Google Inc.

Rami Botros

Tara N Sainath

Trevor D Strohman

Ruoming Pang

Arun Narayanan

Yanzhang He

Ehsan Variani

David Rybach

Weiran Wang

Cyril Allauzen

Rohit Prakash Prabhavalkar

Jiahui Yu

Qiao Liang

Anmol Gulati

Krzysztof Marcin Choromanski

Ian Mcgraw

Yangzhang He

Shaojin Ding

Ding Zhao

Xin Wang

Dongseong Hwang

Robert David

Rina Panigrahy

An automated speech recognition (ASR) model includes a first encoder, a first encoder, a second encoder, and a second 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 first decoder receives, as input, the first higher order feature representation generated by the first encoder, and generates a first probability distribution over possible speech recognition hypotheses. The second encoder receives, as input, the first higher order feature representation generated by the first encoder, and generates a second higher order feature representation for a corresponding first higher order feature frame. The second decoder receives, as input, the second higher order feature representation generated by the second encoder, and generates a second probability distribution over possible speech recognition hypotheses.

Unified cascaded encoder ASR model for dynamic model sizes

A computer-implemented method includes receiving a sequence of acoustic frames as input to an automatic speech recognition (ASR) model. Here, the ASR model includes a causal encoder and a decoder. The method also includes generating, by the causal encoder, a first higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The method also includes generating, by the decoder, a first probability distribution over possible speech recognition hypotheses. Here, the causal encoder includes a stack of causal encoder layers each including a Recurrent Neural Network (RNN) Attention-Performer module that applies linear attention.

Optimizing inference performance for conformer

A recurrent neural network-transducer (RNN-T) model improves speech recognition by processing sequential non-blank symbols at each time step after an initial one. The model's prediction network receives a sequence of symbols from a final Softmax layer and employs a shared embedding matrix to create and map embeddings to each symbol, associating them with unique position vectors. These embeddings are weighted according to their similarity to their matching position vector. Subsequently, a joint network of the RNN-T model uses these weighted embeddings to output a probability distribution for potential speech recognition hypotheses at each time step, enabling more accurate transcriptions of spoken language.

Tied and reduced RNN-T

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 RNN-T model includes a prediction network configured to, at each of a plurality of times steps subsequent to an initial time step, receive a sequence of non-blank symbols. For each non-blank symbol the prediction network is also configured to generate, using a shared embedding matrix, an embedding of the corresponding non-blank symbol, assign a respective position vector to the corresponding non-blank symbol, and weight the embedding proportional to a similarity between the embedding and the respective position vector. The prediction network is also configured to generate a single embedding vector at the corresponding time step. The RNN-T model also includes a joint network configured to, at each of the plurality of time steps subsequent to the initial time step, receive the single embedding vector generated as output from the prediction network at the corresponding time step and generate a probability distribution over possible speech recognition hypotheses.

Growing community of inventors

Mountain View, CA, United States of America

Rami Botros