KR102175163B1 - Method and apparatus of noise reduction and audio playing apparatus with non-magnetic speaker - Google Patents

Abstract

An input unit for receiving a speech signal including noise, a period estimating unit for estimating a period of a noise pattern in the speech signal, and subtracting and removing the noise pattern from the speech signal in a frequency domain using the estimated period of the noise pattern Disclosed is a sound device including a noise removal unit, a noise update unit for updating the noise pattern according to a change in the amount of the noise, and an output unit for outputting the voice signal from which the noise pattern is removed.

Description

A noise reduction method and apparatus, and a sound reproduction apparatus using a non-magnetic speaker TECHNICAL FIELD

The present disclosure relates to a method and apparatus for voice communication enabling smooth communication without being affected by noise generated by repeating patterns having a constant period. In addition, the present disclosure provides entertainment, communication, and patient diagnosis services using a display device, a headset, or a speaker device during magnetic resonance imaging (MRI). The present invention relates to a voice communication method that enables medical staff and patients to communicate smoothly without being affected by noise generated by MRI, and to a sound reproducing apparatus for reproducing multimedia to a patient during photographing.

The present disclosure relates to a sound reproducing apparatus for transmitting sound to a patient in an MRI noisy environment, the structure of a non-magnet speaker that works well in all positions inside or outside the MRI bore and an algorithm for operating it The present invention relates to a system capable of reproducing sound signals such as voice and audio and actively attenuating the noise of MRI.

As one of the noise reduction methods for voice communication, it is a noise reduction method using the beamforming method. It receives an input mixed with noise and voice signals from multiple microphones, and uses the MKDR (Maximum-Kurtosis Distortionless Response) algorithm and MVDR (Minimum Variance). There is a method of restoring the speech signal by using the Distortionless Response) algorithm.

The MRI environment is an environment in which a strong magnetic field exists, and materials containing general magnetic substances cannot be used. Therefore, a general dynamic speaker cannot be used in a strong magnetic field environment of MRI. Conventionally, a pneumatic speaker or a piezo-electro speaker has been used to transmit acoustic signals to patients in such a strong magnetic field environment. Speakers of this type do not use magnetic materials, so they can operate in a strong magnetic field environment without affecting the MRI image. However, these speakers have a bad overall characteristic because their ability to reproduce low frequencies is limited and their output is limited.

Through some embodiments, an overall system and method required for voice communication between a speaker and a counterpart in an environment with a high noise level having a periodic pattern are presented. In addition, according to some embodiments, a method for removing noise required for voice communication between patients and medical staff in a high noise level MRI environment having a periodic pattern, and various intensities distributed inside/outside the MRI bore to provide additional services to the patient during imaging. We present a sound reproducing apparatus having a structure that can operate both in the magnetic field environment of To this end, a sound reproducing apparatus using a magnetic-free dynamic speaker operating using an MRI magnetic field, a control method for maintaining a constant acoustic performance by adapting to a change in magnetic field strength, and a sound reproducing system including a control device are proposed. And through this, the MRI sound reproducing system is constructed that operates anywhere without being affected by the change of the static magnetic field according to the position of the headset.

The acoustic device according to some embodiments includes an input unit receiving a speech signal including noise, a period estimating unit estimating a period of a noise pattern in the speech signal, and the speech in a frequency domain using the estimated period of the noise pattern. A noise canceling unit that subtracts and removes the noise pattern from a signal, a noise update unit that updates the noise pattern according to a change in the size of the noise, and an output unit that outputs the voice signal from which the noise pattern has been removed. .

The period estimating unit may obtain period information of the noise from the device generating the noise, or may calculate period information through data of the voice signal acquired for a predetermined time.

The noise canceling unit includes an alignment unit that matches the noise pattern and the timing index of the current frame of the voice signal, removes the spectrum of the noise frame with timing from the current frame of the voice signal, and removes residual noise through post-processing. It may include a calculation unit to allow.

The noise update unit determines the presence or absence of speech in the current frame of the speech signal, updates the noise pattern when there is no speech in the current frame of the speech signal, and updates the noise pattern from the speech signal to the noise reduction unit. Request to remove the noise pattern, determine whether or not the output from which the noise pattern is removed is amplified and emit larger than the input noise, and if the output is emanating, the noise pattern information is initialized to the period estimation unit. Can be asked to do.

The noise reduction method according to some embodiments includes receiving a voice signal including noise, estimating a period of a noise pattern in the voice signal, and using the estimated period of the noise pattern in the frequency domain. And subtracting and removing the noise pattern from, updating the noise pattern according to a change in the size of the noise, and outputting the voice signal from which the noise pattern has been removed.

Estimating the period of the noise pattern may include obtaining period information of the noise from the device generating the noise, or calculating period information through data of the voice signal acquired for a predetermined time. .

The subtracting and removing the noise pattern may include matching the noise pattern with the timing index of the current frame of the voice signal, removing the spectrum of the timing noise frame from the current frame of the voice signal, and remaining It may include removing the noise through post-processing.

The updating of the noise pattern may include determining the presence or absence of voice in the current frame of the voice signal, and if voice does not exist in the current frame of the voice signal, updating the noise pattern and from the voice signal Removing the noise pattern, determining whether the output from which the noise pattern has been removed is emitted, and when the output is emitted, initializing noise pattern information.

The sound reproducing apparatus according to some embodiments includes an audio output unit that outputs an acoustic signal by mounting a non-magnetic speaker, and selects a length of a coil or an intensity of a current according to a change in output of the non-magnetic speaker due to a change in the intensity of a static magnetic field. It may include a selector, and a reproducing unit for reproducing the sound signal using the length of the selected coil or the strength of the selected current.

The sound output unit includes a magnet-free dynamic transducer including a multi-coil having a plurality of lengths, and an analysis sensor for measuring a change in output of the magnet-free speaker, and the selection unit includes: At least one of the multi-coils may be selected, and the reproducing unit may reproduce the sound signal according to characteristics of the non-magnetic dynamic transducer using the selected coil.

The sound output unit includes a magnetic-free dynamic transducer including a current intensity adjustment unit for adjusting the intensity of the current flowing through the voice coil, and an analysis sensor for measuring a change in output of the non-magnetic speaker, and the selection unit, the The current intensity adjustment unit of the magnetic-free dynamic transducer may select a current intensity, and the reproducing unit may reproduce the acoustic signal according to characteristics of the non-magnetic dynamic transducer using the selected current intensity.

The multi-coil of the magnetic-free dynamic transducer of the sound output unit may include a plurality of annular coils having different diameters, and the plurality of annular coils may be connected to one diaphragm.

The multi-coil of the magnetic-free dynamic transducer of the sound output unit divides the coil connected to the diaphragm into a plurality of bundles, and the number of bundles of the coils operated according to the selection of the selection unit may vary.

The current intensity adjustment unit of the magnetic-free dynamic transducer of the sound output unit includes a variable resistor or resistance size selector and a gain controller, and adjusts the intensity of the current selected by the selection unit through selection of a resistance size. I can.

The active noise control method according to some embodiments includes selecting an output device according to the strength of a static magnetic field according to a position of the output device, adjusting a gain used in the selected output device, and determining the selected output device and the gain. Reflecting and calculating an active noise control signal, and outputting the calculated active noise control signal to the selected output device may include performing noise control.

The selecting of the output device may include sequentially operating the shortest coil from the multi-coil included in the output device, and analyzing the output signal obtained through an analysis sensor for measuring the response of the output device. Steps, determining whether distortion occurs in which the waveform of the output signal is transformed from the waveform of the input signal input to the output device, and whether the output can be reproduced louder than noise, and the output is output without distortion. It may include the step of selecting the shortest coil among the reproducible coils.

The selecting of the output device includes sequentially operating from the largest resistance in the current intensity adjustment unit included in the output device, and analyzing the output signal obtained through an analysis sensor for measuring the response of the output device. The step of, determining whether distortion is generated in which the waveform of the output signal is transformed from the waveform of the input signal input to the output device, and whether the output can be reproduced louder than noise, and output without distortion It may include the step of selecting the largest resistance value among the reproducible resistance values.

The step of adjusting the gain may include calculating a gain value based on an output of a non-magnetic dynamic transducer using a selected coil among multi-coils included in the output device and level information of an input noise. .

The step of adjusting the gain may include calculating a gain value based on an output based on a selected resistance value among resistance values of a current intensity adjusting unit included in the output device and level information of an input noise.

The calculating of the active noise control signal may include output characteristic information between an analysis sensor for measuring a response of the output device and the selected output device, and noise information input through the analysis sensor, the analysis sensor It may include the step of configuring a filter for generating the active noise control signal to reduce the amount of noise input through.

In the calculating of the active noise control signal, the selected sound is based on output characteristic information between the analysis sensor for measuring the response of the output device and the selected output device, and noise information input through the analysis sensor. It may include calculating the active noise control signal using a fixed filter previously input to the output device.

The non-magnetic speaker according to some embodiments includes a voice coil, a vibration plate connected to the voice coil, and is attached to one side of the vibration plate between the center of the vibration plate and the edge of the vibration plate, and absorbs vibration for absorbing the vibration of the vibration plate. And a stopper positioned at a front or rear side of the speaker and provided at a distance to support the vibration absorbing unit attached to the diaphragm.

The vibration absorbing unit may be a damper made of a compressible material.

The magnetic resonance imaging (MRI) system according to some embodiments is a first voice input device receiving an audio signal including a bore generating a magnetic field, a voice of a first user on the bore side, and controlling the MRI A second voice input device that receives an audio signal including the voice of the second user, and a first output device that outputs the sound signal including the voice of the second user to the first user using a non-magnetic speaker, A second output device that outputs an acoustic signal including the voice of the first user to the second user, and according to the strength of the MRI static magnetic field according to the position of the first output device, the sound signal of the second voice input device An active noise control device that performs active noise control and transmits it to the first output device, and removes the noise pattern from the frequency domain by estimating a noise pattern from the sound signal of the first voice input device and delivers it to the second output device. It may include a noise cancellation signal processing device.

The first voice input device may include a directional microphone.

1A is a diagram illustrating a method of processing noise in a sample unit in a time domain according to some embodiments.
1B is a diagram showing a result of a noise reduction simulation using a method of processing noise in units of samples in a time domain.
2A is a diagram illustrating a method of processing noise in a buffer (or frame) unit in a frequency domain according to some embodiments.
2B is a diagram illustrating a result of a noise removal simulation through a method of processing noise in a frequency domain in units of buffers (or frames) according to some embodiments.
3 is a diagram showing a change in MRI static magnetic field strength according to a location.
4A to 4H are diagrams showing results of an active noise reduction performance test for a non-magnetic speaker for each location.
5 is a schematic block diagram of an acoustic device according to some embodiments.
6 is a diagram illustrating a signal processing method for removing periodic noise according to some embodiments.
7 is a schematic block diagram of a sound reproducing apparatus according to some embodiments.
8 is a detailed block diagram of a sound reproducing apparatus according to some embodiments.
9 is a diagram showing a direction of an MRI magnetic field and a structure of a non-magnetic dynamic speaker.
10 is a diagram showing an embodiment of a multi-coil method of a magnetic-free dynamic speaker.
11 is a diagram showing another embodiment of a multi-coil method of a magnetic-free dynamic speaker.
12 is a view showing an embodiment of the current intensity adjustment unit of the non-magnetic dynamic speaker shown in FIG.
13 is a flowchart illustrating an operation according to an embodiment of the static magnetic field strength analyzer illustrated in FIG. 8.
14 is a diagram illustrating a structure of a damper and a stopper of a magnetic-free speaker according to some embodiments.
15 is a flow chart of a noise reduction method according to some embodiments.
16 is a schematic flowchart illustrating a voice communication noise reduction algorithm according to some embodiments.
17 is a detailed flowchart illustrating a voice communication noise reduction algorithm according to some embodiments.
18 is a flowchart illustrating an active noise control method according to some embodiments.
19 is a diagram illustrating an active noise reduction algorithm according to some embodiments.
20 illustrates a magnetic resonance imaging (MRI) system according to some embodiments.
21 shows a simulation result of blocking noise by applying an algorithm to MRI noise.

Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described later together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in a variety of different forms, and only these embodiments make the disclosure of the present invention complete, and are common knowledge in the technical field to which the present invention pertains. It is provided to completely inform the scope of the invention to those who have, and the invention is only defined by the scope of the claims.

The terms used in the present specification will be briefly described, and the present invention will be described in detail.

The terms used in the present invention have been selected from general terms that are currently widely used while considering functions in the present invention, but this may vary depending on the intention or precedent of a technician working in the field, the emergence of new technologies, and the like. In addition, in certain cases, there are terms arbitrarily selected by the applicant, and in this case, the meaning of the terms will be described in detail in the description of the corresponding invention. Therefore, the terms used in the present invention should be defined based on the meaning of the term and the overall contents of the present invention, not a simple name of the term.

When a part of the specification is said to "include" a certain component, it means that other components may be further included rather than excluding other components unless otherwise stated. In addition, the term "unit" used in the specification refers to a hardware component such as software, FPGA, or ASIC, and the "unit" performs certain roles. However, "unit" is not meant to be limited to software or hardware. The “unit” may be configured to be in an addressable storage medium or may be configured to reproduce one or more processors. Thus, as an example, "unit" refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, procedures, Includes subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, database, data structures, tables, arrays and variables. The functions provided within the components and "units" may be combined into a smaller number of components and "units" or may be further separated into additional components and "units".

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present invention. In addition, in the drawings, parts not related to the description are omitted in order to clearly describe the present invention.

Among the imaging equipment used in medicine-related fields, MRI is widely used because it has little effect on the human body, but medical staff and patients suffer from discomfort due to loud noise and long scan time. During the MRI operation, the patient experiences fear with loud noise in a narrow bore, and the medical staff listens to the voice along with the MRI noise to monitor the patient's condition, causing stress.

Earplugs or earmuffs are worn to block the noise audible to the patient in the scan room.In this way, the voice spoken by the medical staff cannot be heard in the control room, and the noise that the medical staff hear is reduced. can not do. In addition, even in a noisy environment such as helicopter or jet noise or factory noise, communication is experiencing great inconvenience.

Some embodiments propose a method of enabling two-way voice communication while attenuating periodic noise by using a noise reduction technology and an active noise control device. The noise reduction technology is applied to deliver a voice signal that blocks noise, and an active noise control device is used to reduce the noise audible to the speaker, while listening to the other party's voice. Through this, the situation of the other party can be continuously monitored, and the speaker can also communicate such as asking about the current progress.

In addition, as advanced imaging equipment such as CT, MRI, and ultrasound develops, the necessity and importance of imaging medicine are increasing day by day. Among these advanced imaging equipment, MRI provides more information to doctors and patients by showing detailed human body structure more clearly than other equipment, and is being used more and more because it is harmless to the human body. In addition, with the advancement of technology, MRI is being developed to obtain more precise images in a shorter time by using a stronger magnetic field. However, as the strength of the magnetic field used increases, the noise of the MRI is getting louder, and the noise level exceeding 100dB causes great discomfort to the patient, and furthermore, damage to the auditory organs is concerned due to the noise that continues for a long time. It is reality. Accordingly, there is a need for a device or method capable of protecting an auditory organ in such a noisy MRI environment, or providing external voice information or music to a subject.

In some embodiments, a method for enabling two-way voice communication and other services while attenuating MRI noise that patients and medical staff hear for a long time by using a noise canceling technology, an active noise control headset or speaker device, and a display device, and Using the structure of a non-magnet speaker that works well in all locations inside and outside the MRI bore and the algorithm to operate it, it reproduces sound signals such as voice and audio, and actively attenuates MRI noise. Present a device that can be used. The noise canceling technology is applied to transmit the voice signal of the patient who has blocked MRI noise to the medical staff in the control room, and the patient wearing the active noise control headset hears the medical staff's voice through the headset. This allows the medical staff to continuously monitor the patient's condition, and the patient can also communicate with the medical staff, such as asking about the current progress. In addition, using this method and apparatus, a service that provides multimedia contents to a patient during shooting, a doctor remotely diagnoses a patient, or provides various assistance information tailored to the patient is possible.

As a noise reduction and communication method in an MRI environment, there is a method of inputting a patient's voice signal mixed with micro MRI noise in the scan room, subtracting the noise component from the input signal by sample unit and outputting the voice signal to the control room. In addition, there is a technology on the method and device for transmitting the patient's voice to the medical staff and hearing the external audio signal to the patient using an input/output device configured in the form of earplugs. When transmitting audio signals, input from the external microphone of the earplugs Noise is eliminated by Active Noise Control (ANC) method using the generated signal.

In some embodiments, as a speaker for sound reproduction in an MRI environment, a non-magnet dynamic speaker using a strong magnetic field generated in MRI is used. Magnetic-free dynamic speaker is a speaker of a structure that does not use magnets, and generates sound by moving the diaphragm by Lorentz force generated between the static magnetic field existing in the MRI environment and the current flowing through the voice coil. A speaker with this structure can effectively reproduce a wide range of sound by using a strong static magnetic field existing in MRI even if there is no permanent magnet existing in a general speaker.

Regarding other services that can be provided during MRI, a display device/head-mounted audio playback device is used to provide entertainment services for the patient's stability during imaging, or while the patient and the doctor communicate. There is a way to provide feedback such as patient motion information.

In the case of a negative SNR (Signal to Noise Ratio) because the noise level is high and the distance between the speaker's position and the voice acquisition microphone is large, as in the MRI environment, there is a disadvantage that voice transmission is difficult due to the large amount of noise compared to the voice. The effect of noise reduction is not great. In addition, since the technology for determining the presence or absence of a voice signal in such an environment is inaccurate, there is a limitation in using the method of updating the noise component to be removed.

1A is a diagram illustrating a method of processing noise in a sample unit in a time domain according to some embodiments. 1B is a diagram showing a result of a noise reduction simulation using a method of processing noise in units of samples in a time domain. In Fig. 1B, the upper waveform represents the input, and the lower waveform represents the output.

In step 110 of FIG. 1A, period information of an input signal is input.

In step 120, noise from the input signal is stored as a periodic noise pattern.

In step 130, samples of the noise pattern are removed in units of samples in the time domain.

In step 140, it is determined whether or not there is a voice. If it is determined that voice is present, the process returns to step 130 for processing the next sample.

When it is determined in step 140 that there is no voice, in step 150, since it can be estimated that only noise is present, the noise pattern is updated. Return to step 130 for processing of the next sample.

If a periodic noise pattern is stored from the input signal and processed in a sample unit in the time domain as shown in FIG. 1A, the noise reduction effect may not be great or the noise may be amplified as the result of FIG. 1B for portions in which the noise pattern is not completely identical. There is no judgment as to whether or not the output signal is emitted for such a situation. That the output signal is'dispersed' indicates a case in which the output is greatly amplified compared to the input as shown in FIG. 1B.

2A is a diagram illustrating a method of processing noise in a buffer (or frame) unit in a frequency domain according to some embodiments. FIG. 2B is a diagram illustrating a result of a noise reduction simulation through a method of processing noise in a frequency domain in units of buffers (or frames) according to some embodiments. In FIG. 2B, the upper waveform represents the input, and the lower waveform represents the output.

In step 210 of FIG. 2A, the noise pattern is stored in the buffer.

In step 220, a period of noise is estimated from the noise pattern. Alternatively, a period of noise may be input from the outside.

In step 230, FFT (Fast Fourier Transform) is performed on the noise pattern and the input buffer to switch to the frequency domain.

In step 240, the size of the noise pattern is subtracted from the input buffer in units of buffers. Further, residual noise can be further removed from the subtracted result.

In step 250, IFFT (Inverse Fast Fourier Transform) is performed on the noise-removed result to convert to the time domain.

In step 260, it is determined whether or not there is a voice. If it is determined that voice is present, the process returns to step 230 for processing of the next buffer.

If it is determined in step 260 that there is no voice, in step 270, since it can be estimated that only noise is present, the noise pattern is updated.

In step 280, it is determined whether the updated noise pattern is emitted. When the updated noise pattern is emitted, when noise is removed using this noise pattern, the output can be significantly amplified, that is, emitted, compared to the input. Thus, in order to again estimate the period of the noise pattern, it returns to step 210.

If the updated noise pattern does not emit, the process returns to step 230 so that the next buffer can be processed using the updated noise pattern.

As shown in FIG. 2A, according to the method of processing noise in the frequency domain in units of buffers (or frames), as shown in FIG. 2B, noise is removed from the input and an output that is not emitted can be obtained.

3 is a diagram showing a change in MRI static magnetic field strength according to a location. In FIG. 3, the horizontal axis (Z) is a direction from the center of the bore toward the entrance of the bore, for example, when the patient is lying in the bore of the MRI and the patient's head is in the center of the bore, it may be the direction toward the patient's leg. have. In FIG. 3, the vertical axes R and Y may be in a radial direction from the center of the bore and may be a direction perpendicular to the ground surface. In FIG. 3, the direction penetrating the ground (X-axis direction) may be a radial direction from the center of the bore and a direction horizontal to the ground surface. In FIG. 3, the numerical value of each line of magnetic force represents Gauss. The stray field line represents 0.05→0.1→0.2→0.5→1→2→5→10→20→50→100→200(mT) from the outside.

In the case of a non-magnetic dynamic speaker, since the static magnetic field formed inside the bore of the MRI is used, the strength of the static magnetic field has an important influence on the output of the speaker. However, as shown in Figure 3, outside the bore of the MRI, the strength of the static magnetic field decreases sharply as the distance from the center of the MRI decreases, and in the case of 3T (Tesla) MRI, the strength of the magnetic field is at a distance of about 1.8m from the center of the MRI. In comparison, it is reduced to about 1/60 of 50mT.

4A to 4H are diagrams showing results of an active noise reduction performance test for a non-magnetic speaker for each location. Figure 4a shows the location of the bore center (0 cm from the bore center). 4B shows a location of 10 cm (60 cm from the center of the bore) into the bore from the bore end. Figures 4c, 4d, 4e, 4f, 4g, and 4h are, respectively, 20cm, 50cm, 60cm, 70cm, 80cm, and 90cm from the bore end to the bore outward (90cm, 120cm, 130cm, 140cm, 150cm, and 160cm). In FIGS. 4A to 4H,'ANC off' (a) indicates MRI noise when active noise reduction is not performed, and'ANC on' (b) indicates reduced noise when active noise reduction is performed.

In FIGS. 4A to 4H,'ANC on' (b) shows a result of more noise reduction than'ANC off' (a). However, as the distance from the center of the bore gradually increases, the difference between the outputs of'ANC off' (a) and'ANC on' (b) gradually decreases, and in FIGS. 4E to 4H,'ANC on' (b ) Shows the result that the noise of'ANC off' (a) gradually increases.

Outside the bore of the MRI, the speaker does not produce sufficient output due to the weak magnetic field strength. Due to this problem, as shown in Figs. 4E to 4H, when the head is bore outside, such as when photographing a knee or foot, the speaker does not output an output corresponding to the size of the MRI noise. distortion), and eventually, due to distortion of the speaker, normal sound cannot be reproduced or active noise control performance is reduced. 4E to 4H, it can be seen that, as the distance from the bore of the MRI increases, the noise of the signal (b) with reduced MRI noise is greater than the signal (a) including the MRI noise.

Multimedia provided by the patient during shooting is a service provided by a passive method that the patient cannot select, and the content is also limited to the range previously entered into the entertainment system.

5 is a schematic block diagram of an acoustic device 500 according to some embodiments.

Referring to FIG. 5, the acoustic device 500 may include an input unit 510, a period estimation unit 520, a noise canceling unit 530, a noise update unit 540, and an output unit 550.

The input unit 510 may receive an audio signal including noise. The input unit 510 may include a microphone, and may receive a voice signal recorded in a storage medium or a voice signal transmitted through a network.

The period estimating unit 520 estimates the period of the noise pattern in the voice signal. The period estimating unit 520 may obtain period information of a noise from a device that generates noise, or may calculate period information through data of a voice signal acquired for a predetermined time.

The noise canceling unit 530 subtracts and removes the noise pattern from the voice signal in the frequency domain using the estimated period of the noise pattern. The noise canceling unit 530 is an alignment unit that matches the noise pattern and the timing index of the current frame of the voice signal, removes the spectrum of the noise frame with timing from the current frame of the voice signal, and removes residual noise through post-processing. It may include a calculation unit to allow.

The noise update unit 540 updates the noise pattern according to the change in the noise level. The noise update unit 540 determines whether there is a voice in the current frame of the voice signal. The noise update unit 540 updates the noise pattern when there is no voice in the current frame of the voice signal. The noise update unit 540 requests the noise canceling unit 530 to remove the noise pattern from the voice signal.

The noise update unit 540 determines whether the output from which the noise pattern is removed is amplified and emitted larger than the input noise, and when the output is emitted, the period estimating unit 520 requests the period estimating unit 520 to initialize the noise pattern information. I can. That the output signal is'dispersed' indicates a case in which the output is greatly amplified compared to the input as shown in FIG. 1B.

The output unit 550 may output a voice signal from which the noise pattern has been removed. The output unit 550 may include a speaker that outputs an audio signal as sound, may record an audio signal in a storage medium, or may transmit an audio signal through a network.

In some embodiments, a speaker (e.g., a patient) is a voice input device composed of one or more microphones that acquires the speaker's (e.g., patient) voice in order to communicate with the other party, while reproducing a sound source signal to the speaker (e. It consists of a sound reproducing device including an offset ANC function, and a noise reduction system that removes noise from the input voice signal. In addition, in order for the other party to communicate with a speaker (eg, a patient), a voice input device configured with one or more microphones for acquiring a voice of the other party and a speaker or headphone device for reproducing a sound transmitted to the other party may be provided.

-Noise reduction for voice communication

A signal input from a speaker to an input device passes through a noise reduction system, and a voice signal from which noise is removed is transmitted, and the transmitted signal is reproduced to the other party through a speaker or headset device. In addition, the signal input from the other party to the input device is also transmitted and output to the ANC device or speaker worn or mounted by the speaker.

6 is a diagram illustrating a signal processing method for removing periodic noise according to some embodiments. The noise reduction system for removing the noise signal from the voice signal input to the input device is a periodic noise signal by analyzing the noise period information input from the device generating periodic noise and the noise signal inputted into the microphone in real time as shown in FIG. It is a signal processing device that extracts features and transfers only voice signals by removing only noise signals from input signals using the extracted feature information.

When using a method of subtracting the noise previously stored in the time domain from the current input signal as shown in FIG. 1A, when the noise pattern corresponding to the current sample value in the time domain does not match, the noise signal is reduced. There is a problem that cannot be done or can be amplified.

In some embodiments, periodic characteristic information of a noise signal analyzed in advance or in real time as shown in FIG. 2A is obtained. Also, by classifying a frame in a form suitable for speech signal processing, frequency-processed noise signal data of a frame suitable for each period of the input noise signal is obtained. In addition, a method of subtracting a spectrum of a signal corresponding to a current frame is used by using periodic characteristic information and noise signal data. Therefore, the algorithm works robustly without amplification even in situations where the patterns do not match exactly. In order to process the stored noise signal data according to the time information of the current frame, the period characteristics are determined by finding period information of periodic noise using noise patterns or processing noise period information input from the device in the time domain or frequency domain. To grasp.

7 is a schematic block diagram of a sound reproducing apparatus 700 according to some embodiments.

Referring to FIG. 7, the sound reproducing apparatus 700 includes a sound output unit 710, a selection unit 720, and a reproduction unit 730.

The sound output unit 710 is equipped with a non-magnetic speaker and outputs an sound signal. The sound output unit 710 includes a non-magnetic dynamic transducer including a multi-coil having a plurality of lengths, for example, a multi-voice coil, and an analysis sensor for measuring an output change, a speaker response, and a noise level of the non-magnetic speaker, For example, it may include a microphone. The multi-coil of the magnetic-free dynamic transducer of the sound output unit 710 includes a plurality of annular coils having different diameters, and a plurality of annular coils may be connected to one diaphragm. The multi-coil of the magnetic-free dynamic transducer of the sound output unit 710 divides the coil connected to the diaphragm into a plurality of bundles and operates. The number of bundles of coils operated here may vary depending on the selection of the selection unit 720.

In addition, the sound output unit 710 may include a non-magnetic dynamic transducer including a current intensity adjustment unit that adjusts the intensity of the current flowing through the voice coil. The current intensity adjusting unit of the non-magnetic dynamic transducer of the sound output unit 710 includes a variable resistor or resistance size selector and a gain controller, and the current selected by the selection unit 720 is You can adjust the intensity.

The selector 720 selects the length of the coil or the strength of the current according to the change in the output of the non-magnetic speaker due to the change in the strength of the static magnetic field. The selection unit 720 may select one or more coils from among multiple coils of the non-magnetic dynamic transducer by the coil length adjustment unit. In addition, the selector 720 may select the current intensity of the current intensity adjustment unit of the non-magnetic dynamic transducer.

The reproducing unit 730 reproduces the sound signal by using the selected coil length or the selected current strength. The reproducing unit 730 may reproduce an acoustic signal according to characteristics of a non-magnetic dynamic transducer using the selected coil. In addition, the reproducing unit 730 may reproduce an acoustic signal according to characteristics of the non-magnetic dynamic transducer using the selected current intensity.

8 is a detailed block diagram of a sound reproducing apparatus according to some embodiments.

The sound reproducing apparatus 800 using a speaker in an MRI environment according to some embodiments includes: 1) a magnet-free speaker including a multi-voice coil 810, 2) a coil length adjusting unit 830, 3) a current intensity It consists of an adjustment unit 840, 4) a static magnetic field strength analysis unit 850.

The magnet-free speaker includes a multi-voice coil 810. In addition, an analysis sensor 820 for measuring speaker response and noise level, for example, may include a microphone. The multi-voice coil 810 may include a plurality of annular coils, and a plurality of annular coils may be connected to one diaphragm 860.

The coil length adjustment unit 830 may adjust the total length of the used coil by inputting an audio signal to be output to a selected coil among each of the coils constituting the multi-layer voice coil.

The current intensity adjustment unit 840 may adjust the current intensity through selection of a resistance size.

The static magnetic field strength analyzer 850 is a part that determines the strength of the current or the length of the coil according to the strength of the magnetic field.

Hereinafter, each configuration of FIG. 8 will be described with reference to FIGS. 9 to 14.

1) Magnetic-free speaker including a multi-voice coil 810

9 is a diagram showing a direction of an MRI magnetic field and a structure of a non-magnetic dynamic speaker. Referring to FIG. 9, the non-magnetic speaker includes a vibration plate 920 for generating sound and a voice coil 910 for moving the vibration plate. In particular, the voice coil 910 may be formed in the form of a multi-layered coil so that various coil lengths can be implemented. In addition, the non-magnetic speaker may include a speaker frame 930 and a headset case 940 for supporting and protecting an operating part of the speaker, use of the speaker, and external appearance of the speaker.

10 is a diagram showing an embodiment of a multi-coil method of a magnetic-free dynamic speaker.

When the voice coil 910 of FIG. 9 is implemented in the form of a multi-layer coil, as shown in FIG. 10, an annular-shaped voice coil having different diameters and number of turns may be used.

11 is a diagram showing another embodiment of a multi-coil method of a magnetic-free dynamic speaker. As shown in FIG. 11, the voice coil 910 of FIG. 9 may be used in an overlapping form of several voice coils of an annular shape having the same diameter and number of turns.

The input signal entering each coil is received from the coil length adjustment unit 830 of FIG. 8. Each of the coils constituting a multi-layered voice coil is connected to one diaphragm, and the central axis direction of all coils is arranged so that the direction is not parallel to the direction of the static magnetic field B0 of the MRI. do.

Non-magnetic dynamic speakers using a multi-layered voice coil of FIGS. 9 to 11 may be installed in a headset or a non-headset type output device. In the case of the headset type, the diaphragm of the speaker is disposed toward the ear of the person. In the case of a non-headset type, a non-magnetic dynamic speaker may be installed in the shape of a structure located near a person's head in a shape such as a headrest, etc. toward the ear.

2) Coil length adjustment unit 830

The coil length adjusting unit 830 shown in FIG. 8 serves to input an acoustic signal to an input portion of a multi-layer voice coil. It inputs an audio signal to be output to a selected coil among each coil constituting the multi-layer voice coil of the magnetic-free dynamic speaker. Therefore, it is possible to adjust the length of the entire coil used by selecting the coil.

3) Current intensity adjustment unit 840

12 is a view showing an embodiment of the current intensity adjustment unit 840 of the non-magnetic dynamic speaker shown in FIG. The current intensity adjustment unit 840 includes a resistance selector 1210 for selecting a variable resistance or resistance size, and a gain controller 1220, and may adjust the intensity of current through selection of the resistance size. The current intensity adjustment unit 840 serves to maintain the output of the non-magnetic speaker by changing the current intensity of the voice coil attached to the diaphragm in the MRI magnetic field environment of various intensity as shown in FIG. 12. To this end, a variable resistor or the like may be used, and the strength of the current flowing through the voice coil may be changed by changing the resistance value according to the strength of the static magnetic field.

4) Static magnetic field strength analysis unit (850)

13 is a flowchart illustrating an operation of the static magnetic field strength analyzer 850 shown in FIG. 8 according to an embodiment. The static magnetic field strength analyzer 850 is a part that determines the strength of the current or the length of the coil according to the strength of the magnetic field. To this end, the strength of the coil or the current may be selected using information on the strength of the magnetic field analyzed in advance according to the location.

In step 1310, the coil length adjustment unit 830 of FIG. 8 initializes the coil length (ie, selection of the coil), or the current intensity adjustment unit 840 initializes the intensity of the current flowing through the coil.

In step 1320, the test signal is output to the speaker by the static magnetic field strength analyzer of FIG. 8.

In step 1330, an input signal input through a sensor near the diaphragm of the speaker, such as a microphone, is analyzed. Therefore, the output of the speaker can be analyzed through the sensor.

In step 1340, it is determined whether distortion has occurred and whether the output is insufficient from the analysis result of step 1330. Distortion indicates that the waveform of the input signal of the speaker, for example, the waveform of the test signal and the waveform of the output signal of the speaker, is not the same, and deformation occurs.

If distortion occurs or the output is insufficient, the length of the coil and the strength of the current flowing through the coil are adjusted in step 1350, and then the static magnetic field strength analysis is performed again.

If distortion does not occur and the output is not insufficient, the length of the coil and the strength of the current may be determined based on the result of the static magnetic field strength analysis.

Through the method of FIG. 13, a sensor such as a microphone may be mounted near the speaker diaphragm to select the intensity of the coil and current according to the situation.

5) Non-magnetic speaker vibration plate damper part

14 is a diagram illustrating a structure of a damper and a stopper of a magnetic-free speaker according to some embodiments. The non-magnetic speaker is arranged so that the central axes of the diaphragm 1430 and the voice coils 1410 and 1420 face the ears, for example, orthogonal to the static magnetic field, the semicircle of the voice coil (for example, the voice coil first semicircle, 1410). ) To descend in the direction of the central axis and another semicircle (eg, the second semicircle of the voice coil, 1420) ascending in the opposite direction of the central axis, simultaneously generating the resultant force of the force to move the coil. Extinguish. Accordingly, the entire voice coil has a motion mode in which it rotates by itself without linear movement. Since the circumference of the coil and the diaphragm 1430 are fixed to each other, and the supporting force of the diaphragm 1430 is uneven along the circumference of the coil (the material elastic repulsion force is different depending on the shape of the diaphragm), the diaphragm 1430 is Although linear motion is performed, the magnitude of the vibration is small, and the sound cannot be reproduced smoothly due to the effect that the local vibrations cancel each other. In order for the coil to make a sound according to the +/- current signal, it is necessary to make the coil move up and down with a large displacement in parallel with the central axis. When the motion of one coil semicircle part 1420 is restrained and the other semicircle part 1410 is moved, sound is reproduced by the displacement movement movement of the coil according to the current signal.

In FIG. 14, a vibration absorbing part for absorbing the vibration of the diaphragm 1430 on one side of the diaphragm 1430 as a method of constraining the coil semicircular part 1420, for example, a damper having constant elasticity and self-deformation such as a sponge ( 1440) is attached, and a stopper 1450 is installed thereon as a rigid boundary wall of plastic or the like. By configuring a damper 1440-stopper 1450 structure in the upward direction of the diaphragm 1430, it is possible to suppress torsional vibration of the diaphragm 1430 and increase the vertical linear motion. The spacing between the sponge damper 1440 and the plastic stopper 1450 and the thickness and density of the sponge damper 1440 may be important factors. When the rise of the diaphragm 1430 is prevented by the damper 1440-stopper 1450, if excessive force is applied, a problem of not being able to reproduce some frequencies due to distortion or abnormal deformation of the diaphragm 1430 occurs. There is a problem in that the sound output is weakened if the impeding force is very small or there is a large gap between the 1440 and the stopper 1450. In addition, since the weight of the damper 1440 attached to the diaphragm 1430 affects the natural frequency of the diaphragm and affects the change in frequency response characteristics, the appropriate size and weight may be adjusted according to the mechanical shape and structure of the diaphragm 1430. It must be attached.

Hereinafter, a noise reduction method according to some embodiments will be described with reference to FIGS. 15 and 16.

15 is a flow chart of a noise reduction method according to some embodiments.

In step 1510, a voice signal including noise is input.

In step 1520, the period of the noise pattern in the voice signal is estimated. In this case, period information of noise may be obtained from a device that generates noise, or period information may be calculated based on data of a voice signal acquired for a predetermined time.

In step 1530, the noise pattern is subtracted and removed from the voice signal in the frequency domain by using the estimated period of the noise pattern. In step 1530, matching the noise pattern with the timing index of the current frame of the speech signal, removing the spectrum of the timing noise frame from the current frame of the speech signal, and removing residual noise through post-processing. Can include.

In step 1540, the noise pattern is updated according to the change in the noise level. In step 1540, determining the presence or absence of speech in the current frame of the speech signal, if there is no speech in the current frame of the speech signal, updating the noise pattern and removing the noise pattern from the speech signal, the noise pattern is It may include determining whether the removed output is amplified and emitted larger than the input noise, and initializing noise pattern information when the output is emanating.

In step 1550, a voice signal from which the noise pattern has been removed is output.

16 is a schematic flowchart illustrating a voice communication noise reduction algorithm according to some embodiments.

17 is a detailed flowchart illustrating a voice communication noise reduction algorithm according to some embodiments. In FIG. 17, the period estimation step 1640, the noise removal step 1650, and the noise update step 1660 of FIG. 16 are shown in more detail.

-Speaker (eg, patient) voice acquisition step

The voice and noise signals of the micro speaker are simultaneously acquired and input to the noise reduction processing step. For example, in the MRI scan room, a patient's voice and noise signals are simultaneously acquired and input to the noise blocking process. At this time, since the environment has a high noise level (eg, an MRI environment), a speaker (eg, a patient) voice is input at a small level compared to the noise. Therefore, in some embodiments, a directional microphone is used to increase the SNR.

-Voice communication noise blocking step

Hereinafter, a voice communication noise reduction algorithm according to some embodiments will be described with reference to the flowcharts of FIGS. 16 and 17.

In step 1610 a program loop, e.g., a voice communication denoising algorithm, is started. The program loop may be performed repeatedly and continuously as shown, after ending at step 1690, and returning to step 1610.

In step 1620, a noise reduction algorithm according to some embodiments is executed on the input signal obtained in the speaker's speech acquisition step described above.

In step 1630, it is determined whether period estimation is completed, and if period estimation is not completed, a period estimation step is performed in step 1640. When period estimation is completed, a noise removal step is performed in step 1650.

In step 1640, when the algorithm of the period estimation step is operated, in step 1642, a sample count is started while storing the input signal in a noise buffer.

In step 1644, until the sample count reaches a predetermined threshold to estimate the period by storing the noise buffer by a sufficient size, for example, until the sample count has a value greater than a predetermined threshold. Repeat saving.

In step 1646, the period is estimated by calculating the period value of the input noise using the noise buffer, or period information is received and stored from the system of the device generating the noise.

Next, in step 1650, a noise removal step is performed.

When the storage of the period information value is completed in step 1646, the input is stored in a frame unit in step 1652. In this case, as the frame size is smaller, distortion can be reduced in noise removal.

In step 1654, in order to process a signal at a small frame size, information indicating which timing of the current frame matches the buffer in which the noise pattern is stored, that is, index information, is calculated as shown in Equation 1 below.

Here, the Index is information indicating at which timing the current frame matches the buffer in which the noise pattern is stored.

In step 1656, the noise buffer signal corresponding to the size of the input frame and the frame whose timing is matched is converted into the frequency domain using the index information value (for example, FFT (Fast Fourier Transform) is performed), and each magnitude is calculated. do.

는 추정된 음성,

는 잡음이 섞인 음성, 그리고

는 소음 패턴 신호를 의미한다.In step 1658, the calculated size information is subtracted as shown in Equation 2 below to separate the speech signal. At this time

Is the estimated voice,

Is a noisy voice, and

Means the noise pattern signal.

After removing the residual noise from the separated speech signal, in step 1659, it is converted to the time domain (for example, Inverse Fast Fourier Transform (IFFT)).

Next, in step 1660, a noise update step is performed.

In step 1662, it is determined whether there is a voice. If there is no voice in the current frame, in step 1664, the current frame of the noise pattern buffer is replaced with an input frame, and noise information is updated.

In step 1670, it is determined whether this output is diverging.

If the output diverges in step 1670, the process returns to step 1640 and the period estimation step is performed again.

If the output does not diverge in step 1670, the output is sent to the audio output in step 1680 and the program loop is terminated. In the program loop end step of step 1690, the program loop may be repeated by returning to step 1610.

-The other party's voice and content playback stage

Another voice input device acquires the other party's voice, and the speaker in the output device including the ANC function worn by the speaker performs active noise control and simultaneously outputs the voice.

An embodiment of the present invention is to propose an acoustic system capable of operating anywhere in an MRI environment by changing the strength of the current or the total length of the wire according to the strength of the B0 static magnetic field of the MRI that is different for each location. The speaker constituting the sound system generates sound by using a phenomenon in which a current flowing by the surrounding magnetic field moves as shown in Equation 3 below.

Here, F is the force received by the wire, L is the length of the wire, and I is the current flowing through the wire.

In addition, among the elements of Equation 3 above, the magnetic field (B) is an element that basically exists strongly in the MRI environment, and the non-magnetic dynamic speaker uses the static magnetic field of the MRI instead of the normal magnet. However, since the strength of the static magnetic field of the MRI varies from position to position, the force F received by the voice coil varies from position to position. In order to keep the force F received by the voice coil constant in an environment in which the strength B of the magnetic field is changed, in some embodiments, the length L of the conducting wire or the current I flowing through the conducting wire is changed.

The operation of some embodiments for this will be described with reference to FIG. 18.

18 is a flowchart illustrating an active noise control method according to some embodiments.

In step 1810, the output device is selected according to the strength of the static magnetic field according to the location of the output device.

In step 1810, the steps of sequentially operating the coils of the shortest length from the multi-coils included in the output device, analyzing the output signal acquired through an analysis sensor for measuring the response of the output device, for example, a microphone , Determining whether distortion occurs in which the waveform of the output signal is transformed from the waveform of the input signal input to the output device and whether the output can be reproduced louder than noise, and the output is reproducible without distortion. Selecting the shortest coil may be included.

Further, in step 1810, the current intensity adjustment unit included in the output device operates in order from the largest resistance, and an analysis sensor for measuring the response of the output device, for example, an output signal acquired through a microphone is analyzed. The step of determining whether the output signal waveform is transformed from the waveform of the input signal input to the output device, determining whether distortion occurs and whether the output can be reproduced louder than noise, and the output can be reproduced without distortion. It may include the step of selecting the largest resistance value among the resistance values.

In step 1820, the gain used for the selected output device is adjusted. In step 1820, a gain value may be calculated based on the output of the non-magnetic dynamic transducer using the selected coil among the multi-coils included in the output device and the level information of the input noise. In addition, in operation 1820, a step of calculating a gain value based on level information of the input noise and output based on a selected resistance value among the resistance values of the current intensity adjusting unit included in the output device may be included.

In step 1830, an active noise control signal is calculated by reflecting the selected output device and the gain. In step 1830, based on the analysis sensor for measuring the response of the output device, for example, output characteristic information between the microphone and the selected output device, and noise information input through the analysis sensor, noise input through the analysis sensor It may include configuring a filter for generating an active noise control signal to reduce the size of the. In addition, in step 1830, based on the analysis sensor for measuring the response of the output device, for example, output characteristic information between the microphone and the selected output device, and noise information input through the analysis sensor, the selected sound output device is It may include the step of calculating an active noise control signal using the fixed filter input in advance.

In step 1840, noise control is performed by outputting the calculated active noise control signal to the selected output device.

Hereinafter, an active noise control method will be described in detail.

First, it goes through a process of selecting an appropriate coil length and current strength according to the strength of the static magnetic field. Next, the selected coil and current strength are used to output a sound having an appropriate output for the strength of the static magnetic field at the location.

In the process of selecting an appropriate coil length and current strength according to the strength of the static magnetic field, information about the strength of the magnetic field analyzed in advance may be used according to the location. In the process of using the previously analyzed information, as shown in FIG. 3, the length of the coil or the strength of the current is determined for the static magnetic field strength information measured in advance, and the length of the coil and the strength of the current are changed according to the position where the speaker is placed. And use it.

In addition, the length of the coil or the strength of the current can be automatically determined using a sensor such as a microphone. When the speaker is located inside the static magnetic field of the MRI and starts to operate, the current intensity according to the static magnetic field intensity and the automatic selection process of the coil begin. As shown in FIG. 10, for example, in a multi-coil including coil 1, coil 2, and coil 3 having different lengths, first, the length of the shortest coil is selected. Next, a pre-input test signal is input to the system, and the signal coming out of the speaker is received through a sensor such as a microphone. The received signal is analyzed to measure whether or not distortion of the speaker occurs and the size of the output. If distortion occurs or the output is insufficient, the strength of the MRI static magnetic field is weak, so the next coil length is selected. The length of the coil undergoes a process of analyzing the microphone input signal by reproducing the test signal, and repeating the process by increasing the length of the coil until an appropriate output is possible. The process of determining the current strength according to the strength of the static magnetic field is the same, and the current strength is determined from the minimum level to an appropriate level through the process of repeatedly measuring until no distortion or sufficient output is produced.

Hereinafter, an operation of an active noise control (ANC) system according to an embodiment of the present invention will be described.

The active noise control system is activated before starting the MRI imaging. Before taking an MRI image, an appropriate coil or transducer is selected according to the position of the magnetic-free dynamic speaker. In the case of a multi-layer coil type non-magnetic dynamic speaker, the shortest coil is operated in order. The operation uses test signals such as random noise or sweep sine signals. After outputting the test signal of maximum output, the input signal coming through the microphone is analyzed.

As a result of analyzing the input signal of the micro, if distortion occurs or the output is insufficient, the next short coil is sequentially operated. For the coils operated in sequence, characteristics are tested in order of coil length by analyzing whether distortion occurs and the magnitude of the output for the test signal with the maximum output. When it is confirmed that distortion does not occur in the coil and the output (Level _speaker ) is sufficient compared to the level of _{MRI noise} (Level _{MRI _ noise} ) as shown in Equation 4 below, the coil currently being tested is Decide to use it and move on to the next step.

In the case of a multi-transducer type non-magnetic dynamic speaker, first input a test signal to the coil and analyze the output signal of the speaker coming from the microphone. A piezoelectric element may be selected if distortion occurs by analyzing the output signal that varies depending on the position of the speaker and the angle with the B0 static magnetic field of the MRI, or if the size of the output is insufficient compared to the level of the MRI noise. When the coil is operated, if there is no distortion and the output is sufficient compared to the MRI noise, select the coil and move on to the next step.

When a coil or transducer is selected, a gain for the coil or transducer is calculated and reflected. The gain is determined by comparing the level of the MRI noise and the output of the speaker, which is determined due to the distorted angle of the non-magnetic dynamic speaker with respect to the speaker position and MRI static magnetic field B0. Alternatively, the gain is determined by comparing the magnitude of the output of the piezoelectric speaker and the level of the MRI noise. At this time, the level information of the MRI noise is pre-measured and statistical information is used, and the output size of the speaker is the data information input to the microphone when analyzing the test signal for selecting a coil or transducer. In this way, the gain is calculated with a value between 0 and 1 based on the ratio information of the sound pressure level based on the microphone.

19 is a diagram illustrating an active noise reduction algorithm according to some embodiments.

When the coil or transducer is determined and the gain is determined, MRI imaging is started, and an active noise control filter (W, 1960) is calculated for the generated noise as shown in FIG. 19.

In step 1910, output is performed from a non-magnetic dynamic speaker.

In step 1920, output characteristic information of the magnetic-free dynamic speaker 1910 and input information of a microphone near the ear (error microphone) are acquired.

In step 1930, the noise control filter is updated to reduce the MRI noise coming into the microphone (error microphone, 1920) near the ear, and the input information of the near-ear microphone 1920 obtained in step 1920 and the non-magnetic dynamic speaker 1910 Use the output characteristic information of.

In step 1940, input signal information of a second microphone (reference microphone, 1940) for analyzing characteristics of MRI noise is also obtained.

In step 1950, the input from the reference microphone 1940 is corrected using a transfer function between the dynamic speaker 1910 and the error microphone 1920. sh' is the block representing this correction.

In step 1960, a final MRI noise control signal is calculated by synthesizing and multiplying the input signal of the second microphone 1940 or the near-ear microphone 1920 with the control filter W calculated and updated in the process described in step 1930.

In step 1970, the gain is reflected in the calculated MRI noise control signal.

In step 1980, a control signal is sent to an input line of a coil or transducer already selected by a coil selector or a transducer selector.

In step 1910, a control signal is output through a non-magnetic dynamic speaker. The control signal output through the magnetic-free dynamic speaker cancels the MRI noise near the ear and reduces the amount of MRI noise reaching the ear. The reduced MRI noise level is analyzed again through a microphone near the ear in step 1920 and used again to update the control filter in step 1930. Through this repetitive process, active noise control for the MRI noise is performed.

Some embodiments may provide a system and method for communicating between a speaker and a counterpart during operation of a device that generates a periodic noise, thereby enabling smooth communication. In addition, some embodiments provide a system and method for communicating between a patient and a medical staff during an MRI operation, so that the medical staff can monitor a patient's situation during a long MRI imaging time. In addition, there is an advantage of reducing fatigue and stress in both directions (for example, patients and doctors) communicating by blocking noise signals and transmitting voice signals in a high-noise environment. The method for blocking noise suggested in some embodiments applies a method of removing the spectrum of only the noise signal from the input signal in the frequency domain by using the pattern of the noise signal in the noise blocking step, even when the noise pattern does not match exactly. The noise is not amplified and it shows stable performance. In addition, it has the advantage of adapting to rapidly changing noise patterns through the operation of removing residual noise and updating the noise pattern through monitoring.

In addition, some embodiments relate to a method of effective active noise control against low-frequency noise even in an MRI environment by using a magnetic-free dynamic speaker. Unlike the conventional method of using a magnetic-free dynamic speaker that does not work properly outside the bore where the strength of the static magnetic field is weak, in some embodiments, a magnetic-free dynamic speaker having a structure that operates regardless of the strength of the static magnetic field is used. Noise can be reduced through active noise control. To this end, a structure capable of controlling the total amount of current flowing through the voice coil was proposed, and a method through adjusting the length of the voice coil or adjusting the current intensity was proposed. By combining this method and the noise control algorithm, an active noise control system using a magnetic-free dynamic speaker that can operate both inside and outside the bore of the MRI can be used, and through this, the change of the static magnetic field strength according to the position inside or outside the bore of the MRI or MRI noise can be reduced regardless of the change in the output of the speaker according to the angle with the B0 static magnetic field formed by the arrangement of the speaker.

20 illustrates a magnetic resonance imaging (MRI) system according to some embodiments. 21 shows a simulation result of blocking noise by applying an algorithm to MRI noise. The magnetic resonance imaging (MRI) system includes a bore 2010, a first voice input device 2020, a second voice input device 2030, a first output device 2040, a second output device 2050, and active noise control. An apparatus 2060, and a noise reduction signal processing apparatus 2070.

Bohr (2010) generates a magnetic field. The magnetic field is generated by a magnet or an electromagnet, and when the magnetic field is generated, information about the inside of the patient's human body in the bore can be obtained using electromagnetic waves.

The first voice input device 2020 receives an audio signal including the voice of a first user at the bore 2010 side, for example, a patient in the scan room. The first audio input device 2020 can acquire audio with high SNR even in an environment with a high noise level using a directional microphone.

The second voice input device 2030 receives an audio signal including voice from a second user controlling the MRI, for example, a medical staff in a control room.

The first output device 2040 outputs an acoustic signal including the voice of the second user (medical staff) to the first user (patient) using a magnetic-free speaker.

The second output device 2050 outputs an acoustic signal including the voice of the first user (patient) to the second user (medical staff).

The active noise control device 2060 performs active noise control on the sound signal of the second voice input device 2030 according to the strength of the MRI static magnetic field according to the position of the first output device 2040, and the first output device 2040 To pass on.

The noise canceling signal processing apparatus 2070 estimates a noise pattern from the acoustic signal of the first voice input apparatus 2020, removes the noise pattern in the frequency domain, and transmits the noise pattern to the second output apparatus 2050.

According to some embodiments, a two-way communication operation in an MRI device is possible as shown in FIG. 20. The voice spoken by the patient inside the MRI bore (2010) is input with the MRI noise through the first voice input device (2020), and the voice signal from which the noise is removed is transmitted to the medical staff through the noise reduction signal processing (2070). Voice of the patient can also be output to the patient through the headset 2040 including the ANC function. In this case, the microphone 2020 for voice acquisition and the headset 2040 used inside the MRI scan room should be configured as devices that do not affect the MRI scan. The headset 2040 device used inside the MRI scan room may be replaced with a speaker 2040 mounted in the MRI scan room or the main body.

21 is a result of blocking MRI noise and obtaining a voice signal by applying a noise reduction algorithm in an actual MRI equipment.

In addition, entertainment services based on patient characteristics during shooting are possible. By using a display and a sound reproducing device, the patient can select the contents of the electronic device linked to the outside, and it is possible to provide music and images that induce the patient's stability. In addition, it is possible to shorten the entire hospital treatment time through additional diagnosis of the patient and remote consultation with the doctor during the long shooting time, and provide customized medical information and advertisements for patients.

In addition, when communicating with an outside person in a helicopter or a factory that generates a periodic and high level of noise, it is possible to transmit a voice signal to the other party while blocking the noise by using the method of some embodiments. In addition, when communicating in a situation that generates periodic noise, such as transportation means such as automobiles, motorcycles, and trains, and household appliances such as vacuum cleaners and washing machines, voice signals may be transmitted by removing noise using some embodiments.

Meanwhile, an embodiment of the present invention may be implemented as a computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all types of recording devices storing data that can be read by a computer system.

Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tapes, floppy disks, optical data storage devices, and the like, and are implemented in the form of carrier waves (for example, transmission over the Internet). Include. Further, the computer-readable recording medium is distributed over a computer system connected by a network, so that the computer-readable code can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily inferred by programmers in the technical field to which the present invention belongs.

So far, we have looked at the center of the preferred embodiment for the present invention. Those of ordinary skill in the art to which the present invention pertains will understand that the present invention can be implemented in a modified form without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered from an illustrative point of view rather than a limiting point of view. The scope of the present invention is shown in the claims rather than the above description, and all differences within the scope equivalent thereto should be interpreted as being included in the present invention.

Claims (25)

An input unit that receives a voice signal including noise,
A period estimating unit that estimates a period of a noise pattern in the speech signal,
A noise removal unit for subtracting and removing the noise pattern from the voice signal in a frequency domain using the estimated noise pattern period,
A noise update unit for updating the noise pattern according to a change in the amount of the noise, and
Using a magnetic-free speaker including a plurality of coils, including an output unit for outputting the sound signal from which the noise pattern has been removed,
The output unit determines the length of the coil by selecting at least one of the plurality of coils according to a change in the output of the non-magnetic speaker. The method of claim 1,
The period estimating unit acquires period information of the noise from the device generating the noise, or calculates period information based on data of the audio signal acquired for a predetermined time. The method of claim 1,
The noise reduction unit,
An alignment unit that matches the noise pattern and the timing index of the current frame of the voice signal,
And a calculation unit that removes a spectrum of a noise frame with timing from the current frame of the speech signal and removes residual noise through post-processing. The method of claim 1,
The noise update unit,
It is determined whether there is a voice in the current frame of the voice signal,
If there is no voice in the current frame of the voice signal, update the noise pattern, and request the noise removal unit to remove the noise pattern from the voice signal,
It is determined whether the output from which the noise pattern is removed is amplified and emitted larger than the input noise,
When the output is diverged, the sound device is requested to initialize the noise pattern information to the period estimation unit. Receiving a voice signal including noise,
Estimating a period of a noise pattern in the voice signal,
Subtracting and removing the noise pattern from the voice signal in a frequency domain using the estimated period of the noise pattern,
Updating the noise pattern according to the change in the amount of the noise, and
Including the step of outputting the voice signal from which the noise pattern has been removed using a magnetic-free speaker including a plurality of coils,
In the outputting step, the length of the coil is determined by selecting at least one of the plurality of coils according to a change in the output of the non-magnetic speaker. The method of claim 5,
Estimating the period of the noise pattern,
And obtaining period information of the noise from the device generating the noise, or calculating period information based on data of the voice signal acquired for a predetermined time. The method of claim 5,
The step of subtracting and removing the noise pattern,
Matching the noise pattern with the timing index of the current frame of the speech signal,
Removing a spectrum of a timed noise frame from the current frame of the speech signal, and
Noise reduction method comprising the step of removing residual noise through post-processing. The method of claim 5,
Updating the noise pattern,
Determining the presence or absence of voice in the current frame of the voice signal,
If there is no voice in the current frame of the voice signal, updating the noise pattern and removing the noise pattern from the voice signal,
Determining whether the output from which the noise pattern has been removed is amplified and emitted larger than the input noise, and
And initializing noise pattern information when the output is divergent. delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete delete

Images