Google Inc.

Eugene Weinstein

Pedro Jose Moreno Mengibar

Ignacio Lopez Moreno

Tara N Sainath

Bo Li

Yonghui Wu

Ron J Weiss

Zhifeng Chen

Patrick An Phu Nguyen

Khe Chai Sim

Jeffrey Scott Sorensen

Olivier Siohan

Olga Kapralova

John Paul Alex

Andrew William Senior

Sanjiv Kumar

Johan Schalkwyk

Michael Dennis Riley

Pedro J Mengibar

Austin Severn Waters

Boulos Harb

Anaya Misra

Pavel Golik

Alex Rudnick

Nikhil Prasad Bhat

Methods, systems, and apparatus, including computer programs encoded on a computer-readable media, for speech recognition using multi-dialect and multilingual models. In some implementations, audio data indicating audio characteristics of an utterance is received. Input features determined based on the audio data are provided to a speech recognition model that has been trained to output score indicating the likelihood of linguistic units for each of multiple different language or dialects. The speech recognition model can be one that has been trained using cluster adaptive training. Output that the speech recognition model generated in response to receiving the input features determined based on the audio data is received. A transcription of the utterance generated based on the output of the speech recognition model is provided.

Multi-dialect and multilingual speech recognition

Characteristics of a speaker are estimated using speech processing and machine learning. The characteristics of the speaker are used to automatically customize a user interface of a client device for the speaker.

User interface customization based on speaker characteristics

The characteristics of the speaker may be used to automatically customize a user interface of a client device for the speaker. For instance, a mobile device may generate a user interface that does not permit access to particular applications on the mobile device, that only includes options to call home and access a camera application, or that includes an override option to cause generation of a user interface for a different class of user than a user class for the automatic customizations applied to the user interface.

The present disclosure relates to training a speech recognition system. A system that includes an automated speech recognizer and receives data from a client device. The system determines that at least a portion of the received data is likely sensitive data. Before the at least a portion of the received data is deleted, the system provides the at least a portion of the received data to a model training engine that trains recognition models for the automated speech recognizer. After the at least a portion of the received data is provided, the system deletes the at least a portion of the received data.

Privacy-preserving training corpus selection

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for caching speech recognition scores. In some implementations, one or more values comprising data about an utterance are received. An index value is determined for the one or more values. An acoustic model score for the one or more received values is selected, from a cache of acoustic model scores that were computed before receiving the one or more values, based on the index value. A transcription for the utterance is determined using the selected acoustic model score.

Caching speech recognition scores

The present disclosure relates to training a speech recognition system. One example method includes receiving a collection of speech data items, wherein each speech data item corresponds to an utterance that was previously submitted for transcription by a production speech recognizer. The production speech recognizer uses initial production speech recognizer components in generating transcriptions of speech data items. A transcription for each speech data item is generated using an offline speech recognizer, and the offline speech recognizer components are configured to improve speech recognition accuracy in comparison with the initial production speech recognizer components. The updated production speech recognizer components are trained for the production speech recognizer using a selected subset of the transcriptions of the speech data items generated by the offline speech recognizer. An updated production speech recognizer component is provided to the production speech recognizer for use in transcribing subsequently received speech data items.

Acoustic model training corpus selection

A processing system receives an audio signal encoding a portion of an utterance. The processing system receives context information associated with the utterance, wherein the context information is not derived from the audio signal or any other audio signal. The processing system provides, as input to a neural network, data corresponding to the audio signal and the context information, and generates a transcription for the utterance based on at least an output of the neural network.

Context-based speech recognition

Examples of methods and systems for implementing environmentally aware speech recognition are described. In some examples, a method may be performed by a computing device within a system to adapt an acoustic model for a particular language to one or more environmental conditions. A device may receive one or more spoken utterances and based on the utterances, a system containing the device may determine an acoustic model for the particular language. The system may adapt the acoustic model using one or more data sets depending on the environmental conditions at the location of the device or may obtain another acoustic model that is adapted to the environmental conditions. In some examples, the system may also adapt the acoustic model using one or more data sets based on the voice characteristics of the speaker of the one or more spoken utterances.

Environmentally aware speech recognition

This document describes methods, systems, techniques, and computer program products for generating and/or modifying acoustic models. Acoustic models and/or transformations for a target language/dialect can be generated and/or modified using acoustic models and/or transformations from a source language/dialect.

Generating acoustic models

Systems and methods for identifying music in a noisy environment are described. One of the methods includes receiving audio segment data. The audio segment data is generated from the portion that is captured in the noisy environment. The method further includes generating feature vectors from the audio segment data, identifying phonemes from the feature vectors, and comparing the identified phonemes with pre-assigned phoneme sequences. Each pre-assigned phoneme sequence identifies a known music piece. The method further includes determining an identity of the music based on the comparison.

Growing community of inventors

New York, NY, United States of America

Eugene Weinstein