Abstract: An automatic speech recognition engine receives an acoustic-echo processed signal from an acoustic-echo processing (AEP) module, where said echo processed signal contains mainly the speech from the near-end talker. The automatic speech recognition engine analyzes the content of the acoustic-echo processed signal to determine whether words or keywords are present. Based upon the results of this analysis, the automatic speech recognition engine produces a value reflecting the likelihood that some words or keywords are detected. Said value is provided to the AEP module. Based upon the value, the AEP module determines if there is double talk and processes the incoming signals accordingly to enhance its performance.

Abstract: An audio device may be configured to produce output audio and to capture input audio for speech recognition. In some cases, a second device may also be used to capture input audio to improve isolation of input audio with respect to the output audio. In addition, acoustic echo cancellation (AEC) may be used to remove components of output audio from input signals of the first and second devices. AEC may be implemented by an adaptive filter based on dynamically optimized filter coefficients. The filter coefficients may be analyzed to detect situations in which the first and second devices are too close to each other, and the user may then be prompted to increase the distance between the two devices.

Abstract: Acoustic signals may be localized such that their position in space is determined. Time-difference-of-arrival data from multiple microphones may be used for this localization. Signal data from the microphones may be degraded by reverberation and other environmental distortions, resulting in erroneous localization. By detecting a portion of the signal resulting from sound directly reaching a microphone rather than from a reverberation, accuracy of the localization is improved.

Abstract: The location of a sound within a given spatial volume may be used in applications such as augmented reality environments. An artificial neural network processes time-difference-of-arrival data (TDOA) from a known microphone array to determine a spatial location of the sound. The neural network may be located locally or available as a cloud service. The artificial neural network is trained with perturbed and non-perturbed TDOA data.

Abstract: A passive projection screen presents images projected thereon by a projection system. A surface of the screen includes elements that are reflective to non-visible light, such as infrared (IR) light. When non-visible light is directed to the screen, the non-visible light is reflected by the reflective elements back. Part of the reflected light may contact and reflect from a user's fingertip or hand (or other object, such as a stylus) while another part is reflected to the projection system. The projection system differentiates among distances to the surface and distances that include the additional travel to the fingertip. As the fingertip moves closer to the surface, the distances approach equality. When the distances are approximately equal, the finger is detected as touching the surface. In this manner, a projection surface equipped with reflective elements facilitates more accurate touch detection.

Abstract: A plurality of microphones of a communication device is grouped into multiple microphone groups, such that each microphone group includes two or more microphones. For each microphone group, output of the corresponding microphones is processed to form an acoustic null in a corresponding spatial direction, such that sound from the corresponding spatial direction is attenuated in the processed output. One of the microphone groups is selected based on various factors leading to maximal echo attenuation and rejection of reverberant components of the room. The selected microphone group is then used to detect sound from a near end talker of the communication device.

Abstract: Techniques for utilizing blind source separation as a front-end to an acoustic echo canceller are described herein. The techniques include removing a first portion of an acoustic echo from an audio signal using blind source separation and a reference signal. The techniques then further remove a second portion of the acoustic echo using an acoustic echo canceller and the reference signal. Further, output of the blind source separation may be used to improve double-talk detection.

Abstract: The location of a sound within a given spatial volume may be used in applications such as augmented reality environments. An artificial neural network processes time-difference-of-arrival data (TDOA) from a known microphone array to determine a spatial location of the sound. The neural network may be located locally or available as a cloud service. The artificial neural network is trained with perturbed and non-perturbed TDOA data.

Abstract: Techniques are provided for sending and receiving key frames and key frame request messages. At a video conference bridge, a key frame request message is received from a first endpoint device. The key frame request message comprises a request for a key frame from a second endpoint device. When a prior key frame request message is received before the key frame request message, a key frame request time value is determined that corresponds to an amount of time between receiving the key frame request message and receiving the prior key frame request message. This value is compared to a threshold time value. When the key frame request time is greater than the threshold time, a key frame request forwarding message is generated, and the key frame request forwarding message is sent to the second endpoint device to request the key frame from the second endpoint device.

Abstract: Techniques are provided for sending and receiving key frames and key frame request messages. At a video conference bridge, a key frame request message is received from a first endpoint device. The key frame request message comprises a request for a key frame from a second endpoint device. When a prior key frame request message is received before the key frame request message, a key frame request time value is determined that corresponds to an amount of time between receiving the key frame request message and receiving the prior key frame request message. This value is compared to a threshold time value. When the key frame request time is greater than the threshold time, a key frame request forwarding message is generated, and the key frame request forwarding message is sent to the second endpoint device to request the key frame from the second endpoint device.

Abstract: An augmented reality environment allows interaction between virtual and real objects. Beamforming techniques are applied to signals acquired by an array of microphones to allow for simultaneous spatial tracking and signal acquisition from multiple users. Localization information such as from other sensors in the environment may be used to select a particular set of beamformer coefficients and resulting beampattern focused on a signal source. Alternately, a series of beampatterns may be used iteratively to localize the signal source in a computationally efficient fashion. The beamformer coefficients may be pre-computed.

Abstract: Acoustic signals may be localized such that their position in space is determined. Time-difference-of-arrival data from multiple microphones may be used for this localization. Signal data from the microphones may be degraded by reverberation and other environmental distortions, resulting in erroneous localization. By detecting a portion of the signal resulting from sound directly reaching a microphone rather than from a reverberation, accuracy of the localization is improved.