removed. .

The periodic sound analysis and attenuation control unit 120 minimizes a signal perceived as an error of a mean square by using an active algorithm.

Here, the part recognized as an error signal is (error signal) = (block-based mental code) + (periodic noise reduction signal). Here, if the transfer function of the signal block sum processing function is Hs(n) and the transfer function of the signal block sum inverse processing function is H's(n),

(Block unit mental code) = (matched signal) × Hs(n), similarly, (block unit inverse signal) = (matched signal) × H's(n).

In the end, (error signal) = (block unit mental code) + (periodic noise reduction signal), and taking the root mean square of the active algorithm, it is summarized as follows.

▽²(Error signal) = 2 (Error signal)×(d(Error signal)/dW(n))

= 2 (error signal)×d{(block unit mental code) + ((block unit inverse signal) × [W _m (n)] _T )}/dW(n)

That is, the block unit jeongsinho and blocks reverse signal, so each signal in the reverse phase relations, [W _m (n)] Do [W _m (n)] _T periodically cancel each other out processing of by the active factor vector table as _T If so, the error signal will approach zero, and finally only aperiodic signals can be reproduced in the periodic noise attenuation signal.

On the other hand, in general, it is possible to remove a periodic signal using only the LMS active algorithm. However, it is necessary to determine to what extent the periodic signal is to be regarded as a periodic signal, and to which degree to cancel the determined periodic signal.

This determination can be adjusted by the system operation unit 190 controlling T (period) and G (weight) applied to the signal block sum processing function, the signal block sum processing function, and the adaptive Fir Filter LMS function.

As the degree of canceling the periodic signal from the audio signal increases, the induced noise causing the afterimage effect increases. So, in order to solve this side effect, the following adaptation level range controller 130 is applied.

The adaptive level range control unit 130 determines a signal provided by the periodic sound analysis and attenuation control unit 120, that is, a signal from which a periodic signal (noise) is attenuated, and determines a level-adapted signal by tone compensation processing and a level scale range. It performs the function of printing.

First, a preprocessed signal (preprocessed signal) is generated by adjusting a preprocessing gain in order to additionally filter a periodicity inducing factor in a signal from which the periodic signal is attenuated.

The pre-processed signal is converted into a level-adapted signal by compensating for a periodicity inducing component by performing tone compensation processing and level adaptive curve control processing for adjusting the level scale range, and correcting the post-processing gain again.

When the voice signal passes through the periodic sound analysis and attenuation control unit 120, a signal attenuated around a periodic component (ie, a periodic noise attenuation signal) is output, and the adaptive level range control unit 130 still remains in the periodic noise attenuation signal. The part of the induced noise is processed.

The adaptive level range control unit 130 includes a gain pre-processing function, a level control curve control function, and a gain post-processing function. The gain pre-processing function and the gain post-processing function are applied to level the signal level. The level adaptive curve control function is applied to process at least a portion of the induced noise portion.

The level adaptive curve control function includes serially connected multi-stage filters (Filter1, Filter2 ... FilterN) for equalization processing and a gain slope control function in order to attenuate the induced noise that was unintentionally added by processing in the periodic sound attenuation control unit. It is composed by

The periodic noise attenuation signal passes through the adaptive level range controller 130 to adjust the volume, control the afterimage noise, and control the level of the frequency band in which the induced noise appears, thereby becoming a level adapted signal. The level adapted signal is processed as follows. (Level-adapted signal) = (pre-processing gain) × [gN FilterN] × GSC × (post-processing gain). Here, [gN FilterN] is for controlling the level of the frequency band in which the induced noise appears, and may be composed of a notch filter, a chef filter, a peak filter, and the like.

In the gain slope control function, the afterimage noise control process is performed by controlling the deformation of the sound wave according to the Attack, Release, Sustain, and Decay Time, which controls the temporal transformation of the sound, and the gain slope according to the height of the volume.

The attenuation adjustment selection unit 140 includes a matched signal output from the sound matching unit 110, a signal from which a periodic signal output from the periodic sound analysis and attenuation control unit 120 is attenuated, and an adaptation level range control unit 130 The level-adapted signal output from is output alone or two or more are weighted and mixed to generate a processed audio signal.

The attenuation control selection unit 140 includes a signal (ie, a matched signal) from which the tone is controlled through the sound matching unit 110 and unnecessary sound sources (ie, feedback oscillation) are removed, and a periodic sound analysis and attenuation control unit 120 A signal in which a periodic component is processed through (i.e., noise attenuated signal) and a signal in which the volume control, afterimage noise control, and induced noise frequency band are controlled through the adaptive level range controller 130 (i.e., level-adapted signal) Each includes input terminals for receiving inputs, selects one or more of each signal, and outputs a finally processed audio signal.

Amplification stages (G1, G2, G3) for gain adjustment for each of the input signals and switches (sw1, sw2, sw3) for selecting a signal input are arranged at each input terminal of the attenuation adjustment selection unit 140. I can. The signals selected by the switches are mixed at the mixing stage and output.

The processed audio signal, which is a result signal selected and mixed by the amplification stages G1, G2, G3 and switches sw1, sw2, sw3, is expressed as follows.

(Processed speech signal) = (G1 × sw1 × (level adapted signal)) + (G2 × sw2 × (periodic noise reduction signal)) + (G3 × sw3 × (matched signal)). At this time, the gain of the amplification stage may be adjusted according to a desired degree of noise attenuation.

Conventional noise attenuation devices fix the amplification gain of a signal and only turn it on/off, so it is difficult to adapt to changes in surrounding environment and to respond to the needs of listeners. However, the noise reduction device according to the present invention selects various signals and attenuates noise using the amplification stages G1, G2, G3 of the attenuation control selection unit 140 and switches sw1, sw2, sw3. Since the degree can be varied, it is possible to effectively respond to the surrounding environment and listening needs.

As an example of the control operation, when the noise reduction degree is turned off, only the matched signal can be output by setting only (processed voice signal) = (G3 × sw3 × (matched signal)). At this output, the tone and feedback oscillations are controlled voice can be heard.

In addition, when the attenuation degree is on, (processed audio signal) = (G1 × sw1 × (level adapted signal)) + (G2 × sw2 × (periodic noise reduction signal)) + (G3 × sw3 × (matching) Signal)), and apply the values of G1 = 0, G2 = α, and G3 = β, and sw1 can be selected from Off, and sw2 and sw3 can be selected from On or Off. Accordingly, the periodic noise attenuation signal and the level-adapted signal are mixed and output after the noise is attenuated. At this output, tone and feedback oscillations are controlled, periodic noise is attenuated, and level adjusted voices can be heard.

On the other hand, if you want to listen to a sentimental voice by accepting the noise of the original sound to some extent, you can listen by partially mixing the signal from which noise has not been removed by setting G1 to a slight weight of 0.1 or less instead of 0. .

The spatial voice position control unit 150 generates spatial sound by arranging the position of the sound source in a virtual space based on the listener by using the noise attenuated voice signal (ie, the processed voice signal). In the case of a voice signal input from one microphone, the listener may recognize that the sound source is located between the listener's eyes. Accordingly, the listener can feel strabismus and dizziness.

The spatial voice position control unit 150 converts the processed voice signal into an analog signal to be input to the left speaker and an analog signal to be input to the right speaker so that the sound source is three-dimensionally arranged in the space surrounding the listener. do. In this way, the listener can hear the voice stably and comfortably.

That is, the spatial voice position control unit 150 spatially controls the processed voice signal output through the attenuation control selection unit 140, so that the spatial sense recognized from the voice signal and the visual sense recognized by the spatial sense It is possible to eliminate the possible disparity, that is, the phenomenon that the sound image and the visual overlap. As a result, visual fatigue of the listener is reduced, and pain and dizziness in the ears and head do not occur even when the audio signal is listened to for a long time.

In this way, the spatial voice position control unit 150, which performs a function of converting a voice signal inputted by only one microphone into a stereoscopic sound in the space, processes the signal using the principle of the HRTF (Head Related Transfer Function). Is done. To this end, the spatial voice position control unit 150 may include a left delay function, a left amplification function, a left phase processing function, a right delay function, a right amplification function, a right phase processing function, and a D/A conversion function. .

In general, the HRTF controls the volume of the direct sound and the indirect sound at the same time. In this embodiment, the indirect sound is additionally processed. In addition, while the general HRTF only provides a function of adjusting only the balance of the spatially transmitted flat sound transmitted to the ear, the HRTF in this embodiment is capable of controlling the spatial position of the spatial transmission as it operates in the spatial sound position control unit 150. You can adjust the position of the sound source you are listening to in three dimensions.

The processing relationship is (left analog signal) = (processed audio signal) + (processed audio signal) × ((left delay amount) + (left amplification amount) + (left phase throughput)), and (right analog signal) = (Processed voice signal) + (Processed voice signal) × ((right delay amount) + (right amplification amount) + (right phase throughput)).

The signal processed by the HRTF is converted into an analog signal while passing through the D/A converter 155.

The main gain amplifying unit 170 controls the volume of an analog signal to be input to the speaker automatically or according to a request of a listener under the control of the system operation unit 190.

The signal analysis unit 160 generates a spectrum signal by analyzing a frequency component of a noise attenuated speech signal (ie, a processed speech signal). Based on this spectral signal, it is possible to search for a feedback oscillation sound included in the processed speech signal.

That is, the signal analysis unit 160 receives and processes a voice signal (that is, a processed voice signal) in which the degree of attenuation for noise is multivariably processed according to the environment or listening request through the attenuation control selector 140. The resulting voice signal is analyzed by frequency and spectrum information is output.

Using the spectrum information analyzed by the signal analysis unit 160, the system operation unit 190 may find a signal source from which the feedback oscillation sound is generated. The spectral information consists of (spectral signal) = [gN FilterN] _T × (processed speech signal). [gN FilterN] _T is arranged in a parallel structure. Each frequency component is analyzed and converted into data to output spectrum information.

The system operation unit 190 may control the operation of each of the above-described units to process a noise reduction operation according to a preset program. That is, processing such as selecting a signal to be mixed in each unit and adjusting an amplification gain (or weight) for the selected signal is performed.

In particular, the system operation unit 190 identifies the frequency at which the feedback oscillation sound appears through the spectrum information provided by the signal analysis unit 160, and issues a feedback oscillation sound control command to control the feedback oscillation sound of the corresponding frequency. It is provided to the matching unit 110. When a control command is received, the sound matching unit 110 performs a process related to the feedback oscillation by using the tone and feedback oscillation control function.

In addition, the system operation unit 190 may receive an operation related to noise reduction from a listener. For example, the attenuation control selector 190 may select which signal to mix or an operation for adjusting a gain for the selected signal may be applied by receiving input from the listener.

Meanwhile, the system operation unit 190 controls the signal selection unit 195 to stop the output of the processed audio signal when the user receives the selection of the hearing management function and the selection of the reference frequency, and the hearing management unit ( 200) can be output from the hearing management sound.

The signal selection unit 195 is mainly disposed in front of the main gain amplification unit 170, and one of the audio management sound from the auditory management unit 200 and the signal from another component (for example, noise reduction Voice signal). The operation of the signal selection unit 195 may be controlled by the system operation unit 190 or may be manually controlled by a user.

The noise reduction apparatus according to an embodiment of the present invention according to this configuration selects only components requiring noise attenuation from the mixed signal introduced into only one microphone, and various output signals generated using the selected signal (matching One or more of the reduced signal, periodic noise attenuation signal, and level-adapted signal) can be weighted and mixed to provide an emotional sound to the listener. In addition, since it can be converted into a stereoscopic sound in a virtual space, even if the listener listens to the sound for a long time, it is not uncomfortable.

Claims (16)

When a single reference frequency for auditory management specialized for the subject is set, a single reference sound related to the single reference frequency is determined, and a plurality of ambient sounds are determined within a range of 1 octave including the single reference sound. A sound range determining unit-the frequencies of the plurality of ambient sounds are calculated by the formula of {(reference frequency)×2^(M/L)}, where M represents the M-th sound from the reference frequency, L denotes the number of notes to divide one octave from the reference frequency;
An output time determination unit determining an output time for each sound for outputting the single reference sound and each of the plurality of ambient sounds;
A sound time matching unit for generating sound output control information by matching the output time of each sound to each of the single reference sound and the plurality of ambient sounds; And
And an output sound generator for generating an auditory management sound to be output to a speaker based on the sound output control information. delete delete delete The method of claim 1,
The sound range determination unit,
The audible sound frequency range is divided into a plurality of bands in which the range is partially overlapped with the adjacent one while being a predetermined unit of one octave,
When a specific band including the single reference frequency is specified, all or some scales included in the specified band are selected as the plurality of ambient sounds. The method of claim 1,
The output time determination unit,
A sweep output time for each sound to output each sound in a sweep method of sequentially outputting the reference sound and the plurality of ambient sounds according to a frequency order,
In a random method of outputting the reference sound and the plurality of ambient sounds in a non-sequential manner not following a frequency order, a random output time for each sound to be outputted is determined. The method of claim 6,
The sound sweep output time is,
Hearing management, characterized in that, in the sweep method, it is determined to be more than a minimum time for recognizing the existence of each output sound or less than a maximum time for not recognizing that each output sound is cut off. Device. The method of claim 1 or 6,
The auditory management sound is output according to a randomly mixed method of a sweep method in which the single reference sound and the plurality of ambient sounds are sequentially output according to a frequency order and a random method that is output in a non-sequential order without following a frequency order. , Characterized in that configured, hearing management device. When a single reference frequency for auditory management specialized for the subject is set, determining a single reference sound related to the single reference frequency, and determining a plurality of ambient sounds within a predetermined range from the single reference sound-the The frequencies of the plurality of ambient sounds are calculated by the formula {(reference frequency)×2^(M/L)}, where M represents the M-th sound from the reference frequency, and L is from the reference frequency Indicates the number of negatives to distinguish 1 octave of -;
Determining an output time for each sound for outputting the single reference sound and each of the plurality of ambient sounds;
Generating sound output control information by matching the output time for each sound to each of the single reference sound and the plurality of ambient sounds; And
Generating an auditory management sound to be output through a speaker based on the sound output control information and applying it to the subject. delete The method of claim 9,
The step of determining the output time for each sound,
Determining a sweep output time for each sound to output each sound in a sweep method in which the reference sound and the plurality of ambient sounds are sequentially output according to a frequency order,
The hearing management method further comprising determining a random output time for each sound to output each sound in a random manner in which the reference sound and the plurality of ambient sounds are output in a non-sequential manner not following a frequency order. The method of claim 9 or 11,
Generating the auditory management sound,
The auditory management sound is generated so that the single reference sound and the plurality of ambient sounds are output according to a randomly mixed method of a sweep method in which the single reference sound and the plurality of ambient sounds are sequentially output in a frequency order and a random method that is output in a non-sequential order without following the frequency order. Characterized in that, hearing management method. delete delete delete delete

Images