Google Inc.

Other

Ignacio Lopez Moreno

Diego Melendo Casado

Quan Wang

Pu-sen Chao

Li Wan

Javier Gonzalez-Dominguez

Andrew William Senior

Alexander Houston Gruenstein

Eugene Weinstein

Jason William Pelecanos

Georg Heigold

Prashant Sridhar

Samuel Bengio

Taral Pradeep Joglekar

Matthew Nirvan Sharifi

Pedro Jose Moreno Mengibar

Yonghui Wu

Yu Zhang

Ron J Weiss

Zhifeng Chen

Ruoming Pang

Aleksandar Kracun

Christopher Thaddeus Hughes

Ye Jia

Jonathan Shen

Meltem Oktem

William Zhang

Fei Ren

Hakan Erdogan

Ludwig Schmidt

Fnu Heryandi

Salil Rajadhyaksha

Hasim Sak

Patrick An Phu Nguyen

Ágoston Weisz

Olivier Siohan

Olga Kapralova

Luis Carlos Cobo Rus

Hyun Jin Park

Yu-Hsin Chen

Hannah Muckenhirn

Yang Yu

John Paul Alex

Sanjiv Kumar

Samy Bengio

Alexandru Ovidiu Dovlecel

Jin Shi

Xin Lei

Ehsan Variani

Erik Mcdermott

Anshuman Tripathi

Han Lu

Niranjan A Subrahmanya

Patrick An Nguyen

Gang Feng

Roza Chojnacka

Yiling Huang

Lukas Lopatovsky

Fang Chu

Alex Seungryong Park

Wei Xia

Evan Clark

Pai Zhu

Hannah Muckenhim

Alan Sean Papir

Nikhil Prasad Bhat

Yeming Fang

Yash Sheth

Mert Saglam

Pu-Sen Chao

Alexandru Dovlecel

A method includes receiving an input audio signal that corresponds to utterances spoken by multiple speakers. The method also includes processing the input audio to generate a transcription of the utterances and a sequence of speaker turn tokens each indicating a location of a respective speaker turn. The method also includes segmenting the input audio signal into a plurality of speaker segments based on the sequence of speaker tokens. The method also includes extracting a speaker-discriminative embedding from each speaker segment and performing spectral clustering on the speaker-discriminative embeddings to cluster the plurality of speaker segments into k classes. The method also includes assigning a respective speaker label to each speaker segment clustered into the respective class that is different than the respective speaker label assigned to the speaker segments clustered into each other class of the k classes.

Speaker-turn-based online speaker diarization with constrained spectral clustering

Techniques are disclosed that enable processing of audio data to generate one or more refined versions of audio data, where each of the refined versions of audio data isolate one or more utterances of a single respective human speaker. Various implementations generate a refined version of audio data that isolates utterance(s) of a single human speaker by processing a spectrogram representation of the audio data (generated by processing the audio data with a frequency transformation) using a mask generated by processing the spectrogram of the audio data and a speaker embedding for the single human speaker using a trained voice filter model. Output generated over the trained voice filter model is processed using an inverse of the frequency transformation to generate the refined audio data.

Targeted voice separation by speaker conditioned on spectrogram masking

Techniques disclosed herein enable training and/or utilizing speaker dependent (SD) speech models which are personalizable to any user of a client device. Various implementations include personalizing a SD speech model for a target user by processing, using the SD speech model, a speaker embedding corresponding to the target user along with an instance of audio data. The SD speech model can be personalized for an additional target user by processing, using the SD speech model, an additional speaker embedding, corresponding to the additional target user, along with another instance of audio data. Additional or alternative implementations include training the SD speech model based on a speaker independent speech model using teacher student learning.

Speaker awareness using speaker dependent speech model(s)

The technology described in this document can be embodied in a computer-implemented method that includes receiving, at a processing system, a first signal including an output of a speaker device and an additional audio signal. The method also includes determining, by the processing system, based at least in part on a model trained to identify the output of the speaker device, that the additional audio signal corresponds to an utterance of a user. The method further includes initiating a reduction in an audio output level of the speaker device based on determining that the additional audio signal corresponds to the utterance of the user.

Recognizing speech in the presence of additional audio

Processing of acoustic features of audio data to generate one or more revised versions of the acoustic features, where each of the revised versions of the acoustic features isolates one or more utterances of a single respective human speaker. Various implementations generate the acoustic features by processing audio data using portion(s) of an automatic speech recognition system. Various implementations generate the revised acoustic features by processing the acoustic features using a mask generated by processing the acoustic features and a speaker embedding for the single human speaker using a trained voice filter model. Output generated over the trained voice filter model is processed using the automatic speech recognition system to generate a predicted text representation of the utterance(s) of the single human speaker without reconstructing the audio data.

Targeted voice separation by speaker for speech recognition

Implementations relate to determining a language for speech recognition of a spoken utterance, received via an automated assistant interface, for interacting with an automated assistant. Implementations can enable multilingual interaction with the automated assistant, without necessitating a user explicitly designate a language to be utilized for each interaction. Selection of a speech recognition model for a particular language can based on one or more interaction characteristics exhibited during a dialog session between a user and an automated assistant. Such interaction characteristics can include anticipated user input types, anticipated user input durations, a duration for monitoring for a user response, and/or an actual duration of a provided user response.

Automatically determining language for speech recognition of spoken utterance received via an automated assistant interface

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech recognition using neural networks. A feature vector that models audio characteristics of a portion of an utterance is received. Data indicative of latent variables of multivariate factor analysis is received. The feature vector and the data indicative of the latent variables is provided as input to a neural network. A candidate transcription for the utterance is determined based on at least an output of the neural network.

Speech recognition using neural networks

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech synthesis. The methods, systems, and apparatus include actions of obtaining an audio representation of speech of a target speaker, obtaining input text for which speech is to be synthesized in a voice of the target speaker, generating a speaker vector by providing the audio representation to a speaker encoder engine that is trained to distinguish speakers from one another, generating an audio representation of the input text spoken in the voice of the target speaker by providing the input text and the speaker vector to a spectrogram generation engine that is trained using voices of reference speakers to generate audio representations, and providing the audio representation of the input text spoken in the voice of the target speaker for output.

Synthesis of speech from text in a voice of a target speaker using neural networks

Techniques are disclosed that enable determining and/or utilizing a misrecognition of a spoken utterance, where the misrecognition is generated using an automatic speech recognition (ASR) model. Various implementations include determining a recognition based on the spoken utterance and a previous utterance spoken prior to the spoken utterance. Additionally or alternatively, implementations include personalizing an ASR engine for a user based on the spoken utterance and the previous utterance spoken prior to the spoken utterance (e.g., based on audio data capturing the previous utterance and a text representation of the spoken utterance).

Identification and utilization of misrecognitions in automatic speech recognition

Text independent speaker recognition models can be utilized by an automated assistant to verify a particular user spoke a spoken utterance and/or to identify the user who spoke a spoken utterance. Implementations can include automatically updating a speaker embedding for a particular user based on previous utterances by the particular user. Additionally or alternatively, implementations can include verifying a particular user spoke a spoken utterance using output generated by both a text independent speaker recognition model as well as a text dependent speaker recognition model. Furthermore, implementations can additionally or alternatively include prefetching content for several users associated with a spoken utterance prior to determining which user spoke the spoken utterance.

Text independent speaker recognition

This document generally describes systems, methods, devices, and other techniques related to speaker verification, including (i) training a neural network for a speaker verification model, (ii) enrolling users at a client device, and (iii) verifying identities of users based on characteristics of the users' voices. Some implementations include a computer-implemented method. The method can include receiving, at a computing device, data that characterizes an utterance of a user of the computing device. A speaker representation can be generated, at the computing device, for the utterance using a neural network on the computing device. The neural network can be trained based on a plurality of training samples that each: (i) include data that characterizes a first utterance and data that characterizes one or more second utterances, and (ii) are labeled as a matching speakers sample or a non-matching speakers sample.

Neural networks for speaker verification

A method for training a neural network includes receiving a training input audio sequence including a sequence of input frames defining a hotword that initiates a wake-up process on a user device. The method further includes obtaining a first label and a second label for the training input audio sequence. The method includes generating, using a memorized neural network and the training input audio sequence, an output indicating a likelihood the training input audio sequence includes the hotword. The method further includes determining a first loss based on the first label and the output. The method includes determining a second loss based on the second label and the output. The method further includes optimizing the memorized neural network based on the first loss and the second loss associated with the training input audio sequence.

Mixing heterogeneous loss types to improve accuracy of keyword spotting

Determining a language for speech recognition of a spoken utterance received via an automated assistant interface for interacting with an automated assistant. Implementations can enable multilingual interaction with the automated assistant, without necessitating a user explicitly designate a language to be utilized for each interaction. Implementations determine a user profile that corresponds to audio data that captures a spoken utterance, and utilize language(s), and optionally corresponding probabilities, assigned to the user profile in determining a language for speech recognition of the spoken utterance. Some implementations select only a subset of languages, assigned to the user profile, to utilize in speech recognition of a given spoken utterance of the user. Some implementations perform speech recognition in each of multiple languages assigned to the user profile, and utilize criteria to select only one of the speech recognitions as appropriate for generating and providing content that is responsive to the spoken utterance.

Teacher-student learning can be used to train a keyword spotting (KWS) model using augmented training instance(s). Various implementations include aggressively augmenting (e.g., using spectral augmentation) base audio data to generate augmented audio data, where one or more portions of the base instance of audio data can be masked in the augmented instance of audio data (e.g., one or more time frames can be masked, one or more frequencies can be masked, etc.). Many implementations include processing augmented audio data using a KWS teacher model to generate a soft label, and processing the augmented audio data using a KWS student model to generate predicted output. One or more portions of the KWS student model can be updated based on a comparison of the soft label and the generated predicted output.

Noisy student teacher training for robust keyword spotting

Growing community of inventors

New York, NY, United States of America

Ignacio Lopez Moreno