Microphone array system
A method and system for enhancing a target sound signal from multiple sound signals is provided. An array of an arbitrary number of sound sensors positioned in an arbitrary configuration receives the sound signals from multiple disparate sources. The sound signals comprise the target sound signal from a target sound source, and ambient noise signals. A sound source localization unit, an adaptive beamforming unit, and a noise reduction unit are in operative communication with the array of sound sensors. The sound source localization unit estimates a spatial location of the target sound signal from the received sound signals. The adaptive beamforming unit performs adaptive beamforming by steering a directivity pattern of the array of sound sensors in a direction of the spatial location of the target sound signal, thereby enhancing the target sound signal and partially suppressing the ambient noise signals, which are further suppressed by the noise reduction unit.
Latest LI Creative Technologies, Inc. Patents:
This application is a reissue application of U.S. patent application Ser. No. 13/049,877, filed Mar. 16, 2011 (now U.S. Pat. No. 8,861,756), which claims the benefit of provisional patent application No. 61/403,952 titled “Microphone array design and implementation for telecommunications and handheld devices”, filed on Sep. 24, 2010 in the United States Patent and Trademark Office.
The specification of the above referenced patent application is incorporated herein by reference in its entirety.
BACKGROUNDMicrophones constitute an important element in today's speech acquisition devices. Currently, most of the hands-free speech acquisition devices, for example, mobile devices, lapels, headsets, etc., convert sound into electrical signals by using a microphone embedded within the speech acquisition device. However, the paradigm of a single microphone often does not work effectively because the microphone picks up many ambient noise signals in addition to the desired sound, specifically when the distance between a user and the microphone is more than a few inches. Therefore, there is a need for a microphone system that operates under a variety of different ambient noise conditions and that places fewer constraints on the user with respect to the microphone, thereby eliminating the need to wear the microphone or be in close proximity to the microphone.
To mitigate the drawbacks of the single microphone system, there is a need for a microphone array that achieves directional gain in a preferred spatial direction while suppressing ambient noise from other directions. Conventional microphone arrays include arrays that are typically developed for applications such as radar and sonar, but are generally not suitable for hands-free or handheld speech acquisition devices. The main reason is that the desired sound signal has an extremely wide bandwidth relative to its center frequency, thereby rendering conventional narrowband techniques employed in the conventional microphone arrays unsuitable. In order to cater to such broadband speech applications, the array size needs to be vastly increased, making the conventional microphone arrays large and bulky, and precluding the conventional microphone arrays from having broader applications, for example, in mobile and handheld communication devices. There is a need for a microphone array system that provides an effective response over a wide spectrum of frequencies while being unobtrusive in terms of size.
Hence, there is a long felt but unresolved need for a broadband microphone array and broadband beamforming system that enhances acoustics of a desired sound signal while suppressing ambient noise signals.
SUMMARY OF THE INVENTIONThis summary is provided to introduce a selection of concepts in a simplified form that are further described in the detailed description of the invention. This summary is not intended to identify key or essential inventive concepts of the claimed subject matter, nor is it intended for determining the scope of the claimed subject matter.
The method and system disclosed herein addresses the above stated need for enhancing acoustics of a target sound signal received from a target sound source, while suppressing ambient noise signals. As used herein, the term “target sound signal” refers to a sound signal from a desired or target sound source, for example, a person's speech that needs to be enhanced. A microphone array system comprising an array of sound sensors positioned in an arbitrary configuration, a sound source localization unit, an adaptive beamforming unit, and a noise reduction unit, is provided. The sound source localization unit, the adaptive beamforming unit, and the noise reduction unit are in operative communication with the array of sound sensors. The array of sound sensors is, for example, a linear array of sound sensors, a circular array of sound sensors, or an arbitrarily distributed coplanar array of sound sensors. The array of sound sensors herein referred to as a “microphone array” receives sound signals from multiple disparate sound sources. The method disclosed herein can be applied on a microphone array with an arbitrary number of sound sensors having, for example, an arbitrary two dimensional (2D) configuration. The sound signals received by the sound sensors in the microphone array comprise the target sound signal from the target sound source among the disparate sound sources, and ambient noise signals.
The sound source localization unit estimates a spatial location of the target sound signal from the received sound signals, for example, using a steered response power-phase transform. The adaptive beamforming unit performs adaptive beamforming for steering a directivity pattern of the microphone array in a direction of the spatial location of the target sound signal. The adaptive beamforming unit thereby enhances the target sound signal from the target sound source and partially suppresses the ambient noise signals. The noise reduction unit suppresses the ambient noise signals for further enhancing the target sound signal received from the target sound source.
In an embodiment where the target sound source that emits the target sound signal is in a two dimensional plane, a delay between each of the sound sensors and an origin of the microphone array is determined as a function of distance between each of the sound sensors and the origin, a predefined angle between each of the sound sensors and a reference axis, and an azimuth angle between the reference axis and the target sound signal. In another embodiment where the target sound source that emits the target sound signal is in a three dimensional plane, the delay between each of the sound sensors and the origin of the microphone array is determined as a function of distance between each of the sound sensors and the origin, a predefined angle between each of the sound sensors and a first reference axis, an elevation angle between a second reference axis and the target sound signal, and an azimuth angle between the first reference axis and the target sound signal. This method of determining the delay enables beamforming for arbitrary numbers of sound sensors and multiple arbitrary microphone array configurations. The delay is determined, for example, in terms of number of samples. Once the delay is determined, the microphone array can be aligned to enhance the target sound signal from a specific direction.
The adaptive beamforming unit comprises a fixed beamformer, a blocking matrix, and an adaptive filter. The fixed beamformer steers the directivity pattern of the microphone array in the direction of the spatial location of the target sound signal from the target sound source for enhancing the target sound signal, when the target sound source is in motion. The blocking matrix feeds the ambient noise signals to the adaptive filter by blocking the target sound signal from the target sound source. The adaptive filter adaptively filters the ambient noise signals in response to detecting the presence or absence of the target sound signal in the sound signals received from the disparate sound sources. The fixed beamformer performs fixed beamforming, for example, by filtering and summing output sound signals from the sound sensors.
In an embodiment, the adaptive filtering comprises sub-band adaptive filtering. The adaptive filter comprises an analysis filter bank, an adaptive filter matrix, and a synthesis filter bank. The analysis filter bank splits the enhanced target sound signal from the fixed beamformer and the ambient noise signals from the blocking matrix into multiple frequency sub-bands. The adaptive filter matrix adaptively filters the ambient noise signals in each of the frequency sub-bands in response to detecting the presence or absence of the target sound signal in the sound signals received from the disparate sound sources. The synthesis filter bank synthesizes a full-band sound signal using the frequency sub-bands of the enhanced target sound signal. In an embodiment, the adaptive beamforming unit further comprises an adaptation control unit for detecting the presence of the target sound signal and adjusting a step size for the adaptive filtering in response to detecting the presence or the absence of the target sound signal in the sound signals received from the disparate sound sources.
The noise reduction unit suppresses the ambient noise signals for further enhancing the target sound signal from the target sound source. The noise reduction unit performs noise reduction, for example, by using a Wiener-filter based noise reduction algorithm, a spectral subtraction noise reduction algorithm, an auditory transform based noise reduction algorithm, or a model based noise reduction algorithm. The noise reduction unit performs noise reduction in multiple frequency sub-bands employed for sub-band adaptive beamforming by the analysis filter bank of the adaptive beamforming unit.
The microphone array system disclosed herein comprising the microphone array with an arbitrary number of sound sensors positioned in arbitrary configurations can be implemented in handheld devices, for example, the iPad® of Apple Inc., the iPhone® of Apple Inc., smart phones, tablet computers, laptop computers, etc. The microphone array system disclosed herein can further be implemented in conference phones, video conferencing applications, or any device or equipment that needs better speech inputs.
The foregoing summary, as well as the following detailed description of the invention, is better understood when read in conjunction with the appended drawings. For the purpose of illustrating the invention, exemplary constructions of the invention are shown in the drawings. However, the invention is not limited to the specific methods and instrumentalities disclosed herein.
The array of sound sensors herein referred to as a “microphone array” comprises multiple or an arbitrary number of sound sensors, for example, microphones, operating in tandem. The microphone array refers to an array of an arbitrary number of sound sensors positioned in an arbitrary configuration. The sound sensors are transducers that detect sound and convert the sound into electrical signals. The sound sensors are, for example, condenser microphones, piezoelectric microphones, etc.
The sound sensors receive 102 sound signals from multiple disparate sound sources and directions. The target sound source that emits the target sound signal is one of the disparate sound sources. As used herein, the term “sound signals” refers to composite sound energy from multiple disparate sound sources in an environment of the microphone array. The sound signals comprise the target sound signal from the target sound source and the ambient noise signals. The sound sensors are positioned in an arbitrary planar configuration herein referred to as a “microphone array configuration”, for example, a linear configuration, a circular configuration, any arbitrarily distributed coplanar array configuration, etc. By employing beamforming according to the method disclosed herein, the microphone array provides a higher response to the target sound signal received from a particular direction than to the sound signals from other directions. A plot of the response of the microphone array versus frequency and direction of arrival of the sound signals is referred to as a directivity pattern of the microphone array.
The sound source localization unit estimates 103 a spatial location of the target sound signal from the received sound signals. In an embodiment, the sound source localization unit estimates the spatial location of the target sound signal from the target sound source, for example, using a steered response power-phase transform as disclosed in the detailed description of
The adaptive beamforming unit performs adaptive beamforming 104 by steering the directivity pattern of the microphone array in a direction of the spatial location of the target sound signal, thereby enhancing the target sound signal, and partially suppressing the ambient noise signals. Beamforming refers to a signal processing technique used in the microphone array for directional signal reception, that is, spatial filtering. This spatial filtering is achieved by using adaptive or fixed methods. Spatial filtering refers to separating two signals with overlapping frequency content that originate from different spatial locations.
The noise reduction unit performs noise reduction by further suppressing 105 the ambient noise signals and thereby further enhancing the target sound signal. The noise reduction unit performs the noise reduction, for example, by using a Wiener-filter based noise reduction algorithm, a spectral subtraction noise reduction algorithm, an auditory transform based noise reduction algorithm, or a model based noise reduction algorithm.
The array 201 of sound sensors, herein referred to as the “microphone array” is in operative communication with the sound source localization unit 202, the adaptive beamforming unit 203, and the noise reduction unit 207. The microphone array 201 is, for example, a linear array of sound sensors, a circular array of sound sensors, or an arbitrarily distributed coplanar array of sound sensors. The microphone array 201 achieves directional gain in any preferred spatial direction and frequency band while suppressing signals from other spatial directions and frequency bands. The sound sensors receive the sound signals comprising the target sound signal and ambient noise signals from multiple disparate sound sources, where one of the disparate sound sources is the target sound source that emits the target sound signal.
The sound source localization unit 202 estimates the spatial location of the target sound signal from the received sound signals. In an embodiment, the sound source localization unit 202 uses, for example, a steered response power-phase transform, for estimating the spatial location of the target sound signal from the target sound source.
The adaptive beamforming unit 203 steers the directivity pattern of the microphone array 201 in a direction of the spatial location of the target sound signal, thereby enhancing the target sound signal and partially suppressing the ambient noise signals. The adaptive beamforming unit 203 comprises a fixed beamformer 204, a blocking matrix 205, and an adaptive filter 206 as disclosed in the detailed description of
The noise reduction unit 207 further suppresses the ambient noise signals for further enhancing the target sound signal. The noise reduction unit 207 is, for example, a Wiener-filter based noise reduction unit, a spectral subtraction noise reduction unit, an auditory transform based noise reduction unit, or a model based noise reduction unit.
The output “y” of the microphone array 201 having N sound sensors 301 is the filter-and-sum of the outputs of the N sound sensors 301. That is, y=Σn=0N−1wnTxn, where xn is the output of the (n+1)th sound sensor 301, and wnT denotes a transpose of a length-L filter applied to the (n+1)th sound sensor 301.
The spatial directivity pattern H (ω, θ) for the target sound signal from angle θ with normalized frequency w is defined as:
where
For example, the angle between the Y-axis and the line joining the origin and the sound sensor 301 M0 is Φ0, the angle between the Y-axis and the line joining the origin and the sound sensor 301 M1 is Φ1, the angle between the Y-axis and the line joining the origin and the sound sensor 301 M2 is Φ2, and the angle between the Y-axis and the line joining the origin and the sound sensor 301 M3 is Φ3. The distance between the origin ◯ and the sound sensor 301 M1, and the origin ◯ and the sound sensor 301 M3 when the incoming target sound signal from the target sound source is at an angle θ from the Y-axis is denoted as τ1 and τ3, respectively.
For purposes of illustration, the detailed description refers to a circular microphone array configuration; however, the scope of the microphone array system 200 disclosed herein is not limited to the circular microphone array configuration but may be extended to include a linear array configuration, an arbitrarily distributed coplanar array configuration, or a microphone array configuration with any arbitrary geometry.
If the target sound source is far enough from the microphone array 201, the time delay between the signal received by the (n+1)th sound sensor 301 “xn,” and the origin of the microphone array 201 is herein denoted as “t” measured in seconds. The sound signals received by the microphone array 201, which are in analog form are converted into digital sound signals by sampling the analog sound signals at a particular frequency, for example, 8000 Hz. That is, the number of samples in each second is 8000. The delay τ can be represented as the product of the sampling frequency (fs) and the time delay (t). That is, τ=fs*t. Therefore, the distance between the sound sensors 301 in the microphone array 201 corresponds to the time used for the target sound signal to travel the distance and is measured by the number of samples within that time period.
Consider an example where “d” is the radius of the circle 302 of the circular microphone array configuration, “fs” is the sampling frequency, and “c” is the speed of sound.
The method of determining the delay (τ) enables beamforming for arbitrary numbers of sound sensors 301 and multiple arbitrary microphone array configurations. Once the delay (τ) is determined, the microphone array 201 can be aligned to enhance the target sound signal from a specific direction.
Therefore, the spatial directivity pattern H can be re-written as:
H(ω,θ)=Σn=0N−1Wn(ω)e−jωτ
where wT=[w0T, w1T, w2T, w3T, . . . , wN−1T] and g(ω,θ)={gi(ω, θ)}i=1 . . . NL={e−jω(k+τ
Consider an example of a microphone array configuration with four sound sensors 301 M0, M1, M2, and M3.
Consider a least mean square solution for beamforming according to the method disclosed herein. Let the spatial directivity pattern be 1 in the passband and 0 in the stopband. The least square cost function is defined as:
Replacing
|H(ω,θ)|2=wTg(ω,θ)gH(ω,θ)w=wT(GR(ω,θ)+jG1(ω,θ))w=wTGR(ω,θ)w and Re(H(ω,θ))=wTgR(ω,θ), J(ω) becomes
J(ω)=wTQw−2wTα+d, where
Q=∫Ω
α=∫Ω
d=∫Ω
where gR(ω,θ)=cos [w(k+τn)] and GR(ω,θ)=cos [w(k−1+τn−τm)].
When ∂J/∂w=0, the cost function J is minimized. The least-square estimate of w is obtained by:
w=Q−1α (5)
Applying linear constrains Cw=b, the spatial response is further constrained to a predefined value b at angle θf using following equation:
Now, the design problem becomes:
and the solution of the constrained minimization problem is equal to:
w=Q−1CT(CQ−1CT)−1(b−CQ−1α)+Q−1α (8)
where w is the filter parameter for the designed adaptive beamforming unit 203.
In an embodiment, the beamforming is performed by a delay-sum method. In another embodiment, the beamforming is performed by a filter-sum method.
For direction i (0≤t≤360), the delay Dit is calculated 801 between the tth pair of the sound sensors 301 (t=1: all pairs). The correlation value corr(Dit) between the tth pair of the sound sensors 301 corresponding to the delay of Dit is then calculated 802. For the direction i (0≤i≤360), the correlation value is given 803 by:
Therefore, the spatial location of the target sound signal is given 804 by:
As exemplarily illustrated in
The output “z” of the blocking matrix 205 may contain some weak target sound signals due to signal leakage. If the adaptation is active when the target sound signal, for example, speech is present, the speech is cancelled out with the noise. Therefore, the adaptation control unit 208 determines when the adaptation should be applied. The adaptation control unit 208 comprises a target sound signal detector 208a and a step size adjusting module 208b. The target sound signal detector 208a of the adaptation control unit 208 detects the presence or absence of the target sound signal, for example, speech. The step size adjusting module 208b adjusts the step size for the adaptation process such that when the target sound signal is present, the adaptation is slow for preserving the target sound signal, and when the target sound signal is absent, adaptation is quick for better cancellation of the ambient noise signals.
The adaptive filter 206 is a filter that adaptively updates filter coefficients of the adaptive filter 206 so that the adaptive filter 206 can be operated in an unknown and changing environment. The adaptive filter 206 adaptively filters the ambient noise signals in response to detecting presence or absence of the target sound signal in the sound signals received from the disparate sound sources. The adaptive filter 206 adapts its filter coefficients with the changes in the ambient noise signals, thereby eliminating distortion in the target sound signal, when the target sound source and the ambient noise signals are in motion. In an embodiment, the adaptive filtering is performed by a set of sub-band adaptive filters using sub-band adaptive filtering as disclosed in the detailed description of
As exemplarily illustrated in
The adaptive filter matrix 206b adaptively filters the ambient noise signals in each of the frequency sub-bands in response to detecting the presence or absence of the target sound signal in the sound signals received from the disparate sound sources. The adaptive filter matrix 206b performs an adaptation step, where the adaptive filter 206 is adapted such that the filter output only contains the target sound signal, for example, speech. The synthesis filter bank 206c synthesizes a full-band sound signal using the frequency sub-bands of the enhanced target sound signal. The synthesis filter bank 206c performs a synthesis step where the sub-band sound signal is synthesized into a full-band sound signal. Since the noise reduction and the beamforming are performed in the same sub-band framework, the noise reduction as disclosed in the detailed description of
In an embodiment, the analysis filter bank 206a is implemented as a perfect-reconstruction filter bank, where the output of the synthesis filter bank 206c after the analysis and synthesis steps perfectly matches the input to the analysis filter bank 206a. That is, all the sub-band analysis filter banks 206a are factorized to operate on prototype filter coefficients and a modulation matrix is used to take advantage of the fast Fourier transform (FFT). Both analysis and synthesize steps require performing frequency shifts in each sub-band, which involves complex value computations with cosines and sinusoids. The method disclosed herein employs the FFT to perform the frequency shifts required in each sub-band, thereby minimizing the amount of multiply-accumulate operations. The implementation of the sub-band analysis filter bank 206a as a perfect-reconstruction filter bank ensures the quality of the target sound signal by ensuring that the sub-band analysis filter banks 206a do not distort the target sound signal itself.
In an embodiment, the noise reduction is performed using the Wiener-filter based noise reduction algorithm. The noise reduction unit 207 explores the short-term and long-term statistics of the target sound signal, for example, speech, and the ambient noise signals, and the wide-band and narrow-band signal-to-noise ratio (SNR) to support a Wiener gain filtering. The noise reduction unit 207 comprises a target sound signal statistics analyzer 207a, a noise statistics analyzer 207b, a signal-to-noise ratio (SNR) analyzer 207c, and a Wiener filter 207d. The target sound signal statistics analyzer 207a explores the short-term and long-term statistics of the target sound signal, for example, speech. Similarly, the noise statistics analyzer 207b explores the short-term and long-term statistics of the ambient noise signals. The SNR analyzer 207c of the noise reduction unit 207 explores the wide-band and narrow-band signal-to-noise ratio (SNR). After the spectrum of noisy-speech passes through the Wiener filter 207d, an estimation of the clean-speech spectrum is generated. The synthesis filter bank 206c, by an inverse process of the analysis filter bank 206a, reconstructs the signals of the clean speech into a full-band signal, given the estimated spectrum of the clean speech.
Consider an example where the microphone array 201 comprises four sound sensors 301 that pick up the sound signals. Four microphone amplifiers 1401 receive the output sound signals from the four sound sensors 301. The microphone amplifiers 1401 also referred to as preamplifiers provide a gain to boost the power of the received sound signals for enhancing the sensitivity of the sound sensors 301. In an example, the gain of the preamplifiers is 20 dB.
The audio codec 1402 receives the amplified output from the microphone amplifiers 1401. The audio codec 1402 provides an adjustable gain level, for example, from about −74 dB to about 6 dB. The received sound signals are in an analog form. The audio codec 1402 converts the four channels of the sound signals in the analog form into digital sound signals. The pre-amplifiers may not be required for some applications. The audio codec 1402 then transmits the digital sound signals to the DSP 1403 for processing of the digital sound signals. The DSP 1403 implements the sound source localization unit 202, the adaptive beamforming unit 203, and the noise reduction unit 207.
After the processing, the DSP 1403 either stores the processed signal from the DSP 1403 in a memory device for a recording application, or transmits the processed signal to the communication interface 1409. The recording application comprises, for example, storing the processed signal onto the memory device for the purposes of playing back the processed signal at a later time. The communication interface 1409 transmits the processed signal, for example, to a computer, the internet, or a radio for communicating the processed signal. In an embodiment, the microphone array system 200 disclosed herein implements a two-way communication device where the signal received from the communication interface 1409 is processed by the DSP 1403 and the processed signal is then played through the loudspeaker or the headphone 1408.
The flash memory 1404 stores the code for the DSP 1403 and compressed audio signals. When the microphone array system 200 boots up, the DSP 1403 reads the code from the flash memory 1404 into an internal memory of the DSP 1403 and then starts executing the code. In an embodiment, the audio codec 1402 can be configured for encoding and decoding audio or sound signals during the start up stage by writing to registers of the DSP 1403. For an eight-sensor microphone array 201, two four-channel audio codec 1402 chips may be used. The power regulators 1405 and 1406, for example, linear power regulators 1405 and switch power regulators 1406 provide appropriate voltage and current supply for all the components, for example, 201, 1401, 1402, 1403, etc., mechanically supported and electrically connected on a circuit board. A universal serial bus (USB) control is built into the DSP 1403. The battery 1407 is used for powering the microphone array system 200.
Consider an example where the microphone array system 200 disclosed herein is implemented on a mixed signal circuit board having a six-layer printed circuit board (PCB). Noisy digital signals easily contaminate the low voltage analog sound signals from the sound sensors 301. Therefore, the layout of the mixed signal circuit board is carefully partitioned to isolate the analog circuits from the digital circuits. Although both the inputs and outputs of the microphone amplifiers 1401 are in analog form, the microphone amplifiers 1401 are placed in a digital region of the mixed signal circuit board because of their high power consumption 1401 and switch amplifier nature.
The linear power regulators 1405 are deployed in an analog region of the mixed signal circuit board due to the low noise property exhibited by the linear power regulators 1405. Five power regulators, for example, 1405 are designed in the microphone array system 200 circuits to ensure quality. The switch power regulators 1406 achieve an efficiency of about 95% of the input power and have high output current capacity; however their outputs are too noisy for analog circuits. The efficiency of the linear power regulators 1405 is determined by the ratio of the output voltage to the input voltage, which is lower than that of the switch power regulators 1406 in most cases. The regulator outputs utilized in the microphone array system 200 circuits are stable, quiet, and suitable for the low power analog circuits.
In an example, the microphone array system 200 is designed with a microphone array 201 having dimensions of 10 cm×2.5 cm×1.5 cm, a USB interface, and an assembled PCB supporting the microphone array 201 and a DSP 1403 having a low power consumption design devised for portable devices, a four-channel codec 1402, and a flash memory 1404. The DSP 1403 chip is powerful enough to handle the DSP 1403 computations in the microphone array system 200 disclosed herein. The hardware configuration of this example can be used for any microphone array configuration, with suitable modifications to the software. In an embodiment, the adaptive beamforming unit 203 of the microphone array system 200 is implemented as hardware with software instructions programmed on the DSP 1403. The DSP 1403 is programmed for beamforming, noise reduction, echo cancellation, and USB interfacing according to the method disclosed herein, and fine tuned for optimal performance.
The computer simulation for verifying the performance of the adaptive beamforming unit 203 when the target sound signal is received from the target sound source in the spatial region centered at 15° uses the following parameters:
Sampling frequency fs=16 k,
FIR filter taper length L=20
Passband (Θp, Ωp)={300-5000 Hz, −5°-35°}, designed spatial directivity pattern is 1.
Stopband (Θs, Ωs)={300˜5000 Hz,−180°˜−15°+45°˜180°}, the designed spatial directivity pattern is 0.
It can be seen that the directivity pattern of the microphone array 201 in the spatial region centered at 15° is enhanced while the sound signals from all other spatial regions are suppressed.
Sampling frequency fs=16 k,
FIR filter taper length L=20
Passband (Θp, Ωp)={300-5000 Hz, 40°-80°}, designed spatial directivity pattern is 1.
Stopband (Θs, Ωs)={300˜5000 Hz, −180°˜30°+90°˜180°}, the designed spatial directivity pattern is 0.
It can be seen that the directivity pattern of the microphone array 201 in the spatial region centered at 60° is enhanced while the sound signals from all other spatial regions are suppressed. The other six spatial regions have similar parameters. Moreover, in all frequencies, the main lobe has the same level, which means the target sound signal has little distortion in frequency.
The microphone array system 200 disclosed herein can be implemented for a square microphone array configuration and a rectangular array configuration where a sound sensor 301 is positioned in each corner of the four-cornered array. The microphone array system 200 disclosed herein implements beamforming from plane to three dimensional sound sources.
For the spatial region centered at 0°:
Passband (Θp, Ωp)={300-4000 Hz, −20°-20°}, designed spatial directivity pattern is 1.
Stopband (Θ, Ωs)={300˜4000 Hz, −180°˜−30°+30°˜180°}, the designed spatial directivity pattern is 0.
For the spatial region centered at 90°:
Passband (Θp, Ωp)={300-4000 Hz, 70°-110°}, designed spatial directivity pattern is 1.
Stopband (Θs, Ωs)={300˜4000 Hz, −180°˜60°+120°˜180°}, the designed spatial directivity pattern is 0. The directivity patterns for the spatial regions centered at −90° and 180° are similarly obtained.
It can be seen from
The foregoing examples have been provided merely for the purpose of explanation and are in no way to be construed as limiting of the present invention disclosed herein. While the invention has been described with reference to various embodiments, it is understood that the words, which have been used herein, are words of description and illustration, rather than words of limitation. Further, although the invention has been described herein with reference to particular means, materials and embodiments, the invention is not intended to be limited to the particulars disclosed herein; rather, the invention extends to all functionally equivalent structures, methods and uses, such as are within the scope of the appended claims. Those skilled in the art, having the benefit of the teachings of this specification, may affect numerous modifications thereto and changes may be made without departing from the scope and spirit of the invention in its aspects.
Claims
1. A method for enhancing a target sound signal from a plurality of sound signals, comprising:
- providing a microphone array system comprising an array of sound sensors positioned in an arbitrary a linear, circular, or other configuration, a sound source localization unit, an adaptive beamforming unit, and a noise reduction unit, wherein said sound source localization unit, said adaptive beamforming unit, and said noise reduction unit are integrated in a digital signal processor, and wherein said sound source localization unit, said adaptive beamforming unit, and said noise reduction unit are in operative communication with said array of said sound sensors;
- receiving said sound signals from a plurality of disparate sound sources by said sound sensors, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- determining a delay between each of said sound sensors and an origin of said array of said sound sensors as a function of distance between each of said sound sensors and said origin, a predefined angle between each of said sound sensors and a reference axis, and an azimuth angle between said reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a two dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for arbitrary numbers of said array of sound sensors and in a plurality of arbitrary configurations of said array of said sound sensors;
- estimating a spatial location of said target sound signal from said received sound signals by said sound source localization unit;
- performing adaptive beamforming for steering a directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal by said adaptive beamforming unit, wherein said adaptive beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and
- suppressing said ambient noise signals by said noise reduction unit for further enhancing said target sound signal.
2. The method of claim 1, wherein said spatial location of said target sound signal from said target sound source is estimated using a steered response power-phase transform by said sound source localization unit.
3. The method of claim 1, wherein said adaptive beamforming comprises:
- providing a fixed beamformer, a blocking matrix, and an adaptive filter in said adaptive beamforming unit;
- steering said directivity pattern of said array of said sound sensors in said direction of said spatial location of said target sound signal from said target sound source by said fixed beamformer for enhancing said target sound signal, when said target sound source is in motion;
- feeding said ambient noise signals to said adaptive filter by blocking said target sound signal received from said target sound source using said blocking matrix; and
- adaptively filtering said ambient noise signals by said adaptive filter in response to detecting one of presence and absence of said target sound signal in said sound signals received from said disparate sound sources.
4. The method of claim 3, wherein said fixed beamformer performs fixed beamforming by filtering and summing output sound signals from said sound sensors.
5. The method of claim 3, wherein said adaptive filtering comprises sub-band adaptive filtering performed by said adaptive filter, wherein said sub-band adaptive filtering comprises:
- providing an analysis filter bank, an adaptive filter matrix, and a synthesis filter bank in said adaptive filter;
- splitting said enhanced target sound signal from said fixed beamformer and said ambient noise signals from said blocking matrix into a plurality of frequency sub-bands by said analysis filter bank;
- adaptively filtering said ambient noise signals in each of said frequency sub-bands by said adaptive filter matrix in response to detecting one of presence and absence of said target sound signal in said sound signals received from said disparate sound sources; and
- synthesizing a full-band sound signal using said frequency sub-bands of said enhanced target sound signal by said synthesis filter bank.
6. The method of claim 3, wherein said adaptive beamforming further comprises detecting said presence of said target sound signal by an adaptation control unit provided in said adaptive beamforming unit and adjusting a step size for said adaptive filtering in response to detecting one of said presence and said absence of said target sound signal in said sound signals received from said disparate sound sources.
7. The method of claim 1, wherein said noise reduction unit performs noise reduction by using one of a Wiener-filter based noise reduction algorithm, a spectral subtraction noise reduction algorithm, an auditory transform based noise reduction algorithm, and a model based noise reduction algorithm.
8. The method of claim 1, wherein said noise reduction unit performs noise reduction in a plurality of frequency sub-bands, wherein said frequency sub-bands are employed by an analysis filter bank of said adaptive beamforming unit for sub-band adaptive beamforming.
9. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors positioned in an arbitrary a linear, circular, or other configuration, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals, by determining a delay between each of said sound sensors and an origin of said array of said sound sensors as a function of distance between each of said sound sensors and said origin, a predefined angle between each of said sound sensors and a reference axis, and an azimuth angle between said reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a two dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for arbitrary numbers of said array of sound sensors and in a plurality of arbitrary configurations of said array of said sound sensors; an adaptive beamforming unit that steers directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal, wherein said adaptive beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals for further enhancing said target sound signal.
10. The system of claim 9, wherein said sound source localization unit estimates said spatial location of said target sound signal from said target sound source using a steered response power-phase transform.
11. The system of claim 9, wherein said adaptive beamforming unit comprises:
- a fixed beamformer that steers said directivity pattern of said array of said sound sensors in said direction of said spatial location of said target sound signal from said target sound source for enhancing said target sound signal, when said target sound source is in motion;
- a blocking matrix that feeds said ambient noise signals to an adaptive filter by blocking said target sound signal received from said target sound source; and
- said adaptive filter that adaptively filters said ambient noise signals in response to detecting one of presence and absence of said target sound signal in said sound signals received from said disparate sound sources.
12. The system of claim 11, wherein said fixed beamformer performs fixed beamforming by filtering and summing output sound signals from said sound sensors.
13. The system of claim 11, wherein said adaptive filter comprises a set of sub-band adaptive filters comprising:
- an analysis filter bank that splits said enhanced target sound signal from said fixed beamformer and said ambient noise signals from said blocking matrix into a plurality of frequency sub-bands;
- an adaptive filter matrix that adaptively filters said ambient noise signals in each of said frequency sub-bands in response to detecting one of presence and absence of said target sound signal in said sound signals received from said disparate sound sources; and
- a synthesis filter bank that synthesizes a full-band sound signal using said frequency sub-bands of said enhanced target sound signal.
14. The system of claim 9, wherein said adaptive beamforming unit further comprises an adaptation control unit that detects said presence of said target sound signal and adjusts a step size for said adaptive filtering in response to detecting one of said presence and said absence of said target sound signal in said sound signals received from said disparate sound sources.
15. The system of claim 9, wherein said noise reduction unit is one of a Wiener-filter based noise reduction unit, a spectral subtraction noise reduction unit, an auditory transform based noise reduction unit, and a model based noise reduction unit.
16. The system of claim 9, further comprising one or more audio codecs that convert said sound signals in an analog form of said sound signals into digital sound signals and reconverts said digital sound signals into said analog form of said sound signals.
17. The system of claim 9, wherein said noise reduction unit performs noise reduction in a plurality of frequency sub-bands employed by an analysis filter bank of said adaptive beamforming unit for sub-band adaptive beamforming.
18. The system of claim 9, wherein said array of said sound sensors is one of a linear array of said sound sensors, a circular array of said sound sensors, and an arbitrarily distributed coplanar array of said sound sensors.
19. The method of claim 1, wherein said delay (τ) is determined by a formula τ=fs*t, wherein fs is a sampling frequency and t is a time delay calculated based on said number of samples within a time period and a time delay for said target sound signal to travel said distance between each of said sound sensors in said microphone array and said origin of said array of said sound sensors, and wherein said distance between said each of said sound sensors in the microphone array and said origin of said array of said sound sensors can be same or different.
20. A method for enhancing a target sound signal from a plurality of sound signals, comprising:
- providing a microphone array system comprising an array of sound sensors positioned in an arbitrary a linear, circular, or other configuration, a sound source localization unit, an adaptive beamforming unit, and a noise reduction unit, wherein said sound source localization unit, said adaptive beamforming unit, and said noise reduction unit are integrated in a digital signal processor, and wherein said sound source localization unit, said adaptive beamforming unit, and said noise reduction unit are in operative communication with said array of said sound sensors;
- receiving said sound signals from a plurality of disparate sound sources by said sound sensors, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- determining a delay between each of said sound sensors and an origin of said array of said sound sensors as a function of distance between each of said sound sensors and said origin, a predefined angle between each of said sound sensors and a first reference axis, an elevation angle between a second reference axis and said target sound signal, and an azimuth angle between said first reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a three dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for arbitrary numbers of said array of sound sensors and in a plurality of arbitrary configurations of said array of said sound sensors;
- estimating a spatial location of said target sound signal from said received sound signals by said sound source localization unit;
- performing adaptive beamforming for steering a directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal by said adaptive beamforming unit, wherein said adaptive beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and
- suppressing said ambient noise signals by said noise reduction unit for further enhancing said target sound signal.
21. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors positioned in an arbitrary a linear, circular, or other configuration, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals by determining a delay between each of said sound sensors and an origin of said array of said sound sensors as a function of distance between each of said sound sensors and said origin, a predefined angle between each of said sound sensors and a first reference axis, an elevation angle between a second reference axis and said target sound signal, and an azimuth angle between said first reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a three dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for arbitrary numbers of said array of sound sensors and in a plurality of arbitrary configurations of said array of said sound sensors; an adaptive beamforming unit that steers directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal, wherein said adaptive beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and
- a noise reduction unit that suppresses said ambient noise signals for further enhancing said target sound signal.
22. A method for enhancing a target sound signal from a plurality of sound signals, comprising:
- providing a microphone array system comprising an array of sound sensors, a sound source localization unit, a beamforming unit, and a noise reduction unit, wherein said sound source localization unit, said beamforming unit, and said noise reduction unit are integrated in a digital signal processor, and wherein said sound source localization unit, said beamforming unit, and said noise reduction unit are in operative communication with said array of said sound sensors;
- receiving said sound signals from a plurality of disparate sound sources by said sound sensors, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point, a predefined angle between each of said sound sensors and a reference axis, and an azimuth angle between said reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a two dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for two or more of said sound sensors;
- estimating a spatial location of said target sound signal from said received sound signals by said sound source localization unit;
- performing beamforming for steering a directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal by said beamforming unit, wherein said beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and
- suppressing said ambient noise signals by said noise reduction unit for further enhancing said target sound signal.
23. The method of claim 22, wherein said beamforming comprises:
- providing a fixed beamformer, a blocking matrix, and an adaptive filter in said beamforming unit;
- steering said directivity pattern of said array of said sound sensors in said direction of said spatial location of said target sound signal from said target sound source by said fixed beamformer for enhancing said target sound signal, when said target sound source is in motion;
- feeding said ambient noise signals to said adaptive filter by blocking said target sound signal received from said target sound source using said blocking matrix; and
- adaptively filtering said ambient noise signals by said adaptive filter in response to detecting one of presence and absence of said target sound signal in said sound signals received from said disparate sound sources.
24. The method of claim 23, wherein said beamforming further comprises detecting said presence of said target sound signal by an adaptation control unit provided in said beamforming unit and adjusting a step size for said adaptive filtering in response to detecting one of said presence and said absence of said target sound signal in said sound signals received from said disparate sound sources.
25. The method of claim 22, wherein said noise reduction unit performs noise reduction in a plurality of frequency sub-bands, wherein said frequency sub-bands are employed by an analysis filter bank of said beamforming unit for sub-band adaptive beamforming.
26. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals, by determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point, a predefined angle between each of said sound sensors and a reference axis, and an azimuth angle between said reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a two dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for two or more of said sound sensors; a beamforming unit that steers directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal, wherein said beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals for further enhancing said target sound signal.
27. The system of claim 26, wherein said beamforming unit further comprises an adaptation control unit that detects said presence of said target sound signal and adjusts a step size for said adaptive filtering in response to detecting one of said presence and said absence of said target sound signal in said sound signals received from said disparate sound sources.
28. The system of claim 26, wherein said noise reduction unit performs noise reduction in a plurality of frequency sub-bands employed by an analysis filter bank of said beamforming unit for sub-band adaptive beamforming.
29. The system of claim 26, wherein said array of said sound sensors is one of a linear array of said sound sensors, and a circular array of said sound sensors, and other types of array of said sound sensors.
30. A method for enhancing a target sound signal from a plurality of sound signals, comprising:
- providing a microphone array system comprising an array of sound sensors, a sound source localization unit, a beamforming unit, and a noise reduction unit, wherein said sound source localization unit, said beamforming unit, and said noise reduction unit are integrated in a digital signal processor, and wherein said sound source localization unit, said beamforming unit, and said noise reduction unit are in operative communication with said array of said sound sensors;
- receiving said sound signals from a plurality of disparate sound sources by said sound sensors, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point, a predefined angle between each of said sound sensors and a first reference axis, an elevation angle between a second reference axis and said target sound signal, and an azimuth angle between said first reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a three dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for two or more of said sound sensors:
- estimating a spatial location of said target sound signal from said received sound signals by said sound source localization unit;
- performing beamforming for steering a directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal by said beamforming unit, wherein said beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and
- suppressing said ambient noise signals by said noise reduction unit for further enhancing said target sound signal.
31. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals by determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point, a predefined angle between each of said sound sensors and a first reference axis, an elevation angle between a second reference axis and said target sound signal, and an azimuth angle between said first reference axis and said target sound signal, when said target sound source that emits said target sound signal is in a three dimensional plane, wherein said delay is represented in terms of number of samples, and wherein said determination of said delay enables beamforming for two or more of said sound sensors; a beamforming unit that steers directivity pattern of said array of said sound sensors in a direction of said spatial location of said target sound signal, wherein said beamforming unit enhances said target sound signal and partially suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals for further enhancing said target sound signal.
32. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals by determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point and an angle of each of said sound sensors biased from a reference axis; a beamforming unit that enhances said target sound signal and suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals.
33. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals by determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point, a predefined angle between each of said sound sensors and a reference axis and an azimuth angle between said reference axis and said target sound signal; a beamforming unit that enhances said target sound signal and suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals.
34. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals by determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point, a predefined angle between each of said sound sensors and a first reference axis, an elevation angle between a second reference axis and said target sound signal and an azimuth angle between said first reference axis and said target sound signal; a beamforming unit that enhances said target sound signal and suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals.
35. A system for enhancing a target sound signal from a plurality of sound signals, comprising:
- an array of sound sensors positioned in a non-circular configuration, wherein said sound sensors receive said sound signals from a plurality of disparate sound sources, wherein said received sound signals comprise said target sound signal from a target sound source among said disparate sound sources, and ambient noise signals;
- a digital signal processor, said digital signal processor comprising: a sound source localization unit that estimates a spatial location of said target sound signal from said received sound signals by determining a delay between each of said sound sensors and a reference point of said array of said sound sensors as a function of distance between each of said sound sensors and said reference point and an angle of each of said sound sensors biased from a reference axis, wherein said distance between each of said sound sensors and said reference point varies from a minimum value to a maximum value, and wherein said minimum value corresponds to zero and said maximum value is defined based on a limitation associated with size of said system; a beamforming unit that enhances said target sound signal and suppresses said ambient noise signals; and a noise reduction unit that suppresses said ambient noise signals.
5315562 | May 24, 1994 | Bradley et al. |
7039199 | May 2, 2006 | Rui |
20030204397 | October 30, 2003 | Amiri et al. |
20040161121 | August 19, 2004 | Chol et al. |
20070076898 | April 5, 2007 | Sarroukh et al. |
20080232607 | September 25, 2008 | Tashev |
20090141907 | June 4, 2009 | Kim et al. |
20090279714 | November 12, 2009 | Kim et al. |
20090304200 | December 10, 2009 | Kim et al. |
20100241426 | September 23, 2010 | Zhang |
20110096915 | April 28, 2011 | Nemer |
20110135125 | June 9, 2011 | Zhan |
20110317522 | December 29, 2011 | Florencio |
WO2008041878 | April 2008 | WO |
WO 2010020162 | February 2010 | WO |
Type: Grant
Filed: Oct 14, 2016
Date of Patent: Sep 18, 2018
Assignee: LI Creative Technologies, Inc. (Florham Park, NJ)
Inventors: Manli Zhu (Pearl River, NY), Qi Li (New Providence, NJ)
Primary Examiner: Ovidio Escalante
Application Number: 15/293,626
International Classification: H04R 25/00 (20060101); H03G 3/20 (20060101); G01S 3/805 (20060101); G01S 3/801 (20060101); H04R 1/40 (20060101); H04R 3/00 (20060101); G01S 5/22 (20060101); H04M 3/56 (20060101);