Nvidia Corporation

Ambrish Dantrey

Revanth Reddy Nalla

Abhijit Patait

Mihir Nyayate

Angshuman Ghosh

Anshul Gupta

Niranjan Avadhanam

Atousa Torabi

Ram Ganapathi

Prakshep Mehta

Jithin Thomas

Anshul Jain

Neilesh Chorakhalikar

Utkarsh Patankar

Apparatuses, systems, and techniques are presented to reduce noise in audio. In at least one embodiment, a sequence of neural networks is used to remove foreground and background noise from audio including a primary audio signal.

Audio noise removal using one or more neural networks

In various examples, audio alerts of emergency response vehicles may be detected and classified using audio captured by microphones of an autonomous or semi-autonomous machine in order to identify travel directions, locations, and/or types of emergency response vehicles in the environment. For example, a plurality of microphone arrays may be disposed on an autonomous or semi-autonomous machine and used to generate audio signals corresponding to sounds in the environment. These audio signals may be processed to determine a location and/or direction of travel of an emergency response vehicle (e.g., using triangulation). Additionally, to identify siren types—and thus emergency response vehicle types corresponding thereto—the audio signals may be used to generate representations of a frequency spectrum that may be processed using a deep neural network (DNN) that outputs probabilities of alert types being represented by the audio data.

Emergency response vehicle detection for autonomous driving applications

Apparatuses, systems, and techniques are presented to reduce noise in audio. In at least one embodiment, one or more neural networks are used to determine a noise signal in one or more speech signals.

Audio noise determination using one or more neural networks

In various examples, game session audio data—e.g., representing speech of users participating in the game—may be monitored and/or analyzed to determine whether inappropriate language is being used. Where inappropriate language is identified, the portions of the audio corresponding to the inappropriate language may be edited or modified such that other users do not hear the inappropriate language. As a result, toxic behavior or language within instances of gameplay may be censored—thereby enhancing the user experience and making online gaming environments safer for more vulnerable populations. In some embodiments, the inappropriate language may be reported—e.g., automatically—to the game developer or game application host in order to suspend, ban, or otherwise manage users of the system that have a proclivity for toxic behavior.

Automatic classification and reporting of inappropriate language in online applications

In various examples, audio alerts of emergency response vehicles may be detected and classified using audio captured by microphones of an autonomous or semi-autonomous machine in order to identify travel directions, locations, and/or types of emergency response vehicles in the environment. For example, a plurality of microphone arrays may be disposed on an autonomous or semi-autonomous machine and used to generate audio signals corresponding to sounds in the environment. These audio signals may be processed to determine a location and/or direction of travel of an emergency response vehicle (e.g., using triangulation). Additionally, to identify siren types&#x2014;and thus emergency response vehicle types corresponding thereto&#x2014;the audio signals may be used to generate representations of a frequency spectrum that may be processed using a deep neural network (DNN) that outputs probabilities of alert types being represented by the audio data. The locations, direction of travel, and/or siren type may allow an ego-vehicle or ego-machine to identify an emergency response vehicle and to make planning and/or control decisions in response.

A virtual reality (VR) audio rendering system and method include spatializing microphone-captured real-world sounds according to a VR setting. In a game streaming system, when a player speaks through a microphone, the voice is processed by geometrical acoustic (GA) simulation configured for a virtual scene, and thereby spatialized audio effects specific to the scene are added. The GA simulation may include generating an impulse response using sound propagation simulation and dynamic HRTF-based listener directivity. When the GA-processed voice of the player is played, the local player or other fellow players can hear it as if the sound travels in the scenery and according to the geometries in the virtual scene. This mechanism can advantageously place the players' chatting in the same virtual world like built-in game audio, thereby advantageously providing enhanced immersive VR experience to users.

Application of geometric acoustics for immersive virtual reality (VR)

In various examples, game session audio data&#x2014;e.g., representing speech of users participating in the game&#x2014;may be monitored and/or analyzed to determine whether inappropriate language is being used. Where inappropriate language is identified, the portions of the audio corresponding to the inappropriate language may be edited or modified such that other users do not hear the inappropriate language. As a result, toxic behavior or language within instances of gameplay may be censored&#x2014;thereby enhancing the user experience and making online gaming environments safer for more vulnerable populations. In some embodiments, the inappropriate language may be reported&#x2014;e.g., automatically&#x2014;to the game developer or game application host in order to suspend, ban, or otherwise manage users of the system that have a proclivity for toxic behavior.

A virtual reality (VR) audio rendering system and method of using HRTF functions to quickly capture new positional cues to pre-computed audio frames responsive to changes in user position relative to sound systems. In a client-server VR system, when a user position change is detected, the client determines an appropriate HRTF based on the new position and convolves them with a set of audio frames that have been generated by the server based on a prior position, resulting in modified frames for rendering. Meanwhile, the client propagates the new position to the server to generate subsequent audio frames for the corrected position. As HRTF convolution is computationally inexpensive, the latency between user position change and the resultant sound change as perceived by the user can be significantly reduced. As a result, an immersive VR experience of the user can be preserved.

Method and system for immersive virtual reality (VR) streaming with reduced audio latency

A virtual reality (VR) audio rendering system and method using pre-computed impulse responses (IRs) to generate audio frames in a VR setting for rendering. Based on a current position of a user or a VR object, a set of possible motions are predicted and a set of IRs are pre-computed by using a Geometric Acoustic (GA) model of a virtual scene. Once a position change is actually detected, one of the pre-computed IRs is selected and convolved with a set of audio frames to generate modified audio frames for rendering. As the modified audio frames are generated by using pre-computed IR without requiring intensive ray tracing computations, the audio latency can be significantly reduced.

Method and system for immersive virtual reality (VR) streaming with reduced geometric acoustic audio latency

A method includes implementing an audio framework to be executed on a data processing device with a virtual audio driver component and a User Mode Component (UMC) communicatively coupled to each other. The virtual audio driver component enables modifying an original default audio endpoint device of an application executing on the data processing device to an emulated audio device associated with a new audio endpoint in response to an initiation through the application in conjunction with the UMC. The virtual audio driver component also enables registering the new audio endpoint as the modified default audio endpoint with an operating system executing on the data processing device. Further, the virtual audio driver component enables capturing audio data intended for the original default audio endpoint device at the new audio endpoint following the registration thereof to enable control of the audio data.

Growing community of inventors

Pune, India

Ambrish Dantrey