KR100553557B1 - Method for Noise Cancelling by Using 2 Microphones for Improve Speech Recognition of Wireless Telematics Services - Google Patents

Abstract

According to the present invention, two microphones are used in hands-free, and one microphone is inputted with voice and noise, and the other microphone is inputted with noise. The present invention relates to a noise removing method.

To realize this, hands-free in the present invention include two microphones, an ADC, a switching circuit, a DSP, a ROM, a RAM, and a DAC.

According to the present invention, when the user uses the telematics service by voice while driving, the user's voice is incorrectly recognized due to noise and the wrong information is provided, and when the user makes a voice call using the hands-free kit. In addition, it can improve the quality of calls by delivering clear voices without noise to the other party. Also, users who use mobile communication terminal can be less likely to disturb the user's nerve because the telematics service or voice call is misrecognized. Focusing on driving has the effect of reducing the incidence of accidents.

Hands-free, Telematics, Microphone, Noise, Speech Recognition Rate, Beamforming, SAD, Infomax, Wiener Filter, Mobile Communication Terminal

Description

Noise for Noise Canceling by Using 2 Microphones for Improve Speech Recognition of Wireless Telematics Services}

1 is a schematic view showing a telematics service system using a hands-free kit,

2 is a block diagram showing a noise removing process in a hands-free kit using a conventional microphone;

3 is a block diagram showing a noise removing process in a hands-free kit using two microphones according to the present invention;

4 is a block diagram illustrating a noise removing process using a beamforming method according to a first embodiment of the present invention;

5 is a block diagram showing a noise removal process using the SAD method according to a second embodiment of the present invention;

6 is an experimental result graph showing that noise is removed by the SAD method,

7 is a block diagram showing a noise removal process using the Infomax method according to a third embodiment of the present invention;

8 is an experimental result graph showing the change after the speech signal passes through the Wiener filter in the Infomax method;

9 is a block diagram illustrating a signal processing procedure according to the present invention applied to a mobile communication terminal.

<Description of Symbols for Main Parts of Drawings>

112: mobile communication terminal 114: hands free kit

122: wireless base station 124: base station transmitter

126: base station controller 128: mobile communication switching center

128: HLR 132: VLR

134: IWF 142: Telematics Service Center

144: router 146: hub

148: subscriber server 150: data data server

202: microphone 204: ADC

206: DSP 208: RAM

210: ROM 212: DAC

302: first microphone 304: first ADC

306: first switching circuit 308: second microphone

310: second ADC 312: second switching circuit

402: voice signal 404: noise signal

406: adder 408: first subtractor

410: second subtractor 412: input voice signal

510: First post-processing filter 512 Decorrelation Criterion

514: second post-processing filter 718: Wiener filter

720: Input voice 726: Entropy Maximization

902 microprocessor 904 digital signal processing unit (DSP)

906: baseband converter 908: first microphone

910: second microphone 912: RF signal processing unit

914: antenna

The present invention relates to a method of removing noise using two microphones when a voice is input using hands free. More specifically, two microphones in hands-free, using two microphones to remove voice and signal from one microphone and noise from one microphone, removing noise from the signal input to the two microphones, and delivering only the voice. It relates to a noise removing method using.

As computers, electronics, and communication technologies have advanced rapidly, various wireless communication services using wireless communication networks have been provided. The most basic wireless communication service is a wireless voice call service that provides a voice call to a mobile terminal user wirelessly, and can provide a service regardless of time and place. In addition, recently, as the number of self-driving drivers increases, telematics services are emerging to guide drivers of road conditions, vehicle accidents, driving routes to destinations, etc. using mobile communication terminals.

A user uses a hands free kit to receive a telematics service or to make a call using a mobile communication terminal while driving. The use of the hands free kit is legally mandated to allow the user to use the mobile terminal safely while driving. Here, Hands Free refers to a device that enables a call without having to hold the phone by hand.

1 is a schematic diagram illustrating a telematics service system using a hands-free kit.

The mobile communication terminal 112 is a terminal capable of making a voice call with the counterpart through the mobile communication network 120 and performing data communication with the mobile communication network 120. The mobile communication terminal 112 through the mobile communication network 120 may include a personal digital assistant (PDA), a cellular phone, a personal communication service (PCS) phone, a global system for mobile (GSM) phone, a wideband CDMA (W-CDMA) phone, CDMA-2000 phone, Mobile Broadband System (MBS) phone and the like. Here, MBS phone refers to a mobile phone to be used in the fourth generation system currently being discussed.

The hands-free kit 114 can make a call using a separate external microphone and speaker to prevent a dangerous situation in which the user needs to operate the device of the vehicle with the other hand while talking on the phone while holding the mobile phone with one hand. It is designed to be composed of circuit part, mechanism part and charging part.

The hands-free kit 114 includes a full kit hands-free, a cradle-integrated hands-free, a cigar jack-type portable hands-free, etc., depending on the circuit mounting method and the configuration of the cradle. , Earphone-type hands-free, contact-type hands-free, and so on. In addition, a boom or necklace type ear microphone may be classified as an earphone hands-free, but it merely serves to amplify and transmit voice.

The signal passing through the mobile communication terminal 112 and the hands free kit 114 in the vehicle is transmitted to the wireless base station 122 of the mobile communication network 120.

The wireless base station 122 transmits the location information of the mobile communication terminal 112, the registrant information, and the call request signal and data request signal from the mobile communication terminal 112 to the mobile communication switching center 128, and the mobile communication switching center 128. And a call request signal and data signal from the mobile communication terminal 112. In addition, the wireless base station 122 is composed of a base station transmitter 124, a base station controller 126 and a repeater (Enhancer: not shown).

In this case, the base station transmitter 124 may be a base transceiver station (BTS) and the base station controller 126 may be a base station controller (BSC). In the asynchronous case, the base station transmitter 124 may be a radio transceiver subsystem (RTS). The base station controller 126 may be a radio network controller (RNC). Of course, the wireless access network is not limited thereto, and may include a GSM network other than a CDMA network and an access network of a fourth generation mobile communication system to be implemented in the future.

The base station transmitter 124 is arranged in units of cells, receives a call request signal and a data request signal from the mobile communication terminal 112 through a traffic channel among signal channels, and receives the received call request. The signal is transmitted to the base station controller 126 or the call request signal and the data signal transmitted from the mobile communication switching station 128 through the base station controller 126 to the mobile communication terminal 112.

In addition, the base station transmitter 124 performs location registration to determine the location of the mobile communication terminal 112 existing in the cell area under its control. In addition, the base station transmitter 122 is a network endpoint device that is directly connected to the mobile communication terminal 112 by performing baseband signal processing, wired and wireless conversion, and transmission and reception of wireless signals.

The base station transmitter 124 may obtain information such as latitude and longitude at which the base station transmitter 124 is located from a global positioning system (GPS), and the position information of the base station transmitter 124 may be converted into a system of a forward link call channel. The message is transmitted to the mobile communication terminal 112 through the parameter message. The mobile communication terminal 112 may register new location information by calculating the moving distance of the mobile communication terminal 112 by using the location information of the base station transmitter 124 of the cell to which it belongs.

Location registration is a processing procedure for informing the mobile telecommunications station 128 of the location, status, identifier, slot period, and other features of the mobile communication terminal 112 via the base station transmitter 124, where the base station transmitter 124 moves. This is a procedure for effectively calling the mobile communication terminal 112 when attempting to set up an incoming call to the communication terminal 112. The location registration of the mobile communication terminal 112 is performed when the mobile communication terminal 112 is powered on or off, when the mobile communication terminal 112 moves between mobile communication switching stations 128, and the mobile communication terminal 112. It is implemented when the parameter of) changes.

The base station controller 126 controls the base station transmitter 124, and assigns and releases radio channels for the mobile communication terminal 112, control of transmission output of the mobile communication terminal 112 and the base station transmitter 124, and inter-cell soft hand. Soft handoff and hard handoff determination, transcoding and vocoding, GPS clock distribution, operation and maintenance of the base station are performed. The base station controller 126 also transmits subscriber information of the location registered mobile communication terminal 112 to the mobile communication switching center 128. The base station controller 126 transfers the call request signal and the data request signal transmitted from the mobile communication terminal 112 through the base station transmitter 124 to the mobile communication switching center 128 or the call transferred from the mobile communication switching center 128. It transmits the request signal and the data signal to the receiving mobile communication terminal 112 through the base station transmitter 124.

The mobile communication switching center 128 performs basic and additional service processing, incoming and outgoing call processing of subscribers, location registration procedure and handoff procedure processing, interworking with other networks, and the like. When the user transmits a call request signal through the hands-free kit 110, the mobile communication switching center 128 connects to another wireless base station so that the call can be made. When the user requests a telematics service, the telematics service center 140 By connecting through the voice or through the IWF (134) allows the mobile communication terminal 112 to receive a telematics service.

The mobile communication switching center 128 of the IS-95 A / B / C system includes an access switching subsystem (ASS) that performs distributed call processing, an interconnection network subsystem (INS) that performs centralized call processing, operation and Subsystems such as Central Control Subsystem (CCS), which is responsible for the centralization of maintenance, and Location Registration Subsystem (LRS), which perform storage and management of information on mobile subscribers.

In addition, the mobile communication switching center 128 for 3rd and 4th generation may include an Asynchronous Transfer Mode (ATM) switch (not shown). The ATM switch increases the transmission speed and the efficiency of line use by the packet transmission per cell. Let

When the location registration of the mobile communication terminal 112 is performed through the wireless base station 122, the mobile communication switching center 128 moves to a visitor location register (VLR) 132. After temporarily storing the subscriber information of the communication terminal 112, the mobile device 112 requests a location registration of the mobile communication terminal 112 to a home location register (HLR).

Here, the HLR 130 is a database that stores a service profile regarding subscriber information of the user of the mobile communication terminal 112. The mobile identification number (MIN) of the mobile communication terminal 112, including the subscriber's telephone number, is included. ), Terminal serial number (ESN) and service type information. HLR 130 performs a function of storing subscriber information including information of wireless base station 122 and mobile communication switching center 128 where mobile communication terminal 112 is located. The mobile communication switching center 128 is composed of a control unit, a call path unit, and a peripheral device, and also has a charging data collection function for the mobile communication terminal 112.

The VLR 132 temporarily stores subscriber information and notifies the HLR 130 when the mobile communication terminal 112 performs location registration, and transmits the terminal identification number of the mobile communication terminal 112 from the HLR 130. It performs a function to utilize the location control, call processing, external operation processing of the mobile communication terminal 112 while receiving and managing a copy of the terminal unique number and service information.

A manganese interworking device (IWF: interworking function, hereinafter referred to as 'IWF') 134 connects a mobile communication switching center 128 and a telematics service provider 140 to exchange data between a wireless communication system and a wired communication system. It provides an interface for converting protocols, signals and data into a form suitable for each network.

The telematics service company 140 is an individual operator that provides a road traffic situation, a vehicle accident status, a driving route to a destination, and the like through a mobile communication network 120 to a telematics service subscriber.

A user requests a telematics service through the mobile communication terminal 112 and can request a service through voice, or a service through the wireless Internet.

When the user requests to provide a service through voice, the service center 142 is connected with the mobile telecommunication switching center 128 and operates the data data server 150 to provide the service requested by the user.

The router 144 and the hub 146 may use the mobile communication terminal 112, the subscriber server 148, and the data data server through the IWF 134 when a user requests a service through the wireless Internet of the mobile communication terminal 112. Connect 150.

The subscriber server 148 stores basic personal information of subscribers subscribed to the telematics service, contents set by the subscriber (mainly used road, home address, main destination address, etc.), terminal identification number of the mobile communication terminal 112, and the like. Doing.

The data data server 150 stores information to be provided to the user in a data format such as a map, road information, and street, and provides information on a region where the user is located when requesting service through voice or wireless internet through data communication.

When the user requests a telematics service by voice as shown in FIG. 1, the hands-free kit 114 is used. In addition, hands-free kit 114 is used even when talking on the phone while driving. The hands-free kit 114 has a microphone and a speaker connected to the mobile communication terminal 112, so that the user can input a voice from a long distance.

However, since the use of the hands-free kit 114 is performed in a vehicle with a high noise level, not in a quiet environment, the hands-free kit 114 removes the ambient noise mixed with the user's voice and inputs it to the mobile communication terminal 112. It includes.

2 is a block diagram illustrating a process of removing noise in the hands-free kit 114.

The microphone 202 is a device that receives a user's voice signal and converts it into an electrical signal.

Analog-to-Digital Converter (ADC) 204 is a device that converts a voice electrical signal input in analog format into a digital format signal. Since the voice electrical signal is an analog signal, the signal is sampled at regular time intervals and then quantized to be converted into a digital voice signal.

The digital signal processor (DSP) 206 is a digital signal processing processor that codes or decodes a digital voice signal input through the ADC 204 and performs an equalization function to remove noise.

The RAM 208 is a memory that temporarily increases the processing speed by storing the result in the DSP 206, and the ROM 210 stores a noise removing algorithm executed in the DSP 206.

The digital-to-analog converter (DAC) 212 is a device for converting the DSP processing signal from which the noise is removed in the DSP 206 into an analog format. Since the signal received through the ear microphone jack of the mobile communication terminal 112 should be an analog signal, a process of converting the DSP processing signal of the digital format processed by the DSP 206 into an analog signal without being input as it is is necessary. .

Through this process, the voice signal input through the hands-free kit 114 is input to the mobile communication network 120 through the mobile communication terminal 112 through the noise removing process, but the voice of the user input in the noisy vehicle It does not cleanly remove the noise contained in the network, which makes it difficult to receive telephony and telematics services.

In addition, some constraints are required to remove noise using a single microphone. For example, a signal of about 100 msec that is initially input is a condition that noise is unconditionally, and a condition that noise does not change suddenly in a section where voice is extracted. Therefore, it is impossible to remove noise such as music with a large deviation or an unexpected car horn.

If the voice is not properly transmitted due to a lot of noise, the driving user may feel distressed by using the hands free and directly use the mobile communication terminal without using the hands free. This can increase the incidence of accidents because the driver can not concentrate on driving.

In order to solve the above problems, the present invention uses two microphones in a hands-free kit, one microphone receives a user's voice including noise, and the other microphone receives only noise and receives two noise signals. It is an object of the present invention to provide a method for reliably removing noise to improve speech recognition rate.

In order to achieve the above object, the present invention provides a method for removing noise input to a mobile phone using two microphones in a hands-free kit, wherein (a) a voice signal and a noise signal are provided in the first microphone and the second microphone. An input step; (b) adding a signal input to the first microphone and a signal input to the second microphone; (c) subtracting the signal input to the second microphone from the signal input to the first microphone; (d) subtracting the signal output as a result of step (c) from the signal output as a result of step (b); And (e) using the resultant signal output from step (d) as an input voice to the mobile phone.

In addition, according to the second object of the present invention, in the method for removing noise input to the mobile phone using two microphones in the hands-free kit, (a) the step of inputting a voice and a distorted noise signal to the first microphone ; (b) inputting a noise and a distorted voice signal to the second microphone; (c) subtracting a signal multiplied by a correction variable from a signal input to the second microphone from a signal input to the first microphone; (d) inputting the signal calculated in step (c) into the post-processing filter; And (e) using the signal passing through the post-processing filter as an input voice to the mobile phone.

According to a third object of the present invention, there is provided a method of removing noise input to a mobile phone using two microphones in a hands-free kit, comprising: (a) a first distorted voice signal and a first distorted signal in a first microphone; Inputting the noise signal; (b) inputting a second distorted noise signal and a second distorted voice signal to the second microphone; (c) correcting the distorted portion of the signal input in steps (a) and (b) by means of the correction variable so that only the voice signal portion is extracted; (d) inputting the signal extracted in step (c) to the Wiener filter; And (e) using the signal passing through the Wiener filter as the input voice to the mobile phone.

According to a fourth aspect of the present invention, there is provided a mobile communication terminal for removing noise input using two microphones, the mobile communication terminal comprising: a microprocessor for controlling a noise processing process in a mobile communication terminal; A first microphone for converting a voice + noise signal into an electrical signal; A second microphone for converting a noise signal into an electrical signal; A baseband converter for converting a signal to be transmitted and received into a baseband signal and performing digital-to-analog conversion and analog-to-digital conversion processing functions; A digital signal processor which performs a noise canceling function of the digital signal received through the baseband converter; An RF signal processor for modulating a transmission signal applied from the baseband converter; And an antenna for transmitting a wireless signal to the mobile communication network.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. First, in adding reference numerals to the components of each drawing, it should be noted that the same reference numerals are used as much as possible even if displayed on different drawings. In addition, in describing the present invention, when it is determined that the detailed description of the related well-known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

3 is a block diagram showing a noise removal process in a hands-free kit using two microphones according to the present invention.

The hands-free kit according to the present invention includes a first microphone 302 for inputting a user's voice signal and a noise signal, and a second microphone 308 for inputting only a noise signal.

The first microphone 302 receives a user's voice signal and noise signal, converts the signal into an electrical signal, and transmits the converted signal to the first ADC 304. The first ADC 304 converts the user's voice electrical signal and noise electrical signal received in an analog form into a digital form.

The first switching circuit 306 is a device for inputting the digital voice signal and the digital noise signal of the user input from the first ADC 304 to the digital format to the DSP 206.

The second microphone 308 is spaced apart from the first microphone 302 so that a user's voice is not directly input and receives a noise signal. The first microphone 302 is installed in the form of a microphone so as to stand out from the user, but the second microphone 308 does not need to know the presence or absence of the user, so the side or top or bottom of the hands-free kit 112, Or it can be installed on the back.

The second ADC 310 converts the analog electrical noise signal input through the second microphone 308 into the digital form, and the second switching circuit 312 transfers the converted digital noise signal to the DSP 206. do.

The DSP 206 receives the digital voice signal and the digital noise signal input through the first switching circuit 306 and the digital noise signal input through the second switching circuit 312, and the RAM 208 and the ROM 210. Noise is removed through the noise reduction algorithm of the A), and only the voice of the user is extracted and transmitted to the DAC 212.

The ROM 210 stores a noise canceling algorithm according to the present invention, rather than a conventional noise canceling algorithm, and is used when removing noise from the DSP 314.

The noise canceling algorithm to be input to the ROM 210 according to the present invention includes a beamforming method, a symmetric adaptive decorrelation (SAD) method, and an information maximization (Infomax) method.

4 is a block diagram illustrating a noise removal process using a beamforming method according to a first embodiment of the present invention.

The beamforming method is an algorithm that assumes noise as a difference between a signal input to the first microphone 302 and a signal input to the second microphone 308. The first microphone 302 and the second microphone 308 are separated by a distance. If there is no noise, only a phase difference is generated between a signal input to the first microphone 302 and a signal input between the second microphone 308 by the time that the voice moves the distance between the two microphones. Therefore, if only the phase difference is corrected, the two signals must match.

However, if there is noise, even if the phase difference is corrected, a difference as much as that noise occurs. Therefore, if the difference value is obtained, the noise value can be measured, and only the voice can be extracted from the signal including the voice and the noise as the noise value. This is the basic concept of the beamforming method.

The voice 402 containing the contents to be delivered by the user and the noise 404 generated by the surrounding environment are input to the first microphone 302 and the second microphone 308.

In the adder 406, a signal input through the first microphone 302 and a signal input through the second microphone 308 are added. The added signal is a mixture of voice and noise.

The first subtractor 408 subtracts the signal input to the second microphone 308 from the signal input through the first microphone 302. Since the signal input to the first microphone 302 and the signal input to the second microphone 308 are assumed to be noise at home, a value output through the first subtractor 408 becomes a value of noise.

The second subtractor 410 subtracts the noise output through the first subtractor 408 from the signal mixed with the voice and the noise output through the adder 406. The noise-free input voice 412 may be extracted through the second subtractor 410.

In the beamforming method, if the distance between the first microphone 302 and the second microphone 308 is too close or the direction of the input voice and the noise is similar, it is difficult to extract the input voice 412.

Sound is about 340 meters per second at room temperature (18 ° C), so when you sample the signal at 8000 Hz, it arrives at the microphone at a distance of about 4 cm per sample. Therefore, the distance between the two microphones should be at least 4 cm. Also, the greater the distance between the two microphones, the better the results.

5 is a block diagram illustrating a noise removal process using the SAD method according to the second embodiment of the present invention.

The SAD method assumes that the noise distorted by the voice 402 and the environment is input through the first microphone 302, and the noise distorted by the noise 404 and the environment is input through the second microphone 308. Algorithm

H 502 is a variable that is distorted by the environment when voice 402 is input to second microphone 308. H 2 504 is a variable that is distorted by the environment when noise 404 is input to the first microphone 302.

Therefore, if the magnitude of the voice 402 signal is V and the magnitude of the noise 404 signal is N, V + H 2 N is input to the first microphone 302 and N + H₁V to the second microphone 308. Is input.

In order to extract the input voice 516 from which the noise is removed from the signals input to the first microphone 302 and the second microphone 308 as described above, in the SAD method, W₁ 506, W₂ 508, and first Post-processing filter 510, second post-processing filter 514, Decorrelation Criterion 512, first subtractor 408 and second subtractor 410 are required.

W₁ 506 is a variable used to remove the H₂ 504 variable input to the first microphone 302, and W2 508 is a variable used to remove the H₁ 502 variable input to the second microphone 308. This variable is used for

The first subtractor 408 is input with V + H 2 N input from the first microphone 302 and -W 2 (N + H ₁ V) output from the second microphone 308 and modified by the W 2 508. As a result, (1-H₁W₂) V + (H₂- W₂) N is input.

The second subtractor 410 receives N + H + V input from the second microphone 308 and -W₁ (V + H2N) modified by the W₁506 output from the first microphone 302. As a result, (H₁- W₁) V + (1-W₁H₂) N is input.

Here, it is ideal that only the audio signal V is input to the first post-processing filter 510, and only the noise signal N is ideally input to the second post-processing filter 514. Therefore, according to the above equation, in order to remove the noise signal part from the signal (1-H₁W₂) V + (H₂- W₂) N input from the first subtractor 408 to the first post-processing filter 510, H₂ = W₂ In order to remove the audio signal part from the signal (H₁-W₁) V + (1-W₁H₂) N input from the second subtractor 410 to the second post-processing filter 514, H₁ = W₁ It is desirable to be.

Decorrelation Criterion 512 controls the size of W₁ 506 and W₂ 508, and controls H₁ = W₁ and H₂ = W₂ so that voice and noise signals do not affect each other.

When H Decor = W₁ and H₂ = W₂ ideally by the Decorrelation Criterion 512, the signal input to the first post-treatment filter 510 is (1-H₁H₂) V and is input to the second post-treatment filter 514. The signal becomes (1-H₁H₂) N.

의 값을 가지게 된다면, 제 1 후처리 필터(510)를 통과한 신호는 V가 되고, 제 2 후처리 필터(514)를 통과한 신호는 N이 되어, 입력 음성(412)은 잡음이 없는 음성 신호 V가 입력된다.Thus, the first post-process filter 510 and the second post-process filter 514

If the value of, the signal passed through the first post-processing filter 510 is V, the signal passing through the second post-processing filter 514 is N, the input voice 412 is a noise-free voice Signal V is input.

However, this value is based on the calculated value on the time-domain, and in actual implementation, the first post-process filter 510 and the second post-process filter 514 may be modified to have different equations.

6 is a graph showing experimental results showing that noise is removed by the SAD method.

The upper graph of FIG. 6 is a signal mixed with noise input by the first microphone 302, and the lower graph shows a signal input to the mobile communication terminal 112 through the first post-processing filter 510. It is a graph. From the graph, it can be seen that much noise is removed by the SAD method.

7 is a block diagram illustrating a noise removal process using the Infomax method according to a third embodiment of the present invention.

The Infomax method is an algorithm that assumes that the distorted voice and the distorted noise by the environment are input through the first microphone 302, and the distorted noise and the distorted voice by the environment is input through the second microphone 308. It is very similar to the real environment.

A ₁₁ 702 is a variable that is distorted by the environment when the voice 402 is input to the first microphone 302, and A ₁₂ 704 is a variable when the voice 402 is input to the second microphone 308. It is a variable that is distorted by the environment.

A ₂₁ 706 is a variable that is distorted by the environment when noise 404 is input to the first microphone 302, and A ₂₂ 708 is a variable when noise 404 is input to the second microphone 308. It is a variable that is distorted by the environment.

In addition, W ₁₁ 710 used as a variable is a variable used to model and remove A ₁₁ 702, W ₁₂ 712 is A ₁₂ (704), and W ₂₁ 714 is A ₂₁ (706). ), W ₂₂ (716) is a variable used for modeling and removing A ₂₁ (708), respectively.

The signal into which the distorted voice and the distorted noise input to the first microphone 302 are input is referred to as x ₁ (n), and the signal into which the distorted voice and the distorted noise input to the second microphone 308 are input. Specify x ₂ (n).

In addition, signals x ₁ (n) and x ₂ (n) modified by W ₁₁ 710, W ₁₂ 712, W ₂₁ 714, and W ₂₂ 716 are respectively u ₁ (n) and u. ₂ (n), u ₁ (n) is passed through g (x) 722, the signal input to Entropy Maximize (726) is called y ₁ (n), u ₂ (n) is g (x A signal input to Entropy Maximization 726 after passing through 724 is designated as y ₂ (n).

Infomax follows the Central Limit Theorem (CLT), which is a more sophisticated Laplacian using a non-linear function whose form is a mixture of voice and noise, which follows a Gaussian distribution. Laplacian).

은 라플라시안 분포를 따르고, 각각의 합인

는 가우시안 분포를 따른다. The central limit theorem is that the sum of the signals along the Laplacian form adds up to a Gaussian distribution. That is, by the center limit theorem

Follows the Laplacian distribution, and the sum of

Follows a Gaussian distribution.

However, since each noise and voice has a Laplacian distribution, it is necessary to make these signals conform to the Laplacian distribution to separate the mixed signals. The Infomax method using this concept makes each signal close to Laplacian by using entropy input to the first microphone 302 and the second microphone 308 to separate the signals.

In other words, according to the central limit theorem, the input is made in the form of Laplacian's Probability Density Function (PDF) using the Infomax method between the input and the output passing through the nonlinear function g (x) (722, 724). Is an algorithm.

Here, according to Papyuli's theorem, the Infomax algorithm is expressed as I (Y, X) = H (Y)-H (Y | X), and the maximum information I (Y, X) is entropy H (Y). Is equivalent to maximizing Therefore, partial weighting of H (Y) by the weight (Weight) and updating the weight in the direction in which H (Y) is maximized is equivalent to Infomax. This is eventually replaced by maximizing the Jacobian matrix | J |.

The equation for maximizing the entropy by partial derivative of the Jacobian matrix by each weight is as follows.

Therefore, a formula in which the weights of W ₁₁ 710, W ₁₂ 712, W ₂₁ 714, and W ₂₂ 716 are updated is derived as in Equation 2.

The Infomax method is more stable and superior to the SAD method. However, since this method is intended to improve speech recognition rate, data having a large energy in the high frequency region or the low frequency region, which is not annoying to the human ear, becomes more misperceived when end point detection is performed. In addition, residual noise in the results of the two microphones also greatly affects the difficulty and misrecognition of these endpoints.

Wiener filter 718 is a filter that removes noise in order to eliminate a lot of misrecognition that occurs frequently in the Infomax method, to easily detect the end point, and to increase the speech recognition rate. The input voice 412 from which noise is removed through the Wiener filter 718 is input through the ear microphone jack of the mobile communication terminal 112.

FIG. 8 is a graph of experimental results showing the change after the speech signal passes through the Wiener filter 718 in the Infomax method.

The graph at the top is a speech signal where noise is first removed from the signal input by two microphones by the Infomax method. This is represented by u ₁ (n) in FIG. 7. The lower graph is a graph showing the signal after the voice signal u ₁ (n) has passed through the Wiener filter 718. Through this, it can be seen that a large amount of noise is removed by passing through the Wiener filter 718 to increase the speech recognition rate.

In the above description, three algorithms for removing noise using two microphones in a hands-free kit have been described. However, in applying the algorithm according to the present invention, it is possible to modify and apply not only the hands-free kit but also the mobile communication terminal.

9 is a block diagram illustrating a signal processing procedure according to the present invention applied to a mobile communication terminal.

In the mobile communication terminal, a signal processing part includes a microprocessor 902, a digital signal processor 904, a baseband converter 906, a first microphone 908, a second microphone 910, and an RF signal processor. 912 and antenna 914.

The microprocessor 902 controls the mobile communication terminal as a whole. When the voice signal and the noise signal are input to the digital signal processor 904, the microprocessor 902 instructs the digital signal processor 904 to drive a noise removal algorithm. It serves to manage the entire system of the mobile communication terminal, such as commanding to transmit the finished voice signal to the baseband converter 906.

The digital signal processor 904 is a digital signal processor that receives the digital voice signal and the digital noise signal transmitted through the baseband converter 906 and removes the noise through an algorithm according to the present invention.

The baseband converter 906 converts signals transmitted and received between the RF signal processor 912 and the digital signal processor 904, the first microphone 908, and the second microphone 910 into baseband signals. It performs functions such as digital-to-analog conversion and analog-to-digital conversion processing.

Accordingly, the baseband converter 906 converts the noise and voice signals input in the analog format through the first microphone 908 into digital signals and transmits the digital signals to the digital signal processor 904. In addition, the noise signal input in the analog format through the second microphone 910 is converted into a digital signal and transmitted to the digital signal processor 904.

In addition, the digital signal processor 904 receives the digital signal processing signal from which the noise is removed through the algorithm of the present invention, and transmits the digital signal to the RF signal processor 912.

The first microphone 908 converts the user's voice signal and the surrounding noise signal into an electrical signal and transmits the electrical signal to the baseband converter 906.

The second microphone 910 is installed at the rear or top of the mobile communication terminal, receives only a noise signal, converts the noise signal into an electrical signal, and transmits it to the baseband converter 906.

The RF signal processor 912 modulates the transmission signal applied from the baseband converter 906 and outputs the modulated signal to the antenna 914, and the antenna 914 transmits a radio signal to the mobile communication network.

Through such a system, it is also possible to apply a method for removing noise using two microphones according to the present invention to a mobile communication terminal.

The above description is merely illustrative of the technical idea of the present invention, and those skilled in the art to which the present invention pertains may make various modifications and changes without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of the present invention.

As described above, according to the present invention, when the user uses the telematics service by voice while driving, the user's voice is incorrectly recognized due to noise and the wrong information is provided. Even when making a voice call, it is possible to improve the quality of the call by delivering a clean voice with no noise.

In addition, a telematics service or a voice call is misrecognized, thereby reducing annoyance of a driving user, and a user using a mobile communication terminal can concentrate on driving, thereby reducing the incidence of an accident.

Claims (41)

In the hands-free kit to remove the noise input to the mobile phone using two microphones, (a) inputting a voice signal and a noise signal to the first microphone and the second microphone; (b) adding a signal input to the first microphone and a signal input to the second microphone; (c) subtracting a signal input to the second microphone from a signal input to the first microphone; (d) subtracting the signal output as a result of step (c) from the signal output as a result of step (b); And (e) using the resultant signal output from step (d) as an input voice to the mobile phone Method for removing noise of the signal input to the mobile phone using two microphones, characterized in that it comprises a. The method of claim 1, wherein step (b) comprises: And the voice signal of the user input to the first microphone and the second microphone, and the noise signal are mixed. The method of claim 1, wherein step (c) comprises: By removing the signal input to the second microphone from the signal input to the first microphone to remove the noise of the signal input to the mobile phone using the two microphones, characterized in that for obtaining the two microphone input noise value Way. The method of claim 1, wherein step (d) Extracting only the voice signals input to the two microphones by subtracting the noise value obtained in the step (c) from the voice signal obtained in the step (b) and the mixed signal of the noise. To remove noise from signals coming into your phone using two microphones. The method of claim 1, wherein in step (e), The output input voice is connected to a mobile communication network through a mobile communication terminal to make a call with another person or to a telematics service center for a telematics service. How to remove it. The method of claim 1, wherein the first microphone and the second microphone Method for removing noise of the signal input to the mobile phone using two microphones, characterized in that installed at least 4 cm apart to facilitate the extraction of the input voice. The method of claim 1, wherein the second microphone, Method for removing noise of the signal input to the mobile phone using two microphones, characterized in that installed on the side, top, bottom, or back of the hands-free kit. The method of claim 1, wherein the hands-free kit, In addition to the first microphone and the second microphone for receiving the voice signal and the noise signal, an analog-to-digital converter (ADC), a digital signal processor (DSP), a digital-to-analog converter (DAC), A method for removing noise in a signal input to a mobile phone using two microphones, characterized by ROM, RAM and switching circuitry. The method of claim 8, wherein the ADC, By using the two microphones, the first microphone and the second microphone input the audio signal and the noise signal of the analog format are sampled at predetermined time intervals, and then quantized and converted into digital signals. How to remove noise from the signal input to the mobile phone. The method of claim 8, wherein the ROM, And storing the noise reduction algorithm in which the steps (b) to (e) are executed in the DSP. 2. The method of claim 8, wherein the DAC, A method of removing noise of a signal input to a mobile phone using two microphones, wherein the DSP converts the digital signal from which the noise is removed to an analog format so that the digital signal can be input through an ear microphone jack of a mobile communication terminal . In the hands-free kit to remove the noise input to the mobile phone using two microphones, (a) inputting a voice and a distorted noise signal into the first microphone; (b) inputting a noise and a distorted voice signal to the second microphone; subtracting a signal multiplied by a correction variable from a signal input to the second microphone from a signal input to the first microphone; (d) inputting the signal calculated in step (c) into a post-processing filter; And (e) using the signal passed through the post-processing filter as an input voice to the mobile phone, wherein the correction parameter multiplied by the signal input to the second microphone is controlled by a Decorrelation Criterion to determine W ₂ . It means that it is used as a value, wherein step (d) means that the calculated value (1-H₁H₂) V calculated in the step (c) is input to the post-treatment filter, The post-processing filter is such that the signal passing through the post-processing filter has only V, which is the voice signal value.

A method of removing noise from a signal input to a mobile phone using two microphones, characterized in that having a value of. The method of claim 12, wherein step (a) comprises: Noise of a signal input to a mobile phone using two microphones is input to the first microphone, wherein the voice signal V inputted and the H ₂ N multiplied by the noise signal N by the distortion variable H ₂ due to the environment are inputted to the first microphone. How to remove it. The method of claim 12, wherein step (b) comprises: The noise of a signal input to a mobile phone using two microphones is input to the second microphone, wherein the input noise N and the H ₁ V multiplied by the environment distortion signal H ₁ are multiplied by the second microphone. How to remove. delete The method of claim 12, wherein step (c) comprises: In conclusion, (1-H₁W₂) V + (H₂- W₂) N by subtracting W₂ (N + H₁V) multiplied by the correction variable from the signal V + H₂N input to the first microphone. A method for removing noise in a signal input to a mobile phone using two microphones, characterized in that it has a calculated value of. The method of claim 16, wherein W₂, To remove the noise of the signal input to the mobile phone using the two microphones characterized in that it is controlled to have a value of H₂ by the Decorrelation Criterion to remove the noise part of the calculated value, that is (H₂-W₂) N Way. delete delete The method of claim 12, wherein in step (e), The output input voice is connected to a mobile communication network through a mobile communication terminal to make a call with another person or to a telematics service center for a telematics service. How to remove it. The method of claim 12, wherein the hands-free kit, In addition to the first microphone and the second microphone for receiving the voice signal and the noise signal, an analog-to-digital converter (ADC), a digital signal processor (DSP), a digital-to-analog converter (DAC), A method for removing noise in a signal input to a mobile phone using two microphones, characterized by ROM, RAM and switching circuitry. The method of claim 21, wherein the ADC, The first microphone and the second microphone input the audio signal and the noise signal of the analog format is sampled at a predetermined time interval and then quantized (Quantization) is converted into a digital signal using two microphones, characterized in that How to remove the noise of the signal input to the. The method of claim 21, wherein the ROM, And storing the noise reduction algorithm in which the steps (c) to (e) are executed in the DSP. 2. The method of claim 21, wherein the DAC, A method of removing noise of a signal input to a mobile phone using two microphones, wherein the DSP converts the digital signal from which the noise is removed to an analog format so that the digital signal can be input through an ear microphone jack of a mobile communication terminal . In the hands-free kit to remove the noise input to the mobile phone using two microphones, (a) inputting a first distorted voice signal and a first distorted noise signal to the first microphone; (b) inputting a second distorted noise signal and a second distorted voice signal to the second microphone; (c) correcting the distorted portion of the signal input in the above steps (a) and (b) by the correction variable to extract only the voice signal portion; (d) inputting the signal extracted in step (c) to a Wiener filter; And (e) using the signal passing through the Wiener filter as an input voice to the mobile phone, wherein the correction variable is used to remove the distortion variables A ₁₁ , A ₁₂ , A ₂₁ , A ₂₂ . Mean, Step (c) is characterized in that the noise signal and the speech signal calculated by the correction parameter value have a form of Laplacian Distribution by a Central Limit Theorem (CLT). To remove noise from signals coming into your phone using two microphones. The method of claim 25, wherein step (a) comprises: By using two microphones, a first distorted speech signal modified by the distortion variable A ₁₁ and a first distorted noise signal modified by the distortion variable A ₂₁ are input to the first microphone. How to remove noise from the signal input to the mobile phone. The method of claim 25, wherein step (b) By using two microphones, a second distorted speech signal modified by the distortion variable A ₁₂ and a second distorted noise signal modified by the distortion variable A ₂₂ are input to the second microphone. How to remove noise from the signal input to the mobile phone. delete delete The method of claim 25, wherein step (d) Remove noise of the signal input to the mobile phone by using two microphones, characterized in that passing through the Wiener filter in order to eliminate the misrecognition generated when detecting the end point of the extracted signal calculated in step (c) and increase the speech recognition rate How to. The process of claim 25, wherein in step (e), The output input voice is connected to a mobile communication network through a mobile communication terminal to make a call with another person or to a telematics service center for a telematics service. How to remove it. The method of claim 25, wherein the hands-free kit, In addition to the first microphone and the second microphone for receiving the voice signal and the noise signal, an analog-to-digital converter (ADC), a digital signal processor (DSP), a digital-to-analog converter (DAC), A method for removing noise in a signal input to a mobile phone using two microphones, characterized by ROM, RAM and switching circuitry. The method of claim 32, wherein the ADC, Sampling the first distorted voice signal, the second distorted voice signal and the first distorted noise signal, and the second distorted noise signal of the first microphone and the second micro-input analog format at predetermined time intervals. After the quantization (Quantization) is converted into a digital signal using two microphones to remove the noise of the signal input to the mobile phone. 33. The method of claim 32, wherein the ROM is And storing the noise reduction algorithm in which the steps (c) to (e) are executed in the DSP. 2. The method of claim 32, wherein the DAC, A method of removing noise of a signal input to a mobile phone using two microphones, wherein the DSP converts the digital signal from which the noise is removed to an analog format so that the digital signal can be input through an ear microphone jack of a mobile communication terminal . delete delete delete delete delete delete

Images