Google Inc.

Arun Narayanan

Tara N Sainath

Ruoming Pang

Rohit Prakash Prabhavalkar

Yanzhang He

Ehsan Variani

Pedro Jose Moreno Mengibar

Yaniv Leviathan

Michiel A U Bacchiani

Tomer Amiaz

Asaf Aharoni

Neeraj Gaur

Parisa Haghani

Austin Severn Waters

Zhongdi Qu

Galen Tsai Chuang

Nir Shabat

Ron J Weiss

Trevor D Strohman

Chung-Cheng Chiu

Kevin William Wilson

Quan Wang

Joseph Caroselli

Tom O'Malley

Bo Li

Richard Cameron Rose

Izhak Shafran

Wei Han

David Rybach

Jiahui Yu

Rami Botros

Cyril Allauzen

Andrew William Senior

Yonghui Wu

Yu Zhang

Rajeev Conrad Nongpiur

Turaj Zakizadeh Shabestary

Antoine Jean Bruguier

Wei Li

Yiteng Huang

Nathan David Howard

Chanwoo Kim

Ananya Misra

Ke Hu

Shuo-Yiin Chang

Qiao Liang

Anmol Gulati

Ian Mcgraw

Liangliang Cao

Rajeev Rikhye

Shaojin Ding

Shuo-yiin Chang

Thibault Doutre

Alex Seungryong Park

Yedid Hoshen

Min Ma

Michiel Au Bacchiani

James Walker

A method for training a generalized automatic speech recognition model for joint acoustic echo cancellation, speech enhancement, and voice separation includes receiving a plurality of training utterances paired with corresponding training contextual signals. The training contextual signals include a training contextual noise signal including noise prior to the corresponding training utterance, a training reference audio signal, and a training speaker vector including voice characteristics of a target speaker that spoke the corresponding training utterance. The operations also include training, using a contextual signal dropout strategy, a contextual frontend processing model on the training utterances to learn how to predict enhanced speech features. Here, the contextual signal dropout strategy uses a predetermined probability to drop out each of the training contextual signals during training of the contextual frontend processing model.

Generalized automatic speech recognition for joint acoustic echo cancellation, speech enhancement, and voice separation

A computer-implemented method includes receiving a sequence of acoustic frames corresponding to an utterance and generating a reference speaker embedding for the utterance. The method also includes receiving a target speaker embedding for a target speaker and generating feature-wise linear modulation (FiLM) parameters including a scaling vector and a shifting vector based on the target speaker embedding. The method also includes generating an affine transformation output that scales and shifts the reference speaker embedding based on the FiLM parameters. The method also includes generating a classification output indicating whether the utterance was spoken by the target speaker based on the affine transformation output.

Optimizing personal VAD for on-device speech recognition

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for an automated calling system are disclosed. In one aspect, a method includes the actions of receiving audio data of an utterance spoken by a user who is having a telephone conversation with a bot. The actions further include determining a context of the telephone conversation. The actions further include determining a user intent of a first previous portion of the telephone conversation spoken by the user and a bot intent of a second previous portion of the telephone conversation outputted by a speech synthesizer of the bot. The actions further include, based on the audio data of the utterance, the context of the telephone conversation, the user intent, and the bot intent, generating synthesized speech of a reply by the bot to the utterance. The actions further include, providing, for output, the synthesized speech.

Automated calling system

An automated speech recognition (ASR) model includes a first encoder, a second encoder, and a decoder. The first encoder receives, as input, a sequence of acoustic frames, and generates, at each of a plurality of output steps, a first higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The second encoder receives, as input, the first higher order feature representation generated by the first encoder at each of the plurality of output steps, and generates, at each of the plurality of output steps, a second higher order feature representation for a corresponding first higher order feature frame. The decoder receives, as input, the second higher order feature representation generated by the second encoder at each of the plurality of output steps, and generates, at each of the plurality of time steps, a first probability distribution over possible speech recognition hypotheses.

Cascaded encoders for simplified streaming and non-streaming ASR

A method for automatic speech recognition using joint acoustic echo cancellation, speech enhancement, and voice separation includes receiving, at a contextual frontend processing model, input speech features corresponding to a target utterance. The method also includes receiving, at the contextual frontend processing model, at least one of a reference audio signal, a contextual noise signal including noise prior to the target utterance, or a speaker embedding including voice characteristics of a target speaker that spoke the target utterance. The method further includes processing, using the contextual frontend processing model, the input speech features and the at least one of the reference audio signal, the contextual noise signal, or the speaker embedding vector to generate enhanced speech features.

Joint acoustic echo cancelation, speech enhancement, and voice separation for automatic speech recognition

A method includes receiving a sequence of acoustic frames and generating, by a first encoder, a first higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The method also includes generating, by a first pass transducer decoder, a first pass speech recognition hypothesis for a corresponding first higher order feature representation and generating, by a text encoder, a text encoding for a corresponding first pass speech recognition hypothesis. The method also includes generating, by a second encoder, a second higher order feature representation for a corresponding first higher order feature representation. The method also includes generating, by a second pass transducer decoder, a second pass speech recognition hypothesis using a corresponding second higher order feature representation and a corresponding text encoding.

Transducer-based streaming deliberation for cascaded encoders

A computer-implemented method of training a streaming speech recognition model that includes receiving, as input to the streaming speech recognition model, a sequence of acoustic frames. The streaming speech recognition model is configured to learn an alignment probability between the sequence of acoustic frames and an output sequence of vocabulary tokens. The vocabulary tokens include a plurality of label tokens and a blank token. At each output step, the method includes determining a first probability of emitting one of the label tokens and determining a second probability of emitting the blank token. The method also includes generating the alignment probability at a sequence level based on the first probability and the second probability. The method also includes applying a tuning parameter to the alignment probability at the sequence level to maximize the first probability of emitting one of the label tokens.

Fast emit low-latency streaming ASR with sequence-level emission regularization utilizing forward and backward probabilities between nodes of an alignment lattice

Two-pass automatic speech recognition (ASR) models can be used to perform streaming on-device ASR to generate a text representation of an utterance captured in audio data. Various implementations include a first-pass portion of the ASR model used to generate streaming candidate recognition(s) of an utterance captured in audio data. For example, the first-pass portion can include a recurrent neural network transformer (RNN-T) decoder. Various implementations include a second-pass portion of the ASR model used to revise the streaming candidate recognition(s) of the utterance and generate a text representation of the utterance. For example, the second-pass portion can include a listen attend spell (LAS) decoder. Various implementations include a shared encoder shared between the RNN-T decoder and the LAS decoder.

Two-pass end to end speech recognition

An ASR model includes a first encoder configured to receive a sequence of acoustic frames and generate a first higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The ASR model also includes a second encoder configured to receive the first higher order feature representation generated by the first encoder at each of the plurality of output steps and generate a second higher order feature representation for a corresponding first higher order feature frame. The ASR model also includes a decoder configured to receive the second higher order feature representation generated by the second encoder at each of the plurality of output steps and generate a first probability distribution over possible speech recognition hypothesis. The ASR model also includes a language model configured to receive the first probability distribution over possible speech hypothesis and generate a rescored probability distribution.

Efficient streaming non-recurrent on-device end-to-end model

A method for Short-Time Fourier Transform-based echo muting includes receiving a microphone signal including acoustic echo captured by a microphone and corresponding to audio content from an acoustic speaker, and receiving a reference signal including a sequence of frames representing the audio content. For each frame in a sequence of frames, the method includes processing, using an acoustic echo canceler configured to receive a respective frame as input to generate a respective output signal frame that cancels the acoustic echo from the respective frame, and determining, using a Double-talk Detector (DTD), based on the respective frame and the respective output signal frame, whether the respective frame includes a double-talk frame or an echo-only frame. For each respective frame that includes the echo-only frame, muting the respective output signal frame, and performing speech processing on the respective output signal frame for each respective frame that includes the double-talk frame.

STFT-based echo muter

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

Implementations disclosed herein are directed to initializing and utilizing a beamformer in processing of audio data received at a computing device. The computing device can: receive audio data that captures a spoken utterance of a user, determine that a first audio data segment of the audio data includes one or more particular words or phrases; obtain a preceding audio data segment that precedes the first audio data segment; estimate a spatial correlation matrix based on the first audio data segment and based on the preceding audio data segment; initialize the beamformer based on the estimated spatial correlation matrix; and cause the initialized beamformer to be utilized in processing of at least a second audio data segment of the audio data. Additionally, or alternatively, the computing device can transmit the spatial correlation matrix to server(s), and the server(s) can transmit the initialized beamformer back to the computing device.

Context aware beamforming of audio data

This specification describes computer-implemented methods and systems. One method includes receiving, by a neural network of a speech recognition system, first data representing a first raw audio signal and second data representing a second raw audio signal. The first raw audio signal and the second raw audio signal describe audio occurring at a same period of time. The method further includes generating, by a spatial filtering layer of the neural network, a spatial filtered output using the first data and the second data, and generating, by a spectral filtering layer of the neural network, a spectral filtered output using the spatial filtered output. Generating the spectral filtered output comprises processing frequency-domain data representing the spatial filtered output. The method still further includes processing, by one or more additional layers of the neural network, the spectral filtered output to predict sub-word units encoded in both the first raw audio signal and the second raw audio signal.

Enhanced multi-channel acoustic models

Growing community of inventors

Santa Clara, CA, United States of America

Arun Narayanan