Google Inc.

David Rybach

Tara N Sainath

Ruoming Pang

Trevor D Strohman

Ehsan Variani

Arun Narayanan

Yanzhang He

Rami Botros

Basilio Garcia Castillo

Cyril Allauzen

Bhuvana Ramabhadran

Bo Li

Rohit Prakash Prabhavalkar

Michiel A U Bacchiani

Neeraj Gaur

Weiran Wang

Tongzhou Chen

Yangzhang He

Charles Caleb Peyser

Ronny Huang

Zelin Wu

Ding Zhao

Deepti Bhatia

Sepand Mavandadi

Basi Garcia

A method includes receiving a training example that includes audio data representing a spoken utterance and a ground truth transcription. For each word in the spoken utterance, the method also includes inserting a placeholder symbol before the respective word identifying a respective ground truth alignment for a beginning and an end of the respective word, determining a beginning word piece and an ending word piece, and generating a first constrained alignment for the beginning word piece and a second constrained alignment for the ending word piece. The first constrained alignment is aligned with the ground truth alignment for the beginning of the respective word and the second constrained alignment is aligned with the ground truth alignment for the ending of the respective word. The method also includes constraining an attention head of a second pass decoder by applying the first and second constrained alignments.

Emitting word timings with end-to-end models

A method includes generating, using an audio encoder, a higher-order feature representation for each acoustic frame in a sequence of acoustic frames; generating, using a decoder, based on the higher-order feature representation, a plurality of speech recognition hypotheses, each hypotheses corresponding to a candidate transcription of an utterance and having an associated first likelihood score; generating, using an external language model, for each speech recognition hypothesis, a second likelihood score; determining, using a learnable fusion module, for each speech recognition hypothesis, a set of fusion weights based on the higher-order feature representation and the speech recognition hypothesis; and generating, using the learnable fusion module, for each speech recognition hypothesis, a third likelihood score based on the first likelihood score, the second likelihood score, and the set of fusion weights, the audio encoder and decoder trained using minimum additive error rate training in the presence of the external language model.

Rare word recognition with LM-aware MWER training

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 includes receiving audio data encoding an utterance, processing, using a speech recognition model, the audio data to generate speech recognition scores for speech elements, and determining context scores for the speech elements based on context data indicating a context for the utterance. The method also includes executing, using the speech recognition scores and the context scores, a beam search decoding process to determine one or more candidate transcriptions for the utterance. The method also includes selecting a transcription for the utterance from the one or more candidate transcriptions.

Using context information with end-to-end models for speech recognition

The technology described herein can be embodied in a method that includes receiving an audio signal encoding a portion of an utterance, and providing, to a first neural network, data corresponding to the audio signal. The method also includes generating, by a processor, data representing a transcription for the utterance based on an output of the first neural network. The first neural network is trained using features of multiple context-dependent states, the context-dependent states being derived from a plurality of context-independent states provided by a second neural network.

Context-dependent state tying using a neural network

Growing community of inventors

Aachen, Germany

David Rybach