Apparatus and method for beamforming to obtain voice and noise signals

Abstract

One method of operation includes beamforming a plurality of microphone outputs to obtain a plurality of virtual microphone audio channels. Each virtual microphone audio channel corresponds to a beamform. The virtual microphone audio channels include at least one voice channel and at least one noise channel. The method includes performing voice activity detection on the at least one voice channel and adjusting a corresponding voice beamform until voice activity detection indicates that voice is present on the at least one voice channel. Another method beamforms the plurality of microphone outputs to obtain a plurality of virtual microphone audio channels, where each virtual microphone audio channel corresponds to a beamform, and with at least one voice channel and at least one noise channel. The method performs voice recognition on the at least one voice channel and adjusts the corresponding voice beamform to improve a voice recognition confidence metric.

Description

FIELD OF THE DISCLOSURE

The present disclosure relates generally to voice processing and more particularly to beamforming systems and methods of applying dual or multi-input noise suppression.

BACKGROUND

Mobile devices such as, but not limited to, mobile phones, smart phones, personal digital assistants (PDAs), tablets, laptops or other electronic devices, etc., increasingly include voice recognition systems to provide hands free voice control of the devices. Although voice recognition technologies have been improving, accurate voice recognition remains a technical challenge when the voice of interest is in the presence of other talkers or ambient noise. These technical challenges exist not only for voice recognition technologies, but also for voice processing such as that used in telephony which today may be performed using almost any electronic device having a suitable telephony application, notwithstanding the prevalence of mobile phones and smart phones.

A particular challenge when implementing voice transmission or voice recognition systems on mobile devices is that many types of mobile devices support use cases where the user (and therefore the user's voice) may be at different positions relative to the mobile device depending on the use case. Adding to the challenge is that various noise sources including other talkers (i.e. jammer voices) may also be located at different positions relative to the mobile device. Some of these noise sources may vary as a function of time in terms of location and magnitude. All of these factors make up the acoustic environment in which a mobile device operates and impacts the sound picked up by the mobile device microphones. Also, as the mobile device is moved or is positioned in certain ways, the acoustic environment of the mobile device also changes accordingly thereby also changing the sound picked up by the mobile device's microphones. Voice sound that may be recognized by the voice recognition system or by a listener on the receiving side of a voice transmission system under one acoustic environment may be unrecognizable under certain changed conditions due to mobile device motion, positioning, or ambient noise levels. Various other conditions in the surrounding environment can add noise, echo or cause other acoustically undesirable conditions that also adversely impact the voice recognition system or voice transmission system.

More specifically, the mobile device acoustic environment impacts the operation of signal processing components such as microphone arrays, noise suppressors, echo cancellation systems and signal conditioning that is used to improve both voice recognition and voice call performance. For mobile devices and also for stationary devices, the speaker and other jammer speakers or other noise sources may also change locations with respect to the device microphones. This also results in undesirable impacts on the acoustic environment and may result in voice being unrecognizable by the voice recognition system or a listener due to noise interference caused by the jammer speakers or other noise sources.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic block diagram of an apparatus in accordance with the embodiments.

FIG. 2 is a flow chart providing an example method of operation of the apparatus of FIG. 1 in accordance with various embodiments.

FIG. 3 is a flow chart showing another example method of operation of the apparatus of FIG. 1 in accordance with various embodiments.

FIG. 4 is a flow chart showing an example method of operation related to formation of a virtual microphone to obtain a voice signal in accordance with various embodiments.

FIG. 5 is a flow chart showing another example method of operation related to formation of a virtual microphone to obtain a voice signal in accordance with various embodiments.

FIG. 6 is a flow chart showing an example method of operation related to formation of a virtual microphone to obtain a noise signal with a jamming voice in accordance with various embodiments.

FIG. 7 is a flow chart showing another example method of operation related to formation of a virtual microphone to obtain a noise signal in accordance with various embodiments.

FIG. 8 is a flow chart showing another example method of operation related to formation of a virtual microphone to obtain a noise signal in accordance with various embodiments.

DETAILED DESCRIPTION

Briefly, a method of operation of the disclosed embodiments includes beamforming a plurality of microphone outputs to obtain a plurality of virtual microphone audio channels. Each virtual microphone audio channel corresponds to a beamform. The virtual microphone audio channels include at least one voice channel and at least one noise channel. The method includes performing voice activity detection on the at least one voice channel and adjusting a corresponding voice beamform until voice activity detection indicates that voice is present on the at least one voice channel.

The method may further include performing voice activity detection on the at least one noise channel and adjusting a corresponding noise beamform until voice activity detection indicates that voice is not substantially present on the at least one noise channel. The method may further include performing energy estimation on the at least one noise channel and adjusting a corresponding noise beamform until energy estimation indicates that the at least one noise channel is receiving audio from a dominant audio energy source. The method may further include performing voice recognition on the at least one voice channel and adjusting the corresponding voice beamform to improve a voice recognition confidence metric of the voice recognition. The method may further include performing voice recognition on the at least one noise channel and adjusting the corresponding noise beamform to decrease a voice recognition confidence metric of the voice recognition performed on the noise beam.

In some embodiments, performing voice recognition on the at least one noise channel may include performing voice recognition on the at least one noise channel using trained voice recognition that is trained to identify a specific speaker. The method may further include configuring the plurality of microphone outputs initially based on a detected orientation of a corresponding group of microphones.

Another method of operation of the disclosed embodiments includes beamforming a plurality of microphone outputs to obtain a plurality of virtual microphone audio channels, where each virtual microphone audio channel corresponds to a beamform, and with at least one voice channel and at least one noise channel. The method includes performing voice recognition on the at least one voice channel and adjusting the corresponding voice beamform to improve a voice recognition confidence metric of the voice recognition.

In some embodiments, performing voice recognition on the at least one voice channel may include performing voice recognition on the at least one voice channel using trained voice recognition that is trained to identify a specific speaker. The method may further include performing voice activity detection on the at least one noise channel and adjusting a corresponding noise beamform until voice activity detection indicates that voice is not substantially present on the at least one noise channel. The method may further include performing energy estimation on the at least one noise channel and adjusting the corresponding noise beamform until energy estimation indicates that the at least one noise channel is receiving audio from a dominant audio energy source. The method may further include performing voice activity detection on the at least one noise channel and adjusting a corresponding noise beamform until voice activity detection indicates that voice is present on the at least one noise channel. The method may further include performing voice recognition on the at least one noise channel and adjusting the corresponding noise beamform to decrease a voice recognition confidence metric of the voice recognition. The method may further include performing voice recognition on the at least one noise channel using trained voice recognition that is trained to identify a specific speaker. The method may further include performing voice recognition on the at least one noise channel in response to voice activity detection indicating that voice is present on the at least one noise channel. The method may further include adjusting the corresponding noise beamform to decrease a voice recognition confidence metric of the trained voice recognition.

The disclosed embodiments also provide an apparatus that includes a beamformer, operatively coupled to a plurality of microphone outputs. The beamformer is operative to provide, as beamformer outputs, a plurality of virtual microphone audio channels where each virtual microphone audio channel corresponds to a beamform and with at least one voice channel and at least one noise channel. A beamformer controller is operatively coupled to the beamformer and is operative to monitor the at least one voice channel and the at least one noise channel to determine if voice is present on either of the at least one voice channel or the at least one noise channel. The beamformer controller is also operative to control the beamformer to adjust a beamform corresponding to the at least one voice channel until voice is present on the at least one voice channel. In some embodiments, the beamformer controller is also operative to control the beamformer to adjust a beamform corresponding to the at least one noise channel until voice is not substantially present on the at least one noise channel.

In one embodiment, a voice activity detector is operatively coupled to the beamformer to receive the at least one voice channel, and to the beamformer controller. The beamformer controller of this embodiment is operative to monitor the at least one voice channel to determine if voice is present by monitoring input received from the voice activity detector. In another embodiment, a voice recognition engine is operatively coupled to the beamformer to receive the at least one voice channel, and to the beamformer controller. The voice recognition engine is operative to perform voice recognition on the at least one voice channel to detect voice, and the beamformer controller is operative to monitor the at least one voice channel to determine if voice is present by monitoring input received from the voice recognition engine. The input may be, for example, voice confidence metrics.

In another embodiment, a voice recognition engine is operatively coupled to the beamformer to receive the at least one voice channel and at least one noise channel. The voice recognition engine is operative to perform voice recognition on the at least one voice channel and at least one noise channel to detect voice. A beamformer controller is operatively coupled to the beamformer, to a voice activity detector and to the voice recognition engine. The beamformer controller is operative to, among other things, monitor the voice activity detector to determine if voice is present on either of the at least one voice channel or the at least one noise channel and control the beamformer to adjust a corresponding voice beamform until voice activity detection or the voice recognition engine indicates that voice is present on the at least one voice channel and adjust a corresponding noise beamform until voice activity detection or the voice recognition engine indicates that voice is not substantially present on the at least one noise channel.

In some embodiments, the apparatus may also include an energy estimator, operatively coupled to the beamformer and to the voice activity detector. In some embodiments, the apparatus may further include microphone configuration logic, operatively coupled to the beamformer. The microphone configuration logic may include switch logic that is operative to switch any microphone output of the plurality of microphone outputs on or off. In some embodiments, the apparatus may also include a noise estimator, operatively coupled to the voice activity detector.

In another embodiment, a method of operation includes beamforming a plurality of microphone outputs to obtain at least one virtual microphone channel, performing voice recognition on the at least one virtual microphone channel, and adjusting a corresponding beamform until voice recognition indicates one of the presence of voice one the at least one virtual microphone channel or that voice is not substantially present on the at least one virtual microphone channel. In some embodiments, performing voice recognition may include performing voice recognition on the at least one virtual microphone channel using trained voice recognition that is trained to identify a specific speaker.

Turning now to the drawings, FIG. 1 is a schematic block diagram of an apparatus 100 in accordance with various embodiments. The apparatus 100 may be incorporated into and used in any electronic device that employs voice-recognition, voice transmission, or voice capture. One application of the apparatus 100 may be used in any of various mobile devices such as, but not limited to, a mobile telephone, smart phone, camera, video camera, tablet, laptop, or some other battery-powered electronic device, etc., however the apparatus 100 is not limited to use in mobile devices. For example, the apparatus 100 may be used in voice controlled television sets, digital video recorders, automobile control systems, or any other device or system that employs voice recognition or voice communication, such as portable or non-portable telephones, speakerphones, etc.

It is to be understood that FIG. 1 is limited, for the purpose of clarity, to showing only those components useful to describe the features and advantages of the various embodiments, and to describe how to make and use the various embodiments to those of ordinary skill. It is therefore to be understood that various other components, circuitry, and devices etc. may be present in order to implement an apparatus and that those various other components, circuitry, devices, etc., are understood to be present by those of ordinary skill. For example, the apparatus may include inputs for receiving power from a power source, a power distribution bus that may be connected to a battery or other power source housed within one of the electronic devices or systems in which the apparatus 100 is incorporated, to provide power to the apparatus 100 or to distribute power to the various components of the apparatus 100. In another example, the apparatus may include one or more communication buses for sending control signals or other information between operatively coupled components, etc. Thus it is to be understood that such various other components, circuitry, or devices are omitted for the purpose of clarity.

Another example is that the apparatus 100 may also include an internal communication bus, for providing operative coupling between the various components, circuitry, and devices. The terminology “operatively coupled” as used herein refers to coupling that enables operational and/or functional communication and relationships between the various components, circuitry, devices etc. described as being operatively coupled and may include any intervening items (i.e. buses, connectors, other components, circuitry, devices etc.) used to enable such communication such as, for example, internal communication buses such as data communication buses or any other intervening items that one of ordinary skill would understand to be present. Also, it is to be understood that other intervening items may be present between “operatively coupled” items even though such other intervening items are not necessary to the functional communication facilitated by the operative coupling. For example, a data communication bus may be present in various embodiments of the apparatus 100 and may provide data to several items along a pathway along which two or more items are operatively coupled, etc. Such operative coupling is shown generally in FIG. 1 described herein.

In FIG. 1 the apparatus 100 may include a group of microphones 110 that provide microphone outputs and that are operatively coupled to microphone configuration logic 120. Although the example of FIG. 1 shows four microphones, with each oriented in a different direction, the embodiments are not limited to four microphones or the example orientations shown and any number of microphones and microphone orientations may be used in the embodiments. It is to be understood that the group of microphones 110 are shown using a dotted line in FIG. 1 because the group of microphones 110 is not necessarily a part of the apparatus 100. In other words, the group of microphones 110 may be part of a mobile device or other electronic device or system into which the apparatus 100 is incorporated. In that case, the apparatus 100 is operatively coupled to the group of microphones 110, which are located within the mobile device, by a suitable communication bus or suitable connectors, etc., such that the group of microphones 110 are operatively coupled to the microphone configuration logic 120.

The microphone configuration logic 120 may include various front end processing, such as, but not limited to, signal amplification, analog-to-digital conversion/digital audio sampling, echo cancellation, etc., which may be applied to the group of microphone 110 outputs prior to performing additional, less power efficient signal processing such as noise suppression. In some embodiments, the microphone configuration logic 120 may also include switch logic operatively coupled to the group of microphones 110 and operative to respond to control signals to individually turn each of the microphones on or off to configure the microphones in various ways. Alternatively, in some embodiments, the microphones may be turned on or off by adjusting a gain or amplifier associated with a corresponding microphone output. For example, a microphone may be turned off by reducing a gain value to zero for the corresponding microphone output. Additionally, in some embodiments, the microphone configuration logic 120 may be operative to receive control signals from other components of the apparatus 100 to adjust front end processing parameters such as, for example, amplifier gain.

The microphone configuration logic 120 is operatively coupled to beamformer 130. In some embodiments, the beamformer 130 may be implemented as a single beamformer with multiple outputs. Each output of the beamformer 130 represents a virtual microphone signal where the virtual microphone is created by beamforming the outputs from one or more physical microphones of the group of microphones 110. In the example embodiment illustrated by FIG. 1, the beamformer 130 is implemented as two or more separate beamformers, beamformer 131 and beamformer 132 in order to increase the speed of operation. Each beamformer 131 and 132 receives inputs from the group of microphones 110 based on the microphone configuration logic 120 or by selecting microphone outputs as needed for given beamform patterns by beamformer controller 190 or by the beamformers independently. That is, in some embodiments, switch logic of microphone configuration logic 120 may switch some or all microphone outputs to beamformer 131 and some or all to beamformer 132 in various combinations and configurations, or in other embodiments the beamformer controller 190, or the beamformers independently, may control which microphones are used as needed for given beamform patterns and may turn microphones on or off by adjusting gain applied within the beamformers. For example, in a mobile device application of the apparatus 100, microphones may be configured by either switch logic, by the beamformer controller 190 or by the beamformers, based on the orientation of the mobile device.

In some embodiments, a device orientation detector 105 is operatively coupled to the microphone configuration logic 120 and to one or more orientation sensors 107. One example of an orientation sensor is a gyroscope, from which the device orientation detector 105 may receive sensor data over connection 106 and determine the positioning of the mobile device. For a given orientation, the device orientation detector 105 may send control signal 108 to the microphone configuration logic 120 to turn off or turn on certain microphones of the group of microphones 110. In other words, various mobile device use cases or mobile device orientations may be associated with certain microphone configurations and such microphone configurations may be triggered by actions taken on the device in conjunction with device orientations. This may be based on pre-determined configuration settings for given orientations in some embodiments, or may be based on other or additional criteria in other embodiments. For example, placing a device in a docking station may trigger engaging a pre-determined microphone configuration. In another example, placing the device in a speakerphone mode and placing the device on a tabletop or desktop may trigger another pre-determined microphone configuration. Thus in some embodiments, the device orientation detector 105, when present, may send orientation information 102 to the beamformer controller 190 such that the beamformer controller 190 may control or override such use case or orientation related settings of the microphone configuration logic 120.

The example apparatus 100 embodiment of FIG. 1 includes two voice detection paths, one for each virtual microphone output of each beamformer 131 and 132. Although the example of FIG. 1 shows two virtual microphone outputs, voice signal 135 and noise signal 136, any number of virtual voice or noise signals may be generated in the various embodiments. In the present example, each of the two virtual microphone outputs is, when needed, provided to a dual input noise suppressor 170. In other embodiments that utilize multiple voice and/or noise signals, a multiple input noise suppresser may be used. In another embodiment, multiple two-input noise suppressors may be used in series to produce a single de-noised output signal. In yet other embodiments, multiple two-input noise suppressors or multiple multi-input noise suppressors may be used in parallel and each output may be sent to the voice recognition engine 180. In such embodiments, whichever output produces the best trained or untrained voice confidence metric may be utilized.

Two symmetrical paths exist between the respective beamformers 131 and 132 and the noise suppressor 170; one for virtual microphone voice signal 135 and one for virtual microphone noise signal 136. The two paths are symmetrical in that they each employ a respective energy estimator 141 and 142 operatively coupled to the beamformers 131 and 132, a respective voice activity detector (VAD) 151 and 152 operatively coupled to the energy estimators 141 and 142, and a noise estimator 161 and 162 operatively coupled to the VAD 151 and 152, respectively. The two noise estimators 161 and 162 are operatively coupled to the noise suppressor 170 to provide respective control signals 149 and 153. The noise estimator 162 receive control signal 143 from VAD 152. The two pathways, including all the components described above, may be considered as a “voice channel” and “noise channel.” That is, a voice signal and a noise signal are sent along the respective pathways through the various components along with control signals between components when appropriate. The voice signal or noise signal may be passed along the pathways and through some of the components without any processing or other action being taken by that component in some embodiments. The voice channel and noise channel are virtual channels that are related to a corresponding voice beamform and noise beamform. The voice beamform may be created by beamformer 131 and the noise beamform may be created by beamformer 132. The voice signal 135 may be considered a voice channel which may also be considered to be one of the virtual microphone outputs. The noise signal 136 may be considered to be noise channel which may also be considered to be another one of the virtual microphone outputs. The “virtual microphones” correspond to beamforms that may incorporate audio from one or more physical microphones of the group of microphones 110. Although FIG. 1 provides an example of one “voice channel” and one “noise channel,” any number of voice channels or noise channels may be created and used in the various embodiments. Also, the various channel components, in some embodiments, may be single integrated components that perform operations for one or more channels. For example, energy estimator 141 and energy estimator 142 may be integrated as a single energy estimator that serves both the voice channel and the noise channel by providing dual inputs or in a time domain multiple access approach or some other suitable approach. The VAD 151 and VAD 152 or the noise estimator 161 and noise estimator 162 may also be implemented in an integrated manner in some embodiments.

Each virtual microphone output is operatively coupled to a respective buffer 133 and 134 which may be a circular buffer to store voice data or noise data while signal examination on the pathways is taking place. That is, signal data may be stored while the signals are being examined to determine if voice is actually present or not in the signals. Thus the signal is buffered as a signal of interest so that if voice or noise is determined to be present the signal can be processed or used accordingly. For example, in some embodiments, voice and noise signals from the beamformers 130 may be buffered and sent to the voice recognition engine 180 while the beamformers 130 continue to adjust beamform patterns to improve the voice and noise signals.

For purposes of explanation, the voice signal 135 pathway will be described first in detail. The symmetrical pathway for the noise signal 136 operates in a similar manner, and any differences will be addressed below. Therefore, beginning with voice signal 135, the energy estimator 141 is operatively coupled to the buffer 133 and to VAD 151. The energy estimator 141 provides a control signal 109 to the buffer 133, a voice and control signal 119 to the VAD 151 and a control signal 111 to the beamformer controller 190. The noise signal 136 energy estimator 142 provides a control signal 113 to buffer 134. In some embodiments, the buffer 133 and buffer 134 may each be controlled by VAD 151 and VAD 152, respectively, and energy estimator 141 and energy estimator 142 may not be present. That is, in some embodiments, VAD 151 and VAD 152 are used to detect voice energy in respective beamform patterns generated by beamformers 130 rather than initially looking for unspecific audio energy as when using the energy estimators. In other embodiments, the VAD may be omitted and, instead, the voice recognition engine 180 and voice confidence metrics alone (without the VAD) may be used as an indicator of the presence of voice in signal. These operations are discussed further herein below with respect to various embodiments and various related methods of operation.

The VAD 151 is further operatively coupled to a noise estimator 161 and provides a voice and control signal 127. The VAD 151 is operatively coupled to the beamformer controller 190 and provides control signal 123 which informs the beamformer controller 190 when the VAD 151 has detected voice. The noise estimator 161 may be a signal-to-noise ratio (SNR) estimator in some embodiments, or may be some other type of noise estimator. The noise estimator 161 is operatively coupled to the beamformer controller 190 and provides control signal 145 which informs the beamformer controller 190 when noise suppression is required for the voice signal 135. In other words, control signal 145 provides information to the beamformer controller 190 which in turn controls the beamformer 131 so that the beamformer 131 may continue to scan or may adjust the beamform pattern in order to reduce some of the noise contained in the voice signal.

Each of the components VAD 151 and 152 and noise estimator 161 and 162, may all be operatively coupled to the respective buffer 133 and buffer 134, to receive buffered voice signal 118 or buffered noise signal 117, respectively. Noise suppressor 170 may be operatively coupled to both buffer 133 and buffer 134 to receive both the buffered voice signal 118 and the buffered noise signal 117. These connections are not shown in FIG. 1 for clarity in showing the various other control connections, etc.

Therefore, noise estimator 161 may receive the buffered voice signal 118 from the buffer 133 and provides control signal 145 to the beamformer controller 190, and voice and control signal 149 to noise suppressor 170. Noise estimator 161 is also operatively coupled to noise estimator 162 by control and data connection 160 such that the two noise estimators can obtain and use information from the other channel to perform various noise estimation operations in some embodiments. The noise suppressor 170 is operatively coupled to the voice recognition engine 180 to provide a noise suppressed voice signal 157, to the beamformer controller 190 to receive control signal 155, and to system memory 103 by read-write connection 173. The noise suppressor 170 may access system memory 103 to read and retrieve noise suppression algorithms, stored in noise suppression algorithms database 171, for execution by the noise suppressor 170. The beamformer controller 190 is operatively coupled to system memory 103 by a read-write connection 193 to access pre-determined beamform patterns stored in a beamform patterns database 191. The system memory 103 is a non-volatile, non-transitory memory.

The noise suppressor 170 may receive the buffered voice signal 118 from the buffer 133 and provide a noise suppressed voice signal 157 to the voice recognition engine 180 and/or to one or more voice transceivers 104 in some embodiments. In some embodiments, the voice recognition engine 180 may not be used and may not be present. That is, in some embodiments, the noise suppressed voice signal 157 may only be provided to one or more voice transceivers 104 for transmission on either by a wired or wireless telecommunications channel or over a wired or wireless network connection if a voice over Internet protocol (VoIP) system is employed by the device into which the apparatus 100 is incorporated. In embodiments having the voice recognition engine 180 present, the voice recognition engine 180 may be operatively coupled to the system control 101, which may be any type of voice controllable system control depending on the device in which the apparatus 100 is incorporated such as, but not limited to, a voice controlled dialer of a mobile telephone, a video recorder system control, an application control of a mobile telephone, smartphone, tablet, laptop, in-vehicle control system, etc., or any other type of voice controllable system control. However, the system control 101 may not be present in all embodiments. The voice recognition engine includes basic voice recognition (VR) logic 181 that recognizes human speech. In some embodiments, the voice recognition engine 180 may additionally, or alternatively, include speaker identification voice recognition logic (SI-VR) 182 which is trained to recognize specific human speech, such as the speech of a specific user.

A control signal 163, sent by the beamformer controller 190, may invoke either the VR logic 181 or the SI-VR logic 182. In response to the control signal 163 instructions, either the VR logic 181 or the SI-VR logic 182 will read either, or both of, the buffered noise signal 117 or buffered voice signal 118. The voice recognition engine 180 will provide a voice-to-text stream with corresponding voice confidence metrics on each phrase or group of words as an indication (i.e. a confidence score) to the beamformer controller 190 of the likelihood of recognized human speech, or the likelihood of a specific user's speech if the SI-VR logic 182 has been invoked. This indication is shown in FIG. 1 as voice confidence metrics 159. The voice recognition engine 180 may also send control signal 165 to the system control 101 in response to detected command words, command phrases or other speech (such as for speech-to-text applications) received on the voice signal 157 or on the buffered voice signal 118 in some embodiments in which the voice recognition engine 180 is also used as a control function for the apparatus 100.

In the various embodiments, the beamformer controller 190 is operative to monitor various control signals which provide various indications of conditions on the voice signal 135 and noise signal 136. In response to the conditions, the beamformer controller 190 is operative to make adjustments to the beamformers 130 to change the beamform directivity. For example, the beamformer controller 190 attempts to adjust the beamformer 131 until the voice signal 135 is substantially the user's voice. Additionally, the beamformer controller 190 attempts to adjust the beamformer 132 until the noise signal 136 is tied to noises and sounds in the acoustic environment of the user other than the user's voice such as a jammer voice or voices or other environmental background noise.

In some embodiments, the formation of a single beamform may be sufficient in some situations. For example, by using a VAD, VR logic 181 or the SI-VR logic 182 (i.e. trained VR) to form a voice beamform channel along with using a noise suppressor may provide sufficient fidelity and de-noising for a given application or for a given acoustic environment. Also, a noise beamform channel using trained VR to substantially eliminate the user's voice and using a noise suppressor may also provide sufficient fidelity and de-noising for a given application or for a given acoustic environment.

The beamformer controller 190 is operative to configure the group of microphones 110 which may be accomplished in some embodiments by controlling the microphone configuration logic 120 to turn microphones on or off according to device orientation detected by device orientation detector 105, or other conditions. In some embodiments, the beamformer controller 190 may generate random beamforms for the voice or noise signal paths where the appropriate signal path components check the results of each. In other embodiments, the beamformer controller 190 may cause the virtual microphone beamforms to change such that the beamforms pan or scan an audio environment until desired conditions are obtained. In yet other embodiments, the beamformer controller 190 may configure the beamformers 130 using pre-determined beamform patterns stored in a beamform patterns database 191 stored in system memory 103. In yet other embodiments, beamformer 131 and beamformer 132 may be adaptive beamformers that are operative to determine the magnitude and phase coefficients needed to combine microphone outputs of the group of microphones 110 in order to steer a beam or a null in a desired direction. In the various embodiments, the beamformer controller 190 is operative to, and may, monitor control signals from any of the following components, in any combination, such as control signal 111 received from energy estimator 141, control signal 115 from energy estimator 142, control signal 123 from VAD 151, control signal 125 from VAD 152, control signal 145 from noise estimator 161 and/or control signal 147 from noise estimator 162. The beamformer controller 190 may also receive voice confidence metrics 159 from the voice recognition engine 180. The beamformer is operative to send a control signal 155 to noise suppressor 170 to invoke noise suppression under certain conditions that are described herein. In some embodiments, the beamformer controller 190 may be integrated into beamformers 130 such that beamformers 130 include all the features of the beamformer controller.

The disclosed embodiments employ VAD 151 and VAD 152 to distinguish voice activity from noise (and vice versa) and accordingly send respective control signals 123 and 125 to the beamformer controller 190. The embodiments also utilize noise estimator 161 and noise estimator 162 to determine when to enable or disable noise reduction if voice cannot be properly distinguished from the signal.

The beamformer 190 accordingly adjusts the beamform directivity of beamformer 131 and beamformer 132 based on energy levels detected by energy estimator 141 and energy estimator 142, voice activity as determined by VAD 151 or VAD 152, and the noise estimators 161 and 162. That is, if the energy level detected exceeds a threshold, the VAD looks for voice. If voice is not detected, the beamformer 190 may adjust the respective beamform pattern. If voice is detected, the noise estimator looks to determine if noise suppression is required or if the signal is sufficient as is. If noise suppression is needed, the beamformer 190 may send control signal 155 to activate the noise suppressor 170 and to perform a voice confidence metric test on the voice signal 157 by the voice recognition engine 180.

Thus, the energy estimators 141 and 142 are operative to detect deviations from a baseline that may be an indicator of voice being present in a received audio signal, or to identify if the beamformers 131 and 132 have a high sensitivity portion of their respective beamforms in a direction of a dominant energy source which may be the primary background noise. If such deviations are detected, the energy estimator 141 may send control signal 119 to activate VAD 151 to determine if voice is actually present in the received audio signal. Short-term deviations exceeding a threshold may also invoke sending control signal 109 to buffer 133 to invoke buffering the signal.

An example method of operation of the apparatus 100 may be understood in view of the flowchart of FIG. 2. The method of operation begins in operation block 201 in which the apparatus 100 uses beamforming to create at least two virtual microphones. One virtual microphone is for the user's voice and the other virtual microphone is for noise. For example as shown in FIG. 1, beamformer 131 outputs the virtual microphone voice signal 135 and beamformer 132 outputs the virtual microphone noise signal 136. In operation block 203, the beamformer controller 190 adjusts one or both of the beamforms to locate dominant energy directions. For example, in some embodiments, the energy estimator 141 may detect an energy level above a threshold and accordingly send the control signal 111 to the beamformer 190 to inform the beamformer controller 190 that a high energy level has been detected. However, in embodiments that do not require the energy estimator 141, the VAD 151 is used to detect voice activity initially instead. Also in some embodiments, a timeout timer may be used such that, if no energy is detected by the energy estimator within a given time period, the beamformer controller 190 may proceed to change the beamform in order to search for a dominant energy source by, for example, employing an adaptive beamformer to determine the magnitude and phase coefficients to steer a beam or a null toward a dominant energy source. In one example of operation, one beamform may be steered in the direction of the user's voice to form the virtual microphone voice channel, and a null may be steered in the direction of the user's voice to form the virtual microphone noise channel.

Acoustic textbook beam-patterns for differential dual-microphone arrays include bidirectional, hyper-cardioid, and cardioid shapes, whose polar patterns have infinite depth nulls. In typical physical systems, the phase and magnitude mismatches between microphone signals are influenced by various factors such as hardware, A/D converter precision, clocking limitations etc. The physical separation distance between microphones and their surrounding structure further reduces the depth of these nulls. In typically realized broad-band signal systems, the null depth of a cardioid pattern may be as little as 10 dB, or as high as 36 dB. Therefore, if a null is directed toward the only jammer talker or noise source present, the expected attenuation of that noise source or jammer could be as least 10 to 12 dB. Note that with perfectly matched microphones and signal processing channels, the attenuation can be much higher. If there are multiple jammer talkers or noise sources oriented in multiple directions, the maximum attenuation realizable with only one steerable null will be less than this 10 to 12 dB value. In one embodiment, in order to form a noise beam, the beamformer controller (190) can steer a null at a desired voice. The desired voice will be attenuated by the aforementioned amounts, and the noise beam will thus be substantially noise. In another embodiment, in order to form a voice beam, the beamformer controller (190) can steer a null at a jammer talker source. The resulting signal will then be substantially voice, having only a small component of jammer signal, as it was attenuated by the aforementioned amount. In yet another embodiment, in the case of a diffused sound field, the beamformer controller (190) can orient a hypercardioid beamform in the direction of a desired talker, thereby forming a signal that is substantially voice due to the −6 dB random energy efficiency of the beam pattern relative to that of an omnidirectional microphone.

In operation block 205, the beamformer controller 190 adjusts at least one beam form until voice is identified on at least one voice virtual microphone signal based on verification by voice activity detection and/or voice recognition confidence metrics. In one example, VAD 151 or VAD 152 will be invoked to determine whether voice is present in the signal or not. For example, if VAD 151 does not detect voice in the signal, then VAD 151 may send control signal 123 to the beamformer controller 190 to indicate that the beamformer controller 190 should re-adapt, or in some other way continue to search for voice by changing the beamform accordingly.

In operation block 207, the beamformer controller 190 adjusts at least a second beamform until either a jammer voice or background noise is identified in at least one noise virtual microphone signal. For example, in one embodiment, VAD 152 may be used to determine whether voice is present in the noise signal 136 or not. In some embodiments, for situations where the VAD 152 detects that voice is present, the VAD 152 may send control signal 125 to beamformer controller 190 to invoke usage of the voice recognition engine 180 to further refine the voice detection. For example, the beamformer controller 190 may send control signal 163 to the voice recognition engine 180 to command the SI-VR 182 logic to analyze the buffered noise signal 117 and determine if any voice detected is that of the user. If the user's voice is detected, based on the voice confidence metrics 159 returned to the beamformer controller 190, the beamformer controller 190 may change the beamform to look for another dominant energy source (i.e. continue to search for noise). If the user's voice is not detected by the SI-VR 182 logic, then in some embodiments the voice activity detected by VAD 152 may be assumed to be jammer voices (i.e. a noise source). Also, if the voice activity detector VAD 152 does not detect voice, then the control signal 125 may indicate to the beamformer controller 190 that only background noise has been detected in the noise signal 136 and that therefore, in either of the above example scenarios the search for a noise source (with either ambient noise, jammer voices, or both) was successful.

In operation block 209, the first and second virtual microphone signals are sent to a dual input noise suppressor. Under certain conditions, the virtual microphone outputs will be sent to the noise suppressor 170. In other words, in some instances, the beamforming of the voice signal 135 may produce an adequately de-noised voice signal such that further noise suppression is not required. The noise estimators 161 and 162 make a determination of whether noise suppression is required or not. That is, the noise estimators 161 and 162 determine whether noise suppression is required for the voice recognition engine 180 to function properly, or if the user's voice will be sufficiently understood by far end listeners (because it has sufficiently little background noise). For example, if voice confidence metrics are too low for the voice signal, then the noise suppressor 170 may need to be applied. In accordance with the embodiments, the beamformed virtual microphone voice signal and the beamformed virtual microphone noise signal are therefore used as inputs to a noise suppressor. That is, once the noise signal 136 is determined to contain only background noise as was described above, or is found to contain a jammer's voice, then the noise signal 136 may be considered adequate for use as an input to the noise suppressor and the beamformer controller 190 will send control signal 155 to noise suppressor 170 to proceed with the dual input noise suppression procedures. The method of operation then ends as shown.

FIG. 3 is a flow chart showing another example method of operation of the apparatus of FIG. 1 in accordance with various embodiments. The method of operation begins and in operation block 301, beamforming is used to create at least two virtual microphones, one for user voice and the other for noise. In operation block 303, the beam forms are adjusted to locate dominant energy source directions. In operation block 305, voice activity detectors are used to identify if voice is present in either signal. In operation block 307, voice recognition confidence metrics are used to identify whether any voice detected is the user's voice or is a jammer voice such that the signal may considered to be noise.

In operation block 309, at least one beamform is adjusted until voice is identified in at least one voice virtual microphone signal based on the voice recognition confidence metrics. In operation block 311, at least a second beamform is adjusted until a jammer voice or background noise is identified in at least one noise virtual microphone signal. In operation block 313, the first and second virtual microphone signals are sent to a dual input noise suppressor, and the method of operation ends as shown.

Further details of operation for obtaining the voice and noise microphone virtual signals and related beamforms are illustrated in FIG. 4 through FIG. 8. Beginning with FIG. 4, a flowchart shows an example method of operation related to formation of a virtual microphone and related beamform to obtain a voice signal in accordance with various embodiments. Initially, the apparatus 100 may determine the orientation of the electronic device or system that incorporates the apparatus 100. For some systems that are relatively stationary, these operations may be omitted since the physical position of the device may be relatively constant. For example, a digital video recorder or television set located in a certain position within a room may remain relatively constant. However, for applications where the apparatus 100 is incorporated into a mobile device, the orientation of the mobile device will change the acoustic environment perceived by the group of microphones 110. Therefore, advantages may be obtained by changing the microphone 110 configuration according to the mobile device orientation. Therefore in some embodiments, the method of operation begins as shown in decision block 401, where device orientation detector 105 may communicate with orientation sensors 107 and obtain the orientation of the device. The orientation information may be sent as orientation information 102 to the beamformer controller 190. In some embodiments, the device orientation detector 105 may send control signal 108 to microphone configuration logic 120 and adjust the microphone configuration accordingly. However, in other embodiments, the beamformer controller 190 will take on this role and will send control signal 194 to microphone configuration logic 120 and change the microphone configuration according to the received orientation information 102. These operations are illustrated in operation block 403.

If orientation information is not available, or is not relevant for the particular device in which the apparatus 100 is incorporated, the method of operation proceeds to operation block 405. In operation block 405, some or all of the microphones, of the group of microphones 110, are combined through the beamformer 130. After the microphone configuration has been selected in either operation block 403 or operation block 405, the method of operation proceeds to decision block 407. The decision of whether noise suppression is required, in decision block 407, is based on the results of the evaluation of noise estimator 161 which evaluates the noise level on the voice signal 135 or the noise level in the user's environment of the signal-to-noise ratio of the user's speech in the user's acoustic environment. If the noise estimator 161 determines that noise suppression is not required in decision block 407, then the control signal 145 will be sent to the beamformer controller 190 to indicate that the current beamform is adequate. In some embodiments, the voice signal may therefore be used for various applications as-is without further noise suppression and the method of operation ends. However, if noise suppression is required in decision block 407, then the resulting noise and voice virtual microphone signals are sent to the noise suppressor 170 in operation block 409.

More particularly, noise estimator 161 sends voice and control signal 149 to the noise suppressor 170. The noise suppressor 170 may obtain the buffered voice signal 118 from buffer 133 and may obtain the buffered noise signal 117 from buffer 134. The noise suppressor 170 may access the system memory 103 over read-write connection 173, and obtain a pertinent noise suppressor algorithm from the noise suppressor algorithms database 171. In some embodiments, the beamformer controller 190 may send the control signal 155 to noise suppressor 170 to indicate a noise suppressor algorithm from the database of noise suppressor algorithms 171 that the noise suppressor 170 should execute.

The noise estimator 161 may check the noise suppressor 170 voice signal 157 to determine if the applied noise suppression algorithm was adequate. If the noise suppression was adequate, and if noise suppression is therefore no longer required in decision block 411, the method of operation ends. However, if noise suppression is still required in decision block 411, then the voice signal 157 may be sent to the voice recognition engine 180. In response, the voice recognition engine will send voice confidence metrics 159 to the beamformer controller 190. If the confidence scores are too low, then the beamformer controller 190 may determine that noise suppression is still required in decision block 415. If the confidence scores are sufficiently high in decision block 415, the noise suppression is no longer required and the method of operation ends. If noise suppression is still required in decision block 415, then the control signal 163 may invoke SI-VR 182 to determine if the user's voice is present in the signal. The method of operation then ends.

In some embodiments, the method of operation illustrated in FIG. 4 may be truncated by omitting operation block 413 and decision block 415 and proceeding from decision block 411 directly to operation block 417. In other words, in some embodiments only the trained speech recognition logic SI-VR 182 is utilized in an attempt to identify the presence of the user's voice in the voice signal. Also, as discussed above with respect to FIG. 2, the trained speech recognition logic SI-VR 182 may also be applied to the noise signal to verify that any voice present in the noise signal is mostly jammer voices and not the user's voice.

FIG. 5 is a flow chart showing another example method of operation related to formation of a virtual microphone and related beamform to obtain a voice signal in accordance with various embodiments. Decision block 501, operation block 503 and operation block 505 involve operations similar to operation blocks 401, 403 and 405 of FIG. 4 and therefore need not be discussed here in detail. Therefore the method of operation proceeds to operation block 507, in which the noise and voice virtual microphone signals are immediately sent to noise suppressor 170. The resulting noise suppressed voice signal 157 is sent to the SI-VR 182 logic in operation block 509. The beamformer controller 190 accordingly receives the voice confidence metrics 159 and determines if further noise suppression is required as shown in decision block 511. If the voice confidence metrics are sufficiently high then the method of operation ends and the voice beamform can be considered adequate. However, if the voice confidence metrics 159 are too low, then this indicates that further noise suppression would be required. The method of operation therefore proceeds to operation block 513. In operation block 513 the beamformer controller sends control signal 194 to the microphone configuration logic 120 and selects a different set of physical microphones from the group of microphones 110 if appropriate. That is, all microphones may already be in use after operation block 503 or operation block 505. In operation block 515, the beamformer controller 190 sends control signal 195 to the beamformer 132 and pans or adapts the beamform or may select a predetermined beamform pattern from system memory 103 in the stored predetermined beamform patterns database 191. This is done in an attempt to steer the peak in sensitivity of the beam toward another location where voice may be detected. Therefore after operation block 515, the method of operation loops back to operation block 507 and the method of operation repeats as shown until success.

FIG. 6 is a flow chart showing an example method of operation related to formation of a virtual microphone and related beamform to obtain a noise signal with a jamming voice in accordance with various embodiments. Operation blocks 601, 603 and 605 are again related to determining a mobile device orientation and have previously been discussed with respect to FIG. 4 and therefore need not be discussed in detail here. The method of operation proceeds to operation block 607, in which some or all of the virtual microphone signals are sent to speech recognition directly. Therefore beamformer controller 190 may send control signal 163 to the voice recognition engine 180 to instruct the voice recognition engine 180 to read the buffered noise signal 117. In decision block 609, the beamformer controller 190 checks the voice confidence metrics 159 to determine if voice appears to be present in any of the signals. Additionally, the beamformer controller 190 may check control signal 125 from VAD 152 to determine if voice activity detection has determined that voice may be present. If voice appears to be present in decision block 609, then the method of operation proceeds to operation block 611 and sends the signal to SI-VR 182 logic. If the user's voice is not detected in decision block 613, based on sufficiently low voice confidence metrics 159, the method of operation ends as shown. However if the voice confidence metrics 159 are high such that the user's voice is likely present, then the method of operation proceeds to operation block 615 in which a different set of physical microphones may be selected if appropriate as was described above, that is, assuming that additional microphones are available (i.e. in situations where only some of the available microphones were initially employed). In operation block 617, the beamformer controller 190 again controls the beamformer 132 to pan or adapt the beamform or selects a predetermined beamformer pattern in order to continue the search for a jammer voice.

FIG. 7 is a flow chart showing another example method of operation related to formation of a virtual microphone and related beamform to obtain a noise signal in accordance with various embodiments. Decision block 701 and operation block 703 and 705 are again similar to other flowcharts in that they are related to determining the orientation of a mobile device and therefore will not be discussed in detail here. The method of operation proceeds to operation block 707 where some or all virtual microphone signals are sent directly to speech recognition, that is, to voice recognition engine 180. In operation block 709, some or all of the virtual microphone signals are also sent to the SI-VR logic 182. In decision block 711, voice activity detectors are checked along with the voice confidence metrics 159 to determine if any of the signals contain voice. If not, then the beamform can be considered to have successfully formed a beam that adequately captures the environmental noise and the method of operation ends as shown. However if voice is detected in decision block 711, then the method of operation proceeds to operation block 713 and a different set of physical microphones is selected if appropriate. In operation block 715, the beamformer controller 190 controls the beamformer 131 and pans or adapts the beamformer, or selects a predetermined beamform pattern from the database of beamformer patterns database 191 stored system memory 103. The method of operation then loops back to operation block 705 and continues until a successful noise beam has been determined.

FIG. 8 is a flow chart showing another example method of operation related to formation of a virtual microphone and related beamform to obtain a noise signal in accordance with various embodiments. Decision block 801 and operation block 803 and operation block 805 are again related to determination of the orientation of a mobile device and will not be discussed herein in detail. The method of operation proceeds to decision block 807 in which the energy estimators are checked to determine whether the virtual microphone signal is in the direction of a dominant energy. If not, the method of operation then proceeds to operation block 815 and may select a different set of physical microphones of the group of microphones 110 if appropriate, that is, assuming that additional microphones are available (i.e. in situations where only some of the available microphones were initially or previously employed). In operation block 817, beamformer controller 190 controls the beamformer 132 to pan or adapt the beamform or selects a predetermined beamform pattern as discussed above, and the method of operation loops back to operation block 805 and continues to search for the noise beam. If the energy estimator determines that the dominant energy source is found in decision block 807, then the method of operation proceeds to operation block 809 where some or all of the virtual microphone signals are sent to the voice recognition engine 180. In operation block 811, some or all of the virtual microphone signals are also sent to the SI-VR logic 182. In decision block 813, if voice is indicated by either the voice activity detector VAD 152, or the voice confidence metrics 159, then the method of operation again proceeds to operation block 815 where a different set of physical microphones may be selected in some situations as discussed above, etc., and in block 817 the beamformer may pan or adapt in the continued search for an environmental noise source. Whenever voice is not indicated in decision block 813, then the beamform can be considered to have successfully captured the environmental noise and the method of operation ends as shown. The noise estimator 162 can send control signal 153 to the noise suppressor 170 when the voice signal and noise signal are to both be sent to the noise suppressor 170. The noise estimator 162 receives control signal 143 from VAD 152.

Thus, in view of the embodiments described in detail above with respect to FIG. 1 and the flowcharts of FIG. 2 through FIG. 8, it is to be understood that various combinations of the components shown in FIG. 1 such as the energy estimators, VADs, noise estimators, voice recognition or speaker identification voice recognition may be used to obtain the features and advantages provided by the present disclosure and that such various combinations are contemplated by the disclosure herein. Also it is to be understood that, in some embodiments, some of the aforementioned components may not be used or may not be present in any particular embodiment. In one example, if VAD is used to detect voice in the voice channel or the noise channel, the voice recognition engine 180 may not be used or may not be present in that embodiment. In another example, if voice recognition is used to detect voice in the voice signal or the noise signal, the VAD 151 or VAD 152 may not be used or may not be present in that embodiment. In either of the above two examples, the energy estimator 141 and energy estimator 142 may not be used or may not be present in either example embodiment. Therefore, based on the description of the embodiments and various examples provide above herein, one of ordinary skill will understand that FIG. 1 contemplates all such various embodiments in view of the present disclosure. Other such contemplated embodiment examples therefore will become apparent to one of ordinary skill in light of the examples and disclosure provided herein.

It is to be understood that the various components, circuitry, devices etc. described with respect to FIG. 1 and the various flowcharts including, but not limited to, those described using the term “logic,” such as the microphone configuration logic 120, beamformers 130, buffers 133 and 134, energy estimators 141 and 142, VAD 151 and 152, noise estimators 161 and 162, noise suppressor 170, voice recognition engine 180, beamformer controller 190, or system control 101 may be implemented in various ways such as by software and/or firmware executing on one or more programmable processors such as a central processing unit (CPU) or the like, or by ASICs, DSPs, FPGAs, hardwired circuitry (logic circuitry), or any combinations thereof.

Also, it is to be understood that the various “control signals” described herein with respect to FIG. 1 and the various aforementioned components, may be implemented in various ways such as using application programming interfaces (APIs) between the various components. Therefore, in some embodiments, components may be operatively coupled using APIs rather than a hardware communication bus if such components are implemented as by software and/or firmware executing on one or more programmable processors. For example, the beamformer controller 190 and the noise suppressor 170 may be software and/or firmware executing on a single processor and may communicate and interact with each other using APIs. In another example, the beamformers 130 and the beamformer controller 190 may be software and/or firmware executing on a single processor and may communicate and interact with each other using APIs. Additional similar examples will be apparent to those of ordinary skill in light of the examples and description provide herein.

Additionally, operations involving the system memory 103 may be implemented using pointers where the components such as, but not limited to, the beamformer controller 190 or the noise suppressor 170, access the system memory 103 as directed by control signals which may include pointers to memory locations or database access commands that access the pre-determined beamform patterns database 191 or the database of noise suppression algorithms 171 or etc., respectively.

It is to be understood that various applications can benefit from the disclosed embodiments, in additions to devices and systems using voice recognition control. For example, the beamforming methods of operations disclosed herein may be used to determine a voice and noise signal for the purpose of identifying a user for a voice uplink channel of a mobile telephone and/or for applying dual or multi-input noise suppression for a voice uplink channel of a mobile telephone. In another example application, a stationary conference call system may incorporate the apparatuses and methods herein described. Other applications of the various disclosed embodiments will be apparent to those of ordinary skill in light of the description and various example embodiments herein described.

While various embodiments have been illustrated and described, it is to be understood that the invention is not so limited. Numerous modifications, changes, variations, substitutions and equivalents will occur to those skilled in the art without departing from the scope of the present invention as defined by the appended claims.

Claims

1. An apparatus comprising: a first voice activity detector, the first voice activity detector configured to generate, based on a voice signal, a first voice control signal to indicate whether a voice is detected in the voice signal;a first noise estimator, the first noise estimator configured to generate, based on a noise level of the voice signal, a first noise control signal to indicate whether to apply noise suppression for the voice signal;a second voice activity detector, the second voice activity detector configured to generate, based on a noise signal, a second voice control signal to indicate whether another voice is detected in the noise signal;a second noise estimator, the second noise estimator configured to generate, based on another noise level of the noise signal, a second noise control signal to indicate whether to apply noise suppression for the noise signal; anda beamformer controller coupled to the first voice activity detector, the first noise estimator, the second voice activity detector, and the second noise estimator, the beamformer controller configured to: adjust, based on the first voice control signal and the first noise control signal, a voice beamform pattern effective to cause the voice to be substantially present in the voice signal;adjust, based on the second voice control signal and the second noise control signal, a noise beamform pattern effective to cause noise to be substantially present in the noise signal; andgenerate, based on the first noise control signal and the second noise control signal, a noise suppression control signal to control noise suppression for the voice signal and the noise signal.

2. The apparatus of claim 1, wherein the beamformer controller is configured to generate, based on the first voice control signal and the second voice control signal, a voice recognition control signal to control voice recognition and speaker identification for the voice signal and the noise signal.

3. The apparatus of claim 2, further comprising: a voice recognition engine coupled to the beamformer controller, the voice recognition engine configured to: detect whether the voice in the voice signal and the other voice in the noise signal are present;responsive to detecting the voice and the other voice, determine whether the voice in the voice signal and the other voice in the noise signal are associated with a same speaker; andgenerate, based on the voice recognition control signal, at least one voice confidence metric that is associated with the detection and the determination, and wherein:the beamformer controller is further configured to: adjust the voice beamform pattern and the noise beamform pattern based on that at least one voice confidence metric; andgenerate the noise suppression control signal based on the at least one voice confidence metric.

4. The apparatus of claim 1, further comprising: a group of microphones having an orientation;an orientation sensor coupled to the beamformer controller, the orientation sensor configured to determine the orientation of the group of microphones, and wherein:the beamformer controller is further configured to adjust the voice beamform pattern and the noise beamform pattern based on the orientation of the group of microphones.

5. The apparatus of claim 4, wherein: the orientation of the group of microphones corresponds to a tabletop orientation in which a side of the apparatus is placed on a tabletop; andthe beamformer controller is further configured to adjust the voice beamform pattern and the noise beamform pattern based on a pre-determined microphone configuration associated with the tabletop orientation.

6. The apparatus of claim 4, wherein: the orientation of the group of microphones corresponds to a docking station orientation in which the apparatus is placed in a docking station; andthe beamformer controller is further configured to adjust the voice beamform pattern and the noise beamform pattern based on a pre-determined microphone configuration associated with the docking station orientation.

7. The apparatus of claim 1, wherein: the adjusting of the voice beamform pattern by the beamformer controller is effective to steer a peak of the voice beamform pattern in a direction of the voice; andthe adjusting of the noise beamform pattern by the beamformer controller is effective to steer a null of the noise beamform pattern in the direction of the voice.

8. The apparatus of claim 1, wherein the noise includes a jammer voice that is not associated with the voice in the voice signal.

9. The apparatus of claim 1, wherein the noise includes environmental background noise.

10. The apparatus of claim 1, wherein the beamformer controller is configured to adjust the voice beamform pattern based on a random set of voice beamform patterns.

11. The apparatus of claim 1, wherein the beamformer controller is configured to scan the voice beamform pattern until the voice is detected in the voice signal by the first voice activity detector.

12. A method comprising: generating at least two voice control signals that respectively indicate whether a voice is present in a voice signal and another voice is present in a noise signal;generating at least two noise control signals that indicate whether to apply noise suppression based on a noise level of the voice signal and another noise level of the noise signal, respectively;adjusting, based on the at least two voice control signals and the at least two noise control signals, a voice beamform pattern effective to cause the voice to be substantially present in the voice signal and a noise beamform pattern effective to cause noise to be substantially present in the noise signal; andgenerating, based on the at least two noise control signals, a noise suppression control signal to activate noise suppression for the voice signal and the noise signal.

13. The method of claim 12, further comprising responsive to the at least two voice control signals indicating a presence of the voice in the voice signal and the other voice in the noise signal, determining whether the voice in the voice signal and the other voice in the noise signal are associated with a same speaker.

14. The method of claim 13, wherein adjusting the voice beamform pattern and the noise beamform pattern is effective to provide higher voice confidence metrics associated with detecting the voice and the other voice and determining whether the voice and the other voice are associated with the same speaker.

15. The method of claim 13, wherein: the voice in the voice signal and the other voice in the noise signal are determined to be associated with the same speaker; andadjusting the noise beamform pattern comprises adjusting the noise beamform pattern effective to cause the other voice to not be substantially present in the noise signal.

16. The method of claim 13, wherein: the voice in the voice signal and the other voice in the noise signal are associated with different speakers;adjusting the voice beamform pattern comprises adjusting a null of the voice beamform pattern in a direction of the other voice; andadjusting the noise beamform pattern comprises adjusting a peak of the noise beamform pattern in the direction of the other voice.

17. The method of claim 12, further comprising: determining an orientation of a group of microphones that produce the voice beamform pattern and the noise beamform pattern; andadjusting the voice beamform pattern and the noise beamform pattern based on the orientation of the group of microphones.

18. A beamformer controller coupled to a first voice activity detector of a voice channel, a first noise estimator of the voice channel, a second voice activity detector of a noise channel, a second noise estimator of the noise channel, a noise suppressor, a voice recognition engine, a voice beamformer, and a noise beamformer, the beamformer controller configured to: receive, from the first voice activity detector, a first voice control signal that indicates whether a voice is detected in a voice signal;receive, from the first noise estimator, a first noise control signal that indicates a noise level of the voice signal;receive, from the second voice activity detector, a second voice control signal that indicates whether another voice is detected in a noise signal;receive, from the second noise estimator, a second noise control signal that indicates another noise level of the noise signal;adjust, based on the first voice control signal and the first noise control signal, a voice beamform pattern for the voice signal via the voice beamformer;adjust, based on the second voice control signal and the second noise control signal, a noise beamform pattern for the noise signal via the noise beamformer;generate, based on the first noise control signal and the second noise control signal, a noise suppression control signal to control whether the noise suppressor is to perform noise suppression using the voice signal and the noise signal; andgenerate a voice recognition control signal based on the first voice control signal and the second voice control signal to control whether the voice recognition engine is to perform voice recognition for the voice signal and the noise signal.

19. The beamformer controller of claim 18, wherein the adjusting of the voice beamform pattern, the adjusting of the noise beamform pattern, the generating of the noise suppression control signal, and the generating of the voice recognition control signal are effective to enable the beamformer controller to respond to a dynamic acoustic environment that changes based on orientation of a group of microphones, motion of the group of microphones, and environmental noise.

20. The beamformer controller of claim 18, further configured to: receive, from a first energy estimator, a first energy control signal that indicates an energy level of the voice signal;receive, from a second energy estimator, a second energy control signal that indicates another energy level of the noise signal;adjust a directivity of the voice beamformer based on the energy level of the voice signal; andadjust another directivity of the noise beamformer based on the other energy level of the noise signal.