TWI458361B - System, method and apparatus with environment noise cancellation - Google Patents

Description

System, method and application device for reducing environmental noise

The present invention is a system, method and application for reducing environmental noise, and particularly to a microphone array, a system for providing better call quality and an application device through an instant noise suppression program.

In order to solve the problem caused by the environmental noise during the call, the prior art proposes a technology for reducing the environmental noise by using a dual microphone array. The principle is to set a main microphone for receiving voice and nearby noise, and then set another position outside the distance. A secondary microphone that receives ambient noise. The signals received by the two microphones can effectively eliminate ambient noise and improve call quality.

Referring to the first figure, a conventional communication device having two microphones is shown, such as U.S. Patent No. 6,549,586. The communication device shown therein has two microphones, a first microphone 101 remote from the mouth and a second microphone 102 near the mouth. Because the first microphone 101 is far away from the mouth, its main job is to collect background noise, but it may also receive the call voice; and the second microphone 102 mainly collects the call voice, and the difference between the two can be used as noise suppression.

In this example, in order to suppress the main operation, the call voice received by the first microphone 101 that collects the background noise, after the signal passes through the first temporary storage memory 103, the first subtraction circuit 105 processes the reduced call voice to strengthen the background. Estimation of noise. In contrast, the voice of the call collected by the second microphone 102 and part of the background noise are temporarily stored in the second temporary memory 104, wherein the second subtraction circuit 106 first refers to the delay circuit by the first subtraction circuit 105. 107 provides the background noise estimated previously (previous time), so the second subtraction circuit 106 can enhance the suppression of the background noise signal collected by the second microphone 102.

Then, the third subtraction circuit 108 receives the background noise estimated by the first subtraction circuit 105 and the voice signal estimated by the second subtraction circuit 106, and after the parameter adjustment, the voice signal processed by the noise suppression process can be obtained. Then, the signal is output to an inverse fast Fourier transform (IFFT) 109, and the discrete time signal is converted into a continuous frequency domain signal, and then the superimposed processor 110 combines the signals to output a voice signal.

Microphone array according to a first conceptual diagram of a general use after more conventional techniques such as Fortemedia U.S. Patent No. 7,587,056 ^TM Company proposed and disclosed a method of inhibiting noise microphone array, wherein further more detailed process flow proposed, including signal Digital processing schemes such as calibration, beamforming, noise estimation and suppression, and time domain-frequency domain conversion for better call quality.

However, there are still some shortcomings in the prior art, such as:

1. Due to the lack of effective adjustment mechanism, the quality of the microphone is relatively high;

2. Using a fixed beamforming circuit to extract the voice signal will require the gain of the microphone to match;

3. Because it is a fixed beamforming technology, the voice will be mixed with more noise, affecting the function of noise reduction; if you want to further deal with noise reduction, it may cause speech distortion.

In order to obtain better call quality, the present invention particularly utilizes instant signal tuning, adaptive beamforming techniques and non-linear noise suppression programs, in addition to eliminating microphone hardware differences or setting position errors, To minimize noise and improve noise suppression performance.

According to an embodiment, the system for reducing ambient noise is mainly applied to a radio module having two or more input terminals, and the input end of the radio module is specifically designed to mainly receive main audio such as voice or specific audio. For example, the device that is implemented by the present invention is an example. The device has a microphone array, and the microphone array includes at least one first microphone module and a second microphone module.

The signals received by the microphone modules are transmitted to the system, and the signals are transmitted to the calibration unit. By adjusting the sensitivity of each microphone module to collect sound, the error caused by the difference between the microphone modules can be reduced. Then, the adaptive beamforming technique can be used to adjust the signal, and the threshold is relatively small, and the signal of the main audio portion is relatively small. The speech is then extracted using a speech extraction unit to derive a signal dominated by the main audio.

Then, the main audio-based signal and the ambient noise-based signal are simultaneously converted from the time domain to the frequency domain, and then the noise suppression unit performs nonlinear noise suppression to generate a noise reduction gain for suppressing noise. Can effectively reduce noise.

In the process of voice extraction and noise suppression, the above information can be fed back to the front end of the system, so that the information of the signal can be referred to when the signal is adjusted, including whether the audio is part of the voice or only ambient noise.

Then, the inverse frequency domain conversion unit performs adjustment using the noise reduction gain, and converts the signal into the time domain, and then uses a process such as superposition and addition to form a continuous output audio.

According to an embodiment, the noise suppression method of the above-mentioned microphone array mainly receives the audio collected by the microphone array, and the system determines whether the main audio component or the ambient noise component determines the gain according to the previous signal, and can be used for tuning. The school is currently audio.

The gain-matched signal then performs a beamforming process that uses a preset threshold to adjust the filtering effect to effectively derive the portion of the ambient noise.

The calibrated main audio portion is compared with the ambient noise portion signal, and the effect of the filtering is adjusted by another preset threshold value, which can be captured in the main audio portion.

After being converted to the frequency domain, the signal smoothing operation and the decimation operation are selectively performed, adjusted to an appropriate signal resolution, and the degree of environmental noise is estimated from the two sets of signals in the frequency domain by a nonlinear noise suppression operation. Then, the gain for adjusting the noise reduction is obtained. Finally, this noise reduction gain is used to perform noise reduction on the continuous audio in the frequency domain.

According to an embodiment of the present invention, in an embodiment, the application is mainly applied to a radio module having two or more input terminals, wherein the input terminal may be two or more microphone modules, and the radio module is The microphone array formed by the two microphone modules is designed to collect the sound of different positions through software or hardware, and can estimate the background noise, thereby obtaining a better quality audio or voice signal. .

The technique implemented in accordance with the present invention has at least several advantages:

1. It can effectively suppress the environmental noise during the call, and improve the clarity and comfort during the audio or voice communication process, so as not to be affected by too much environmental noise;

2. The system for reducing ambient noise proposed by the present invention can perform immediate signal calibration, and experiments can tolerate a gain difference of ±6 dB (decibel) for each microphone module;

3. Introducing adaptive beamforming technology in this system to extract general audio, including voice signals, to minimize the noise in the audio, and to improve the nonlinear noise suppression at the back end (non-linear Noise suppress) The performance of the relevant module.

However, the reduced ambient noise proposed by the present invention can be applied to a device having two sound pickup modules in addition to a specific microphone system, and can still be extended to an array formed by a plurality of microphones, not in the embodiment described herein. Limited.

The embodiment can refer to the second figure.

The figure shows a modular functional block diagram of the system for reducing ambient noise of the present invention. The embodiment uses a two-input radio module for receiving a main audio-based signal and an environmental noise-based signal. The input end may be a microphone array formed by the first microphone module 201 and the second microphone module 202.

In this example, the first microphone module 201 is specifically designed to collect a voice signal or a specific audio to be received. If it is used in a communication device, it can be disposed near the mouth; the second microphone module 202 is It is designed to collect ambient noise. If it is used on a communication device, it can be placed at a distance from the first microphone module 201 to reduce the proportion of collected voice signals or specific audio.

The signals received by the microphone modules are transmitted to the inside of the system. The system proposed by the present invention can be implemented by an integrated circuit (IC) or by software means. The system mainly includes a calibration unit 203 that can be connected to the audio received by the real-time online calibration radio module, and a beamforming unit 204 that can adjust the received energy according to actual needs, and is used for A speech extractor 205 that captures a main audio portion, a frequency domain conversion unit that performs time domain/frequency domain signal conversion (such as a variable frequency resolution transformer, VFRT) 206, and an executable non- The linear noise suppression noise suppression unit 207, the inverse frequency domain resolution transformer (such as the inverse variable frequency resolution transformer) 208 performing the frequency domain/time domain signal conversion, and the overlap of the execution signals (overlap-add) Superposition unit 209 of -sum).

In operation, the input end of the radio module, as shown in the figure, the first microphone module 201 can be disposed at a position closer to the main audio source, and the main collected sound is the audio generated by the audio source; the input end is like the second. The microphone module 202 can be disposed at a position far from the source of the audio, and is mainly used to collect ambient noise and possibly collect audio (including voice). The signals generated by each of the microphone modules 201, 202 are denoted as M1 and M2, respectively.

The signals M1 and M2 are respectively transmitted to the calibration unit 203, and the calibration unit 203 is coupled to the sound collection module, which is actually implemented as a microphone array, mainly based on information of the main audio or environmental noise received in the system (including extraction by voice). The unit generates information SF1 for determining whether the current audio is the main audio portion, and the noise suppression unit generates information for determining whether the current signal is ambient noise. NF1) Adjusting the sensitivity of each microphone module to collect sound, which can reduce each The error caused by the difference between the microphone modules (such as the difference in hardware design, the error caused by the manufacturing process, the difference in the circuit, etc.), the adjustment at the input can ensure the quality of the signal afterwards. The above information of the main audio or ambient noise is the signal information (NF1 and SF1) judged by the back-end components in the system.

After being processed by the calibration unit 203, a signal S1 (main audio portion) and an S2 (ambient audio portion) are respectively generated and transmitted to a beamforming unit 204 coupled to the calibration unit 203. The direction or angle of the sound received by each microphone module represents the received signal energy. In order to obtain a proper receiving energy, in addition to adjusting the microphone angle, the beam forming technology can be used for the microphone formed by the plurality of microphone modules. The sound waves collected by the array interfere with each other and produce an interference pattern, which is adjusted to an appropriate interference image according to design requirements.

After the signal is processed by the beam forming unit 204, a voice signal or a relatively small signal R1 of a specific audio (main audio portion) can be generated. According to the illustration, the signal S1 representing the main audio portion is simultaneously with the signal R1 having few main audio signals. Transfer to the voice extraction unit 205. The voice extraction unit 205 mainly performs a filtering means. After comparing the signals S1 and R1, the output signal SF1 is fed back to the calibration unit 203 as a calibration reference for the front and rear signals, and whether the voice signal mainly has the voice signal to be collected. Or specific audio (such as SF1 = 0 means that the audio is mainly ambient noise; SF1 = 1 means that there is a voice signal or a specific audio to be collected), and the voice extracted signal A1 is output.

The frequency domain converting unit 206 is coupled to the voice extracting unit 205, and receives the relatively small signal R1 of the main audio component and the voice extracted signal A1, that is, the signal of the main audio component and the ambient noise component respectively. Performing a time domain-frequency domain conversion procedure to convert the signal from a time domain to a frequency domain. The main embodiment is to generate a frequency domain signal P1 by using a Fast Fourier Transformation operation. P2.

Next, the noise suppression unit 207 receives the frequency domain signals P1 and P2 from the frequency domain conversion unit 206, thereby estimating the environmental noise and generating a gain for reducing the noise, that is, the signal G1. A signal NF1 for expressing whether the current signal is noise is generated, and is fed back to the calibration unit 203 as a calibration reference for the front and rear signals.

The inverse frequency domain conversion unit 208 receives the gain signal G1 and the frequency domain signal P1, and uses the gain signal G1 as a basis for the interpolation method, and performs inverse fast Fourier Transformation with the signal P1 to convert the frequency domain signal into the time domain. In the middle, the time domain signal SO1 is generated.

Finally, the signals SO1 of the respective periods will be summed by the superimposing unit 209 to form a continuously outputted audio.

The details of the operation of each of the above module units can be followed by the following diagrams and flows, wherein each module can be implemented by software or hardware.

The third figure shows a working flow chart of one of the embodiments of the calibration unit 203 in the system for reducing environmental noise according to the present invention. The system first receives the signal M1 generated by the first microphone module 201 and the second microphone module 202. M2, in this embodiment, M1 is a signal mainly for the main audio portion, and usually includes both a voice signal and ambient noise, and M2 is mainly a signal for ambient noise, but still includes some voice signals.

First, in step S301, the signal SF1 is the signal generated by the voice extraction unit 205. The signal is extracted by the voice extraction process, and the signal content contained in the signal is expressed by the signal SF1. If SF1=0, the audio is mainly ambient noise (No). The step will go directly to step S310; if SF1=1, it means that the signal has a voice signal to be collected or a specific audio (Yes), the message will be brought to step S303 as a calculation reference, and brought to step S304 as the cumulative number of times. Reference.

In contrast, in step S302, the signal NF1 generates a feedback signal for the noise suppression unit 207 to determine whether the signal is ambient noise, such as NF1=0 (No), that is, determining the current signal as the main audio portion. The integration is performed to S310; if NF1=1 (Yes), the message is brought to step S303 to calculate the power as a reference, or to step S305 as a reference for accumulating the number of environmental noises.

In step S303, the powers (energy) of the signals M1 and M2 are calculated, and the energies Po1 and Po2 are respectively generated, and the energies Po1 and Po2 will be used as references for performing the signal judging steps in steps S304 and S305, respectively.

In step S304, if the received signal is a voice signal or a specific audio, the number of times (Cnt1) will be accumulated. If the accumulated number of times has not exceeded a threshold (Cnt1<Th1) (No), step S310 is performed first; if the accumulated number of times exceeds If the threshold is (yes), the gain (Gain1) is calculated in step S308.

The main audio portions of the signals Po1 and Po2 are accumulated by energy (step S304), and are recorded in the signals SS1 and SS2. When it is determined in step S306 that the number of times of the main audio portion (Cnt1) has exceeded a certain threshold (Cnt1>=Th1), the accumulated signals SS1 and SS2 are calculated in step S308 to obtain a gain Gain1.

On the other hand, the flow on the right side of the diagram will process the part of the ambient noise, and the calculated power values Po1 and Po2 will be combined with the information carried by the signal NF1 (NF1=0 means non-environmental noise; NF1=1 means The ambient noise is accumulated in step S305 as the number of environmental noise signals (Cnt2), and the ambient noise power (energy) is accumulated, and the message is carried on signals SN1 and SN2.

When the accumulated number of ambient noise reaches a threshold (Cnt2>=Th2), the step proceeds to S309, and the gain (Gain2) is calculated from the information carried by the signals SN1 and SN2, and the gain Gain2 is generated; if the accumulated number of environmental noise signals has not been reached Threshold (Cnt2<Th2), the steps will be processed directly to S310.

Step S310 is to fuse the gains Gain1 and Gain2, and integrate the information of the main audio part or the environmental noise part of the signal SF1 and NF1 to obtain the gain Gain. There are many ways to determine the gain Gain. One of them is: since the audio continuously enters the system, sometimes step S310 only obtains the information of Gain1, sometimes only Gain2 information, if there is no Gain1 and Gain2, then The gain Gain is 1. In addition to the procedures and judgments described in the third figure, the calculation of the gains Gain, Gain1, Gain2 is a general technique and can be derived by those skilled in the art.

Finally, in step S311, the gain Gain is applied to the signal M1 generated by the first microphone module and the signal M2 generated by the second microphone module, and the gain-adjusted signals S1 and S2 are respectively output.

The fourth figure then shows the modular function block diagram of the beam forming unit 204 in the system of the present invention. The unit blocks shown in the figure can be achieved by software means or can be implemented by a hardware circuit.

The beam forming unit 204 shown in the figure receives the gain-adjusted signals S1 and S2, and first calculates the signal power (energy) through the power calculating unit 401, and generates signals PS1 and PS2. Then, the main audio component in the signal is detected by the voice detecting unit 403, including a voice part, a specific audio, and the like. According to an embodiment, the voice detection unit 403 may first determine whether the energy difference between the PS1 and the PS2 is greater than a preset threshold (first preset threshold), and according to the threshold, determine the parameter V1, and the parameter V1 controls the filtering in the filtering unit 405. Filter coefficient. The signal S1 is delayed by a delay unit 407 and the signal S2 directly entering the filtering unit 405, and the signal R1 with less voice signal is generated by the filtering means.

Then, referring to the fifth figure, a modular function block diagram of the voice extraction unit 205 in the system is shown. The voice extraction unit 205 can also be implemented by a software means, or can be implemented by a hardware circuit.

The voice extraction unit 205 receives the signal R1 (mainly ambient noise) with less voice signal and the previously adjusted signal S1 (mainly the main audio portion), and first calculates the individual power through the power calculation unit 501. The signal PS1 (which has been generated by the power calculation unit 401 in the beam forming unit 204) and PR1, through the voice detecting unit 503, determine whether the difference between the two energies is greater than another preset threshold (second preset threshold), according to The parameter V2 is generated for controlling the filter coefficients in the filtering unit 505 to produce an adaptive filtering effect. In the figure, the filtering unit 505 simultaneously receives the signal S1 and the signal R1 delayed by the delay unit 509, thereby generating a signal having a main audio portion, that is, an output signal A1 mainly for a voice signal and a specific audio.

The speech extracting unit 205 has a speech confirming unit 507, and the speech confirming unit 507 retrieves the signals PS1 and PR1, and determines whether the signal passed at this time is a voice signal or a specific audio, and if so, the signal can be set. SF1=1; otherwise, the setting signal SF1=0. The signal SF1 will be fed back to the calibration unit 203 as a reference for tuning the microphone signal.

Then, referring to the modular function block diagram of the frequency domain conversion unit 206 in the system shown in FIG. 6, the frequency domain converting unit 206 receives the signal R1 with less signal A1 and the voice signal, which is converted from the time domain to the frequency domain. Software means or hardware circuit. In particular, the signals A1 and R1 are fast Fourier transformed by the Fourier transform units 601 and 603, respectively, and the generated frequency domain signals are FA1 and FR1, and when the signal is converted, the calculation amount can be reduced by a sampling mechanism. The frequency domain signals are FA1 and FR1, and then can continue to perform smoothing and decimation operations through the smoothing and decimation unit 602, 604, respectively, and can delete the interference signal and operate less signal reduction without distortion. Cost, can optimize the signal processing process. However, this procedure is optional and not necessary. Finally, signals P1 and P2 are generated separately.

The frequency-domain converted signals P1 and P2 are transmitted to the noise suppression unit 207. Referring to the modular function block diagram of the noise suppression unit 207 in the system shown in FIG.

The noise suppression unit 207 can also be implemented by a software means, or can be implemented by a hardware circuit. In the embodiment of the present invention, the noise suppression means of the latter stage can be ignored.

The noise estimating unit 701 mainly performs non-linear noise suppression, can estimate the environmental noise according to the signals P1 and P2, and calculates the gain G0 for signal adjustment, and simultaneously generates the signal NF1, that is, the input. The reference signal in the calibration unit 203 is used to indicate whether the signal is mainly ambient noise. For example, if it is ambient noise, NF1=1 can be set; if it is the main audio part, NF1=0. The generated gain G0 can be processed by the gain correcting unit 703 to output a gain G1 for noise reduction.

The gain G1 is then passed to the inverse frequency domain converting unit 208, and the inverse frequency domain converting unit 208 receives the signal P1 carrying the main audio portion, and adjusts according to the gain G1 to achieve the purpose of noise reduction. The internal implementation of the inverse frequency domain conversion unit 208 can refer to the modular function block diagram shown in the eighth figure.

The gain G1 generated after the nonlinear noise suppression process can effectively suppress the noise of the main audio portion, and the gain signal G1 is first adjusted by the interpolation unit 801 to return the gain IG1 in the time domain, corresponding to the signal P1 point by point. Multiply, the frequency domain signal GP1 is generated, and finally the signal from the time domain is converted back to the time domain by the inverse Fourier transform unit 803, and the signal SO1 is output.

The superimposing unit 209 is coupled to the inverse frequency domain converting unit 208, and receives the signal SO1 outputted by the signal. The signal is represented by a waveform in the time domain, and the superimposing unit 209 superimposes, adds, and sums the sound waves. Operations such as (summing) form a continuous audio output.

Through the above circuit modules, the method applied by the present invention is summarized in the ninth figure, which is a flow chart of a method for reducing environmental noise performed by the system for reducing ambient noise proposed by the present invention.

According to an embodiment of the invention, the above functional blocks can be executed by software means, and the program can be programmed into an embedded chip or can be loaded into the memory of the processor in the system.

The microphone array has at least one first microphone module mainly receiving the main audio portion and another second microphone module mainly receiving the ambient noise portion, especially applied to the communication device, which can effectively suppress the environmental noise and improve the call. quality. The process of reducing ambient noise performed by the system of the present invention is shown in FIG.

After the audio is collected by the sound receiving module (such as the microphone array) (step S901), at least two sets of signals are respectively adjusted to reduce errors caused by the difference in design of the microphone, including performing gain matching. Receiving information of a main audio and an environmental audio by using the calibration unit mainly includes: determining, by the system, whether the main audio portion or the environmental noise portion has information according to the previous signal (step S903), thereby determining a gain value, and applying the The gain value adjusts the current audio, that is, the audio and ambient noise portion of the main audio portion received by the microphone array (step S905). This adjustment process is an ongoing process, so you can grasp the status of the microphone and the environment at any time, and provide better call quality.

The gain-matched signal is followed by a beamforming process, which is mainly adjusted for the state of the microphones of each microphone module, such as determining whether the difference between the audio received by the two microphone modules exceeds a preset threshold value. The effect of the filtering is to effectively obtain the portion of the environmental noise (step S907).

Next, in step S905, the method step S907 uses the above-mentioned ambient noise portion (the relatively small signal of the main audio portion) to compare the signals of the main audio portion obtained by the first microphone module and adjusted. The difference is also compared with another preset threshold value, which is used to adjust the filtering effect and can be captured in the main audio portion (step S909).

The ambient noise portion and the main audio portion are respectively obtained through the above steps S905 and S907, and then the time domain-frequency domain conversion is performed (step S911), for example, the signal is converted into the frequency domain in the time domain by using a fast Fourier transform program. The signal can be selectively subjected to signal smoothing and decimation operations, adjusted to the appropriate signal resolution, and finally the overlay program is used to restore the signal, and the appropriate saved computing resources are used to generate good call quality.

After the time domain-frequency domain conversion, the degree of environmental noise is estimated by the two sets of signals in the frequency domain by a nonlinear noise suppression operation (step S913), and then the noise reduction gain for adjusting the noise reduction is obtained ( Step S915).

Finally, the noise reduction is performed on the continuous audio to perform noise reduction in the frequency domain (step S917), and then converted into a time domain signal, such as applying an inverse fast Fourier transform (step S919), and finally after the signal overlap and the summation process. Output (step S921).

The above described system for reducing ambient noise and its method are particularly applicable to devices having two inputs.

In summary, the system for reducing ambient noise disclosed in the present invention, after signal tuning, beamforming, voice extraction, frequency domain/time domain conversion, noise suppression and superposition, is output to each microphone in the microphone array. The signal is processed on the fly, and the gain is changed according to the situation at any time. The ambient noise during the call can be effectively suppressed, and the clarity and comfort during the audio or voice communication process can be improved, and the flexibility of the microphone can be selected.

However, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Therefore, equivalent structural changes that are made by using the specification and the contents of the present invention are equally included in the present invention. Within the scope, it is combined with Chen Ming.

101. . . First microphone

102. . . Second microphone

103. . . First temporary memory

104. . . Second temporary memory

105. . . First subtraction circuit

106. . . Second subtraction circuit

107. . . Delay circuit

108. . . Third subtraction circuit

109. . . Inverse Fourier transform circuit

110. . . Superimposed processor

M1, M2, S1, S2, R1, A1, SF1, P1, P2, NF1, G1, SO1, Po1, Po2, SS1, SS2, SN1, SN2, PS1, PS2, PR1, A1, FA1, FR1, GP1. . . Signal

V1, V2. . . parameter

G0, G1, IG1. . . Gain

201. . . First microphone module

202. . . Second microphone module

203. . . Tuning unit

204. . . Beam forming unit

205. . . Voice extraction unit

206. . . Frequency domain conversion unit

207. . . Noise suppression unit

208. . . Anti-frequency domain conversion unit

209. . . Superposition unit

Gain, Gain1, Gain2. . . Gain

401,501. . . Power calculation unit

403,503. . . Speech detection unit

405,505. . . Filter unit

407,509. . . Delay unit

501. . . Power calculation unit

507. . . Voice confirmation unit

601,603. . . Fourier transform unit

602,604. . . Smoothing and extraction unit

701. . . Noise estimation unit

703. . . Gain correction unit

801. . . Interpolation unit

803. . . Inverse Fourier transform unit

Step S301~S311 Workflow of the calibration unit

Step S901~S921 Noise Reduction Process

The first figure shows a block diagram of a conventional technology dual microphone communication device circuit;

The second figure shows a block diagram of the modular function of the system for reducing ambient noise of the present invention;

The third figure shows a working flow chart of an embodiment of the calibration unit in the system of the present invention;

The fourth figure shows a block diagram of the modular function of the beam forming unit in the system of the present invention;

The fifth figure shows a block diagram of the modular function of the voice extraction unit in the system of the present invention;

Figure 6 is a block diagram showing the modular function of the frequency domain conversion unit in the system of the present invention;

Figure 7 is a block diagram showing the modular function of the noise suppression unit in the system of the present invention;

Figure 8 is a block diagram showing the modular function of the inverse frequency domain conversion unit in the system of the present invention;

The ninth diagram shows a method of reducing ambient noise performed by applying the system of the present invention.

M1, M2, S1, S2, R1, A1, SF1, P1, P2, NF1, G1, SO1. . . Signal

201. . . First microphone module

202. . . Second microphone module

203. . . Tuning unit

204. . . Beam forming unit

205. . . Voice extraction unit

206. . . Frequency domain conversion unit

207. . . Noise suppression unit

208. . . Anti-frequency domain conversion unit

209. . . Superposition unit

Claims (12)

A system for reducing ambient noise, comprising: a calibration unit coupled to a microphone array, wherein the microphone array receives a signal mainly based on the main audio and a signal mainly based on ambient noise, and according to the received one The feedback information of the main audio adjusts the sensitivity of a microphone module in the microphone array; a beam forming unit is coupled to the calibration unit to receive the signal mainly based on the adjusted main audio signal and the ambient noise. After processing, a relatively small signal of the main audio portion is generated; a voice extraction unit is coupled to the beam forming unit, and receives a relatively small signal of the main audio portion, and the signal mainly composed of the adjusted main audio Generating the feedback information of the main audio, and performing a filtering means to output a voice extracted signal; a frequency domain converting unit coupled to the voice extracting unit to receive the relatively small signal and the passing of the main audio portion The voice extracted signal performs a time domain-frequency domain conversion process by using a fast Fourier transform; a noise suppression unit is coupled to the frequency domain conversion unit, Performing a nonlinear noise suppression program, receiving a time domain-frequency domain converted signal, and calculating a gain for noise reduction; and an inverse frequency domain conversion unit coupled to the noise suppression unit for utilizing the noise reduction unit The gain performs noise reduction, and performs a frequency domain-time domain conversion by using an inverse fast Fourier transform; and a superposition unit coupled to the inverse frequency domain conversion unit, which is formed by overlapping, adding, and signal summing operations to form a continuous Audio. The system for reducing environmental noise as described in claim 1 of the patent scope, The voice extraction unit has a voice confirmation unit for generating feedback information of the primary audio. The system for reducing ambient noise according to claim 1, wherein the beam forming unit has a filtering unit that determines a filtering coefficient of the filtering unit by using a first preset threshold, thereby generating a relative portion of the main audio portion. Very few signals. The system for reducing ambient noise as described in claim 3, wherein the voice extraction unit has another filtering unit that determines a filter coefficient by using a second preset threshold to generate a voice extracted signal. The system for reducing ambient noise according to claim 1, wherein the calibration unit further adjusts the sensitivity of the microphone module in the microphone array according to the received feedback information of an environmental noise. The system for reducing ambient noise as described in claim 5, wherein the feedback information of the ambient noise is generated by the noise suppression unit when calculating a noise reduction gain. A device having two sound pickup modules, such as the system for reducing environmental noise according to claim 1 of the patent application. A method for reducing ambient noise, the method comprising: receiving audio of an audio and environmental noise portion of a main audio portion received by a microphone array; receiving feedback information of at least one main audio or an environmental audio; according to the primary audio Or the feedback information of the ambient audio determines a gain value; and the gain value is used to adjust the main audio portion received by the microphone array Audio of the audio or ambient noise portion; performing a beamforming process to generate a relatively small portion of the main audio portion; using the relatively small portion of the main audio portion and the calibrated main audio portion The signal generates the feedback information of the main audio, and the signal generated by the filtering to generate a main audio portion; the relatively small signal of the main audio portion and the signal of the main audio portion perform a time domain-frequency domain conversion; The nonlinear noise suppression operation estimates an ambient noise; derives a noise reduction gain, and feedback information for the ambient audio; performs noise reduction; and performs a frequency domain-time domain conversion. The method for reducing ambient noise according to claim 8, wherein the time domain-frequency domain conversion system performs a fast Fourier transform process, and performs a signal smoothing operation via the signal of the fast Fourier transform program. The method for reducing ambient noise according to claim 9 is characterized in that the signal smoothed by the signal is subjected to a decimation operation. The method for reducing ambient noise according to claim 8 is characterized in that the signal after the frequency domain-time domain conversion is output after the signal overlap and the summation process. The method for reducing ambient noise according to claim 8 , wherein the beamforming processing program determines a filter coefficient by using a first preset threshold to generate a relatively small signal of the main audio portion; The relatively small signal in the main audio part is adjusted The signal difference of the main audio part of the school is determined by a second preset threshold, and the signal having the main audio part is generated accordingly.