Google Inc.

Waymo LLC

Ananya Misra

Rajeev Conrad Nongpiur

Chanwoo Kim

Jennifer Taylor

Samuel Liu

Yu Zhang

Ruoming Pang

Arun Narayanan

Chung-Cheng Chiu

Frank Dachille

Jeffrey Scott Sorensen

Wei Han

Sean Rafferty

Liangliang Cao

Zhiyun Lu

Bradley Dodson

Jiajing Wang

Thibault Doutre

Sandro Alighiero Young

Dillon Cower

Min Ma

Ravindran Rajakumar

Megan Teresa Ghastin

Lukasz Kulik

Karan Mahajan

Julia Anna Maria Neidert

Jayant Thatte

The described aspects and implementations enable privacy-respecting detection, separation, and localization of sounds in vehicle environments. The techniques include obtaining, using audio detector(s) of a vehicle, a sound recording that includes a plurality of elemental sounds (ESs) in a driving environment of the vehicle, and processing, using a sound separation model, the sound recording to separate individual ESs of the plurality of ESs. The techniques further include identifying a content of individual ESs and causing a driving path of the vehicle to be modified in view of the identified content of the individual ESs. Further techniques include rendering speech imperceptibly by redacting temporal portions of the speech, using sound recognition models to identify and discard recordings of speech, and driving at speeds that exceed threshold speeds at which speech becomes imperceptible from noise masking.

Privacy-respecting detection and localization of sounds in autonomous driving applications

A system includes a first autonomous vehicle (AV) including a sensing system including a set of sensors, a memory storing instructions and a processing device operatively coupled to the memory, wherein the instructions, when executed by the processing device, cause the processing device to perform operations including observing, using data obtained from the sensing system, an event reflecting a scenario within a driving environment, generating a set of event observation data characterizing the event, and causing the set of event observation data to be shared with at least a second AV that is determined to be in a vicinity of the event after the event is observed.

Real-time multi-vehicle sensing and perception

A method for auralizing a multi-microphone device. Path information for one or more sound paths using dimensions and room reflection coefficients of a simulated room for one of a plurality of microphones included in a multi-microphone device is determined. An array-related transfer functions (ARTFs) for the one of the plurality of microphones is retrieved. The auralized impulse response for the one of the plurality of microphones is generated based at least on the retrieved ARTFs and the determined path information.

Auralization for multi-microphone devices

A method for training a streaming automatic speech recognition student model includes receiving a plurality of unlabeled student training utterances. The method also includes, for each unlabeled student training utterance, generating a transcription corresponding to the respective unlabeled student training utterance using a plurality of non-streaming automated speech recognition (ASR) teacher models. The method further includes distilling a streaming ASR student model from the plurality of non-streaming ASR teacher models by training the streaming ASR student model using the plurality of unlabeled student training utterances paired with the corresponding transcriptions generated by the plurality of non-streaming ASR teacher models.

Streaming automatic speech recognition with non-streaming model distillation

A computer-implemented technique for client/server-based statistical phrase distribution display and text entry includes transmitting a portion of an input entered by a user from a client device to a server having a language model. The input includes a text word or a speech input. The technique further includes receiving responses generated by the language model based on the portion of the input, where each of the responses includes words likely to follow the input, and displaying a tree structure that includes the responses arranged into a plurality of levels. Words in a first portion of each of the responses are included in a first level and are shown on the display. Words in a second portion of each of the responses are included in a second level and subsequent levels and are not shown on the display.

Client/server-based statistical phrase distribution display and associated text entry technique

Growing community of inventors

Mountain View, CA, United States of America

Ananya Misra