CN115361617B - Multi-microphone environment noise suppression method without blind area - Google Patents
Multi-microphone environment noise suppression method without blind area Download PDFInfo
- Publication number
- CN115361617B CN115361617B CN202210977888.9A CN202210977888A CN115361617B CN 115361617 B CN115361617 B CN 115361617B CN 202210977888 A CN202210977888 A CN 202210977888A CN 115361617 B CN115361617 B CN 115361617B
- Authority
- CN
- China
- Prior art keywords
- microphone
- noise
- power spectrum
- auxiliary
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 230000001629 suppression Effects 0.000 title claims abstract description 13
- 238000001228 spectrum Methods 0.000 claims abstract description 118
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 29
- 230000007613 environmental effect Effects 0.000 claims abstract description 16
- 238000013507 mapping Methods 0.000 claims abstract description 16
- 230000002194 synthesizing effect Effects 0.000 claims abstract description 3
- 230000015572 biosynthetic process Effects 0.000 claims description 10
- 238000009499 grossing Methods 0.000 claims description 10
- 238000003786 synthesis reaction Methods 0.000 claims description 10
- 239000002131 composite material Substances 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 8
- 241000209140 Triticum Species 0.000 description 15
- 235000021307 Triticum Nutrition 0.000 description 15
- 230000035945 sensitivity Effects 0.000 description 4
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000003595 spectral effect Effects 0.000 description 3
- 238000001308 synthesis method Methods 0.000 description 3
- 238000012896 Statistical algorithm Methods 0.000 description 2
- 238000004378 air conditioning Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000001052 transient effect Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R1/00—Details of transducers, loudspeakers or microphones
- H04R1/10—Earpieces; Attachments therefor ; Earphones; Monophonic headphones
- H04R1/1083—Reduction of ambient noise
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L21/0232—Processing in the frequency domain
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L2021/02161—Number of inputs available containing the signal or the noise to be suppressed
- G10L2021/02166—Microphone arrays; Beamforming
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2201/00—Details of transducers, loudspeakers or microphones covered by H04R1/00 but not provided for in any of its subgroups
- H04R2201/10—Details of earpieces, attachments therefor, earphones or monophonic headphones covered by H04R1/10 but not provided for in any of its subgroups
- H04R2201/109—Arrangements to adapt hands free headphones for use on both ears
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2460/00—Details of hearing devices, i.e. of ear- or headphones covered by H04R1/10 or H04R5/033 but not provided for in any of their subgroups, or of hearing aids covered by H04R25/00 but not provided for in any of its subgroups
- H04R2460/01—Hearing devices using active noise cancellation
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Circuit For Audible Band Transducer (AREA)
Abstract
The invention discloses a non-blind area multi-microphone environment noise suppression method, which is used for suppressing environment noise of a telephone traffic earphone provided with a main microphone and N auxiliary microphones, wherein N is an even number greater than or equal to 2, at least one auxiliary microphone is arranged on each earphone side, and an auxiliary microphone synthesized signal is obtained by synthesizing the arranged at least one auxiliary microphone on each earphone side; calculating the noise power spectrum of the synthesized signal of the auxiliary microphone, and calculating the noise power spectrum of the main microphone through the mapping of the frequency domain; and removing the noise signal of the main microphone by using a noise reduction algorithm based on the noise power spectrum and the noise signal power spectrum according to the noise power spectrum and the noise signal power spectrum of the main microphone, and converting the obtained noise-reduced frequency spectrum signal into a time domain signal and outputting the time domain signal. The invention can improve the overall noise reduction effect of the environmental noise.
Description
Technical Field
The invention relates to the technical field of audio processing, in particular to a non-blind area multi-microphone environment noise suppression method.
Background
Headphones with conversation may face noise from all directions, and a person may have a telephone headset with noise sources from the left, as shown in fig. 1, or from the right. The telephone earphone with the function of noise reduction of environmental noise on the market has two modes of single microphone and double microphone noise reduction, and usually the main microphone 20 is arranged in a microphone rod near the mouth of a person and used for collecting the voice of the person, and the auxiliary microphone 10 is arranged at the outer side of one of the ear bags and used for collecting the environmental noise.
The main microphone in the conventional double-microphone noise reduction scheme adopts a directional microphone to collect sound emitted by the mouth as much as possible, the collected noise quantity is reduced to improve the signal-to-noise ratio of the main microphone, and the auxiliary microphone is usually an omni-directional microphone to collect noise from all directions. The double-microphone noise reduction algorithm estimates the environmental noise in the main microphone by using the environmental noise acquired by the auxiliary microphone, so that the double-microphone noise reduction is realized by using the conventional single-microphone noise reduction algorithm (such as wiener filtering, a statistical algorithm based on minimum mean square error, a neural network model based on signal to noise ratio and the like).
The auxiliary microphone is used for estimating the power spectrum of the main microphone by the power spectrum of the auxiliary microphone, and usually has a far noise source and basically the same power spectrum when the noise reaches the main microphone and the auxiliary microphone, so that the power spectrum of the main microphone can be estimated by using the power spectrum of the auxiliary microphone through simple spectrum mapping as long as the frequency response from the auxiliary microphone to the main microphone is calculated.
The noise in real life may be steady air conditioning sound or transient noise such as keyboard knocking, so that it is not good to use the steady noise method to estimate the noise power spectrum of the auxiliary microphone, and it is usually to take the power spectrum of the auxiliary microphone directly as the estimate of the noise power spectrum of the current auxiliary microphone.
Practical tests show that if the noise source comes from one side of the auxiliary microphone, the double-microphone noise reduction algorithm has better effect, and the closer the noise source is to the auxiliary microphone, the better the overall noise reduction algorithm performance is. However, if the noise source comes from the other side of the auxiliary microphone, the noise source can not directly reach the auxiliary microphone, but can directly reach the main microphone, so that on one hand, the signal-to-noise ratio of the main microphone is reduced, and on the other hand, the noise-to-signal ratio of the auxiliary microphone is reduced, and the performance of the whole noise reduction algorithm is greatly reduced.
As described above, in applications such as telephone traffic and examination, the noise source may come from various directions, and the conventional manner of the primary microphone and the strong-side auxiliary microphone may improve the overall noise reduction effect of the noise source from one side of the auxiliary microphone, but the noise reduction effect of the noise source from the other side of the auxiliary microphone may be poor.
Disclosure of Invention
The invention aims at solving the problems in the prior art and provides a multi-microphone environment noise suppression method without blind areas, which is characterized in that by increasing the number of auxiliary microphones of telephone traffic type earphone, each side forms a synthesized auxiliary microphone signal through one or more auxiliary microphones, sound emitted by a human mouth in the auxiliary microphones is suppressed, noise signals acquired by Gao Fumai are extracted as much as possible, so that the noise signals of the auxiliary microphones are improved, the estimated auxiliary microphone noise is more accurate, and the overall environment noise reduction effect is improved.
The technical scheme adopted for realizing the purpose of the invention is as follows:
A non-blind area multi-microphone environment noise suppression method is used for suppressing environment noise of a telephone traffic earphone provided with a main microphone and N auxiliary microphones, N is an even number greater than or equal to 2, at least one auxiliary microphone is arranged on each earphone side, and an auxiliary microphone synthesized signal is obtained by synthesizing the arranged at least one auxiliary microphone on each earphone side;
Calculating the noise power spectrum of the synthesized signal of the auxiliary microphone, and calculating the noise power spectrum of the main microphone through the mapping of the frequency domain;
and removing the noise signal of the main microphone by using a noise reduction algorithm based on the noise power spectrum and the noise signal power spectrum according to the noise power spectrum and the noise signal power spectrum of the main microphone, and converting the obtained noise-reduced frequency spectrum signal into a time domain signal and outputting the time domain signal.
Preferably, the auxiliary microphone synthesis signal is formed by a beamforming method or by a smoothing synthesis method.
Preferably, when the method is formed by a smoothing synthesis method, the amplitude mean square error of the auxiliary microphone signal is calculated first, and then the respective mixing proportion is calculated by determining the respective amplitude mean square error, and the signals are formed by inverse mixing.
Preferably, when the signal is formed by a beam forming method, the two auxiliary wheat collected signals are filtered by a beam forming filter matched with the corresponding second-order cone planning according to the noise source positioning result of the auxiliary wheat, and the final auxiliary wheat synthesized signal is output.
Preferably, the auxiliary microphone is updated with the frequency division point voice presence probability of the main microphone, the noise power spectrum of the auxiliary microphone synthesized signal is adjusted, the lower limit value of the noise power spectrum of the auxiliary microphone synthesized signal is set, and the lower limit value of the estimation of the noise power spectrum of the main microphone is performed.
Wherein, the adjusting the auxiliary wheat noise power spectrum includes:
Taking the noise power spectrum Psd n5 (n, k) of the average signal of the auxiliary microphone multiplied by the gain factor g (n, k) as the lower limit of the noise power spectrum of the synthesized signal of the synthesized auxiliary microphone, obtaining the noise power spectrum of the synthesized signal of the auxiliary microphone after the lower limit adjustment Represented as;
g(n，k)＝1.0-0.5*SPPm1(n，k)
In the above equation, SPP m1 (n, k) is the frequency division point speech presence probability of the primary microphone, psd m4 (n, k) is the noisy signal power spectrum of the secondary microphone synthesized signal, and Psd n5 (n, k) is the noise power spectrum of the secondary microphone average signal.
The method comprises the steps of calculating a noise power spectrum of a synthesized signal of an auxiliary microphone, calculating the noise power spectrum of a main microphone through frequency domain mapping, multiplying the noise power spectrum of the main microphone by a gain factor g (n, k) to be used as a lower limit of a final noise power spectrum Psd n1-final (n, k) of the main microphone, and obtaining a noise power spectrum Psd n1-final (n, k) of the main microphone subjected to lower limit adjustment:
Psdn1-final(n,k)＝max(Psdn4-map(n,k),Psdn1(n，k)*g(n，k))
Where Psd n4-map (n, k) is the noise power spectrum of the composite signal of the auxiliary microphone mapped to the noise power spectrum of the main microphone, psd n1 (n, k) is the noise power spectrum of the main microphone, psd m1 (n, k) is the noisy signal power spectrum estimate of the main microphone, psd m4 (n, k) is the noisy signal power spectrum estimate of the composite signal of the auxiliary microphone, β is a parameter between 0 and 1 representing the smoothing factor of the mapping gain factor, the mapping gain factor η (n, k) being performed only if the main microphone speech is detected as being devoid of speech.
According to the multi-microphone environment noise suppression method without the dead zone, the synthesized auxiliary microphone signals are formed through the plurality of auxiliary microphones of the telephone traffic earphone, sound emitted by the human mouth in the auxiliary microphone is suppressed, and the noise signals collected by Gao Fumai are extracted, so that the noise-signal ratio of the auxiliary microphone is improved, the estimated auxiliary microphone noise is more accurate, and the overall environment noise reduction effect is improved.
Drawings
Fig. 1 is a schematic diagram of a microphone arrangement of a prior art dual microphone noise reduction microphone.
Fig. 2 is a schematic diagram of an auxiliary microphone arrangement of the non-blind zone multi-microphone ambient noise suppression system of the present invention.
Fig. 3 is an ambient noise suppression flow chart of the non-blind zone multi-microphone ambient noise suppression system of the present invention.
Fig. 4 is a schematic diagram of a beamforming algorithm for multi-secondary wheat synthesis.
Detailed Description
The invention is described in further detail below with reference to the drawings and the specific examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In the application, the main microphone is synonymous with the main microphone, and the auxiliary microphone, the auxiliary microphone and the auxiliary microphone are synonymous.
For a telephone traffic earphone configured with a main microphone and an auxiliary microphone, assuming that the signal emitted by the human mouth is s (t), the ambient noise is n (t), the transfer function from the human mouth to the main microphone is h1 (t), and the transfer function from the human mouth to the auxiliary microphone is h2 (t), the main microphone and the auxiliary microphone signals are expressed as:
Representing convolution operations
The time domain signals of the main microphone and the auxiliary microphone are obtained by short-time Fourier transformation to the frequency domain:
M1(n，k)＝H1(n，k)*S(n，k)+N1(n，k)＝S1(n，k)+N1(n，k) (3)
M2(n，k)＝H2(n，k)*S(n，k)+N2(n，k)＝S2(n，k)+N2(n，k) (4)
Wherein M1 (N, k), H1 (N, k), S (N, k), N1 (N, k), S1 (N, k), M2 (N, k), H2 (N, k), N2 (N, k), S2 (N, k) are short-time Fourier transform frequency domain representations of time domain signals M1 (t), H1 (t), S (t), N1 (t), S1 (t), M2 (t), H2 (t), N2 (t), S2 (t), respectively, representing multiplication operations.
Calculating the power spectrum PSD of the noisy signals of the main microphone and the auxiliary microphone according to the short-time Fourier transform result;
Psdm1(n,k)＝||M1(n,k)||2＝||S1(n,k)+N1(n,k)||2 (5)
Psdm2(n,k)＝||M2(n,k)||2＝||S2(n,k)+N2(n,k)||2 (6)
the auxiliary microphone is used for noise power spectrum in the main microphone, that is, the power spectrum of the environmental noise N1 (N, k) of the main microphone is estimated by the power spectrum of the environmental noise N2 (N, k) of the auxiliary microphone, the noise source is usually far, and the power spectrum when the noise reaches the main and auxiliary microphones is basically the same, so that the power spectrum of the environmental noise N1 (N, k) can be estimated by using the power spectrum of the environmental noise N2 (N, k) of the auxiliary microphone through simple spectrum mapping as long as the frequency response from the auxiliary microphone to the main microphone is calculated.
For the environmental noise signal n (t), the noise in real life may be steady air conditioning noise or transient noise such as keyboard strike, so it is not good to use the steady noise method to estimate the noise power spectrum of the auxiliary microphone, and it is usually to directly take the noisy signal power spectrum of the auxiliary microphone as the estimation of the current noise power spectrum of the auxiliary microphone, namely:
Thus, the noise-to-signal ratio in the auxiliary microphone, i.e Will become the key to the overall noise reduction algorithm.
Meanwhile, from practical test, if the noise source comes from one side of the auxiliary microphone, the effect of the double-microphone noise reduction algorithm is better, and the closer the noise source is to the auxiliary microphone, the better the overall noise reduction algorithm performance is. However, if the noise source comes from the other side of the auxiliary microphone, the noise source can not directly reach the auxiliary microphone, but can directly reach the main microphone, so that on one hand, the signal-to-noise ratio of the main microphone is reduced, and on the other hand, the noise-to-signal ratio of the auxiliary microphone is reduced, and the performance of the whole noise reduction algorithm is greatly reduced.
Therefore, in the method for suppressing environmental noise of multiple microphones without blind areas, the auxiliary microphones are added at the side of the auxiliary microphone and are configured into a pair of symmetrical auxiliary microphones, as shown in fig. 2, signals are synthesized by the two auxiliary microphones to overcome the problem of noise estimation at the strong and weak sides of the auxiliary microphones, and the auxiliary microphone signals synthesized by the two auxiliary microphones can effectively reduce the frequency spectrum of voice signals sent by the human mouth in the synthesized signals, but not reduce the noise power spectrum in the synthesized signals, so that the noise signal ratio of the synthesized auxiliary microphone signals is improved.
Naturally the invention can also be extended to scenes with more than 2 auxiliary microphones.
According to the embodiment of the invention, the auxiliary microphone signals for estimating the noise power spectrum of the main microphone are obtained by combining the two auxiliary microphone signals, so that the auxiliary microphones from the earmuffs at the left side and the right side have no blind area for collecting noise. By means of the two auxiliary microphones, leakage of pure voice from the direction of the mouth to the auxiliary microphones can be restrained through signal processing methods such as beam forming and the like.
The signal synthesis of the whole auxiliary microphone can be realized through a software algorithm, and can be realized through simple hardware circuit filtering and mixing, and in the hardware implementation, the extra auxiliary microphone is omitted, and the extra auxiliary circuit is low in complexity and easy to realize.
In the embodiment of the invention, after the noise power spectrum of the auxiliary microphone synthesized signal is obtained, the noise power spectrum of the main microphone can be synthesized through the mapping of the frequency domain, and the accurate noise power spectrum of the main microphone is obtained. By obtaining an accurate main microphone noise power spectrum, various algorithms based on signal-to-noise ratio can well remove noise signals of the main microphone.
The obtained main microphone noise power spectrum can be used for conventional single microphone noise reduction algorithms, such as spectral subtraction, wiener filtering, various spectral domain statistical algorithms based on minimum mean square error and the like, and the latest single microphone noise reduction algorithm based on a neural network, so that the main microphone noise reduction processing is performed, which is the prior art and is not repeated.
The noise power spectrum of the auxiliary microphone is obtained, and the noise power spectrum of the main microphone can be synthesized through frequency domain mapping, wherein the mapping factor of the mapping of the frequency spectrum can be obtained through smoothing the power spectrum of the main microphone and the power spectrum mapping factor of the auxiliary microphone when voice is inactive, as shown in a formula (15), which will be described in detail later.
The accurate noise power spectrum of the main microphone also provides a good signal-to-noise ratio reference for detecting the voice activity of the main microphone, and a simple voice detection algorithm based on the signal-to-noise ratio can be designed, so that the implementation complexity of the whole algorithm is reduced. Since the conventional double-microphone noise reduction algorithm and the single-microphone noise reduction algorithm are related in many ways, the noise reduction algorithm is not described in detail.
In the following section of the present invention, how to implement the synthesis of the noise power spectrum of multiple auxiliary microphones will be described with emphasis, and without loss of generality, an implementation of installing an auxiliary microphone on each earphone side will be taken as an example.
Since a plurality of auxiliary microphones are installed and the standards of microphones when leaving the factory are different, the sensitivity error of the microphones is usually +/-2 dB, and the sensitivity error can be improved to +/-1 dB through sorting, but even so, the microphones installed on the auxiliary microphones still have the sensitivity error, and thus when leaving the factory, the sensitivity calibration can be performed through signals right in front of the earphone, and the calibration value can be directly updated to firmware parameters of a factory version in a factory detection program.
Let the signal acquired by the second subsidiary microphone signal on the other earphone side be expressed as:
wherein n3 (t) is an environmental noise signal acquired by a second auxiliary wheat signal, s3 (t) is a human mouth transmitting signal acquired by the second auxiliary wheat, s (t) is a human mouth transmitting signal, and h3 (t) is a transmission function from the human mouth to the second auxiliary wheat.
After the microphone calibration, the amplitude of the auxiliary microphone signal can be used for judging the direction of the noise source, so that the mixing proportion of the two auxiliary microphones can be calculated through the mean square error of the amplitude of the microphone signal. Further, the synthesized signal of the auxiliary microphone can be synthesized by a smoothing synthesis algorithm, which is calculated as follows:
Where α is a parameter between 0 and 1, and represents a smoothing factor of a statistical time slot of the amplitude mean square error Rms (Root mean square), N is a window length for calculating the amplitude mean square error Rms, i is a frame number for framing with a window, rms2 (b), and Rms3 (b) is the amplitude mean square error of the signal collected by the auxiliary microphone at each earphone side, respectively.
The auxiliary microphone signal synthesized by the two auxiliary microphones is:
wherein m2 (t), m3 (t) are noise signals collected by auxiliary microphones at the left side and the right side of the earphone respectively;
In general, the signals leaking from the mouth to the two auxiliary microphones are basically consistent, i.e. s2 (t) and s3 (t) are highly correlated, and the corresponding noise signals n2 (t) and n3 (t) are relatively independent, so that an inversion is made during the above mixing.
The composite signal of the auxiliary microphone may be formed by a beam forming algorithm, as shown in fig. 4, in addition to the smooth composite algorithm described above. According to the noise source positioning results of the two auxiliary microphones (if no proper noise source position exists, the 0-degree direction is used), under the condition of carrying out beam suppression on the 90-degree human mouth direction, the noise source position is subjected to beam forming, and the beam forming of the multi-target constraint can be realized by using conventional second-order cone planning.
In the embodiment of the invention, the sound source positioning accuracy is not required to be very high, for example, about 10 degrees, a second-order cone programming is used in advance to design a beam forming filter with a required angle, and when auxiliary wheat signals are synthesized, proper filter selection can be carried out according to the position of a noise source; the final synthesized integrated auxiliary signals were:
where f3 (t) and f2 (t) are pre-stored second order cone planned beam forming filters selected according to the noise source angle.
Because the voice signal leaks to the auxiliary microphone when the human mouth speaks, in order to improve the accuracy of estimating the noise power spectrum, when the synthetic signal of the auxiliary microphone is formed by the smooth synthesis method, the inverse mixing is adopted, so as to further improve the accuracy of the noise power spectrum of the synthetic signal. Similarly, when the synthesized signal of the auxiliary microphone is formed by the beam forming method, when the beam forming is realized, a signal for suppressing the direction of the mouth of a person by 90 degrees is selected to be a second order cone target in a certain range when the filter is designed, so that the leakage of a voice signal to the auxiliary microphone can be effectively reduced, and the useful signal component in the auxiliary microphone is reduced.
For the auxiliary microphone of the telephone traffic earphone, no matter the wave beam forming algorithm or the smooth synthesis algorithm, the synthesized noise power spectrum obtained by the auxiliary microphone approaches to the result of the inverse mixing of the two auxiliary microphones, if the noise is a low-frequency component and the wavelength is longer, the noise signal obtained by the inverse mixing is greatly reduced, and therefore the problem of noise power spectrum reduction caused by the synthesis of multiple auxiliary microphones is required to be solved.
In order to solve the problem of noise power spectrum reduction caused by the synthesis of the multi-auxiliary wheat, the invention provides the following technology:
The noise power spectrum of the synthesized signal of the auxiliary microphone is set by updating the noise power spectrum of the arithmetic average output of the whole auxiliary microphone by using the voice existence probability SPP of the main microphone, and when the noise power spectrum of the synthesized signal of the auxiliary microphone is lower than the lower limit value, the lower limit value is used as the noise power spectrum to estimate the noise power spectrum of the main microphone.
Setting the average auxiliary wheat signal as follows:
The frequency division point voice existence Probability (SPP, speech Presence Probability) of the main microphone is calculated by a noise estimation power spectrum and a current frequency point power spectrum in the main microphone according to a formula (14).
Where ζ is a preset SNR value (e.g., 15 dB) calibrated to speech, and pi is 180 degrees corresponding to radians.
The estimated noise power spectrum Psd n5 (n, k) of the auxiliary microphone average signal and the noise power spectrum Psd n1 (n, k) of the main microphone can be obtained through a conventional noise power spectrum algorithm based on SPP, that is, after the frequency division point speech presence probability SPP of the main microphone is estimated, the noise power spectrum is obtained by using bayesian estimation based on SPP, and related algorithms are numerous and are not described herein. Of course other single microphone noise power spectral algorithms are possible.
Wherein the estimated noise power spectrum Psd n5 (n, k) multiplied by the gain factor g (n, k) of the auxiliary average signal is regarded as the lower limit of the noise power spectrum of the synthesized auxiliary synthesized signal, namely the noise power spectrum of the auxiliary synthesized signal subjected to lower limit adjustmentExpressed as:
g(n，k)＝1.0-0.5*SPPm1(n，k)
Similarly, after the synthesized signal noise power spectrum of the auxiliary wheat is mapped to the noise power spectrum of the main wheat (Psd n4-map (n, k)), the final noise power spectrum Psd n1 (n, k) in the main wheat multiplied by the gain factor g (n, k) may also be used as the lower limit of the final noise power spectrum Psd n1-final (n, k) of the main wheat, as shown in equation (15):
Where β is a parameter between 0 and 1, representing the smoothing factor of the mapped gain factor, it should be noted that the mapped gain factor η (n, k) is only performed if the host speech is detected as being absent speech.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (5)
1. A multi-microphone environmental noise suppression method without blind areas is characterized in that the method is used for suppressing environmental noise of telephone traffic type earphone provided with a main microphone and N auxiliary microphones, N is an even number greater than or equal to 2, at least one auxiliary microphone is arranged on each earphone side, and an auxiliary microphone synthesized signal is obtained by synthesizing the arranged at least one auxiliary microphone on each earphone side;
Calculating the noise power spectrum of the synthesized signal of the auxiliary microphone, and calculating the noise power spectrum of the main microphone through the mapping of the frequency domain;
Removing the noise signal of the main microphone according to the noise power spectrum and the power spectrum of the noise-carrying signal of the main microphone by using a noise reduction algorithm based on the noise power spectrum and the power spectrum of the noise-carrying signal, and converting the obtained noise-reduced frequency spectrum signal into a time domain signal to be output;
The method comprises the steps of updating a noise power spectrum of an auxiliary microphone synthesized signal for an auxiliary microphone by using the frequency division point voice existence probability of a main microphone, adjusting the noise power spectrum of the auxiliary microphone synthesized signal, setting a lower limit value of the noise power spectrum of the auxiliary microphone synthesized signal, and estimating the lower limit value of the noise power spectrum of the main microphone according to the lower limit value; the adjusting the noise power spectrum of the auxiliary synthesized signal comprises:
Taking the noise power spectrum Psd n5 (n, k) of the average signal of the auxiliary microphone multiplied by the gain factor g (n, k) as the lower limit of the noise power spectrum of the synthesized signal of the synthesized auxiliary microphone, obtaining the noise power spectrum of the synthesized signal of the auxiliary microphone after the lower limit adjustment Represented as;
g(n，k)＝1.0-0.5*SPPm1(n，k)
In the above equation, SPP m1 (n, k) is the frequency division point speech presence probability of the primary microphone, psd m4 (n, k) is the noisy signal power spectrum of the secondary microphone synthesized signal, and Psd n5 (n, k) is the noise power spectrum of the secondary microphone average signal.
2. The non-blind zone multi-microphone ambient noise suppression method of claim 1 wherein the auxiliary microphone composite signal is formed by beamforming or by a smoothing synthesis.
3. The method for suppressing environmental noise of multiple microphones without blind area according to claim 2, wherein said method is characterized in that when said method is formed by smoothing synthesis, the mean square deviation of the amplitude of the auxiliary microphone signal is calculated first, then the mixing proportion is calculated by determining the respective mean square deviation of the amplitude, and the mixture is formed by inverse mixing.
4. The method for suppressing environmental noise of multiple microphones without blind areas according to claim 2, wherein the forming by the beamforming method is based on the noise source positioning result of the auxiliary microphone, and the filtering processing is performed on the signals collected by the two auxiliary microphones by using the beamforming filter matched with the corresponding second order cone plan, so as to output the final synthesized signal of the auxiliary microphone.
5. The method for suppressing noise in a non-blind area environment according to claim 1, wherein the calculating the noise power spectrum of the synthesized signal of the auxiliary microphone calculates the noise power spectrum of the main microphone through the mapping of the frequency domain, and multiplies the noise power spectrum of the main microphone by a gain factor g (n, k) to obtain a lower limit of a final noise power spectrum Psd n1-final (n, k) of the main microphone, thereby obtaining a noise power spectrum Psd n1-final (n, k) of the main microphone with the lower limit adjusted:
Psdn1-final(n,k)＝max(Psdn4-map(n,k),Psdn1(n，k)*g(n，k))
Where Psd n4-map (n, k) is the noise power spectrum of the composite signal of the auxiliary microphone mapped to the noise power spectrum of the main microphone, psd n1 (n, k) is the noise power spectrum of the main microphone, psd m1 (n, k) is the noisy signal power spectrum of the main microphone, psd m4 (n, k) is the noisy signal power spectrum of the composite signal of the auxiliary microphone, β is a parameter between 0 and 1 representing the smoothing factor of the mapping gain factor, the mapping gain factor η (n, k) being performed only if the main microphone speech is detected as being devoid of speech.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210977888.9A CN115361617B (en) | 2022-08-15 | 2022-08-15 | Multi-microphone environment noise suppression method without blind area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210977888.9A CN115361617B (en) | 2022-08-15 | 2022-08-15 | Multi-microphone environment noise suppression method without blind area |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115361617A CN115361617A (en) | 2022-11-18 |
| CN115361617B true CN115361617B (en) | 2024-07-26 |
Family
ID=84033654
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210977888.9A Active CN115361617B (en) | 2022-08-15 | 2022-08-15 | Multi-microphone environment noise suppression method without blind area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115361617B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104424954A (en) * | 2013-08-20 | 2015-03-18 | 华为技术有限公司 | Noise estimation method and device |
| CN111418010A (en) * | 2017-12-08 | 2020-07-14 | 华为技术有限公司 | Multi-microphone noise reduction method and device and terminal equipment |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9966067B2 (en) * | 2012-06-08 | 2018-05-08 | Apple Inc. | Audio noise estimation and audio noise reduction using multiple microphones |
| CN104754430A (en) * | 2013-12-30 | 2015-07-01 | 重庆重邮信科通信技术有限公司 | Noise reduction device and method for terminal microphone |
| US10186278B2 (en) * | 2015-04-29 | 2019-01-22 | Intel Corporation | Microphone array noise suppression using noise field isotropy estimation |
| CN107371079B (en) * | 2017-04-17 | 2019-10-11 | 恒玄科技（上海）有限公司 | A kind of the diamylose gram noise reduction system and noise-reduction method of earphone |
| KR102087019B1 (en) * | 2018-11-23 | 2020-03-10 | 주식회사 위스타 | Method and apparatus for estimating direction of sound source and removing ambient noise using microphone array |
| CN111048106B (en) * | 2020-03-12 | 2020-06-16 | 深圳市友杰智新科技有限公司 | Pickup method and apparatus based on double microphones and computer device |
| CN113096684A (en) * | 2021-06-07 | 2021-07-09 | 成都启英泰伦科技有限公司 | Target voice extraction method based on double-microphone array |
| CN114596874B (en) * | 2022-03-03 | 2025-08-05 | 上海富瀚微电子股份有限公司 | A wind noise suppression method and device based on multiple microphones |
-
2022
- 2022-08-15 CN CN202210977888.9A patent/CN115361617B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104424954A (en) * | 2013-08-20 | 2015-03-18 | 华为技术有限公司 | Noise estimation method and device |
| CN111418010A (en) * | 2017-12-08 | 2020-07-14 | 华为技术有限公司 | Multi-microphone noise reduction method and device and terminal equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115361617A (en) | 2022-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP5762956B2 (en) | System and method for providing noise suppression utilizing nulling denoising | |
| US9635474B2 (en) | Method of processing a signal in a hearing instrument, and hearing instrument | |
| JP4912036B2 (en) | Directional sound collecting device, directional sound collecting method, and computer program | |
| US8606571B1 (en) | Spatial selectivity noise reduction tradeoff for multi-microphone systems | |
| US8204252B1 (en) | System and method for providing close microphone adaptive array processing | |
| US20230186935A1 (en) | Bone conduction headphone speech enhancement systems and methods | |
| DK3008924T3 (en) | METHOD OF SIGNAL PROCESSING IN A HEARING SYSTEM AND HEARING SYSTEM | |
| US8615092B2 (en) | Sound processing device, correcting device, correcting method and recording medium | |
| CN108986832B (en) | Method and device for binaural speech de-reverberation based on speech occurrence probability and consistency | |
| JP2010513987A (en) | Near-field vector signal amplification | |
| CN112331226B (en) | Voice enhancement system and method for active noise reduction system | |
| JPWO2009078105A1 (en) | Noise suppression device, noise suppression control device, noise suppression method, and noise suppression program | |
| CN111951818A (en) | Double-microphone speech enhancement method based on improved power difference noise estimation algorithm | |
| CN110310650A (en) | A Speech Enhancement Algorithm Based on Second-Order Differential Microphone Array | |
| US10674284B2 (en) | Method of operating a hearing device and a hearing device | |
| JP6840302B2 (en) | Information processing equipment, programs and information processing methods | |
| JP6854967B1 (en) | Noise suppression device, noise suppression method, and noise suppression program | |
| CN114724574B (en) | Dual-microphone noise reduction method with adjustable expected sound source direction | |
| CN115361617B (en) | Multi-microphone environment noise suppression method without blind area | |
| Puder et al. | Optimized directional processing in hearing aids with integrated spatial noise reduction | |
| Zhu et al. | Modified complementary joint sparse representations: a novel post-filtering to MVDR beamforming | |
| Kodrasi et al. | Curvature-based optimization of the trade-off parameter in the speech distortion weighted multichannel wiener filter | |
| EP3531719B1 (en) | Dereverberation device and hearing aid | |
| Lotter et al. | A stereo input-output superdirective beamformer for dual channel noise reduction. | |
| CN119155585A (en) | Voice activity detection method and headset |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |