Abstract: Methods, systems, and apparatus, including computer-readable media, for performing speech recognition using sequence-to-sequence models. An automated speech recognition (ASR) system receives audio data for an utterance and provides features indicative of acoustic characteristics of the utterance as input to an encoder. The system processes an output of the encoder using an attender to generate a context vector and generates speech recognition scores using the context vector and a decoder trained using a training process that selects at least one input to the decoder with a predetermined probability. An input to the decoder during training is selected between input data based on a known value for an element in a training example, and input data based on an output of the decoder for the element in the training example. A transcription is generated for the utterance using word elements selected based on the speech recognition scores. The transcription is provided as an output of the ASR system.

Abstract: 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.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-readable storage media, for speech recognition using attention-based sequence-to-sequence models. In some implementations, audio data indicating acoustic characteristics of an utterance is received. A sequence of feature vectors indicative of the acoustic characteristics of the utterance is generated. The sequence of feature vectors is processed using a speech recognition model that has been trained using a loss function that uses N-best lists of decoded hypotheses, the speech recognition model including an encoder, an attention module, and a decoder. The encoder and decoder each include one or more recurrent neural network layers. A sequence of output vectors representing distributions over a predetermined set of linguistic units is obtained. A transcription for the utterance is obtained based on the sequence of output vectors. Data indicating the transcription of the utterance is provided.

Abstract: A method for training a speech recognition model with a minimum word error rate loss function includes receiving a training example comprising a proper noun and generating a plurality of hypotheses corresponding to the training example. Each hypothesis of the plurality of hypotheses represents the proper noun and includes a corresponding probability that indicates a likelihood that the hypothesis represents the proper noun. The method also includes determining that the corresponding probability associated with one of the plurality of hypotheses satisfies a penalty criteria. The penalty criteria indicating that the corresponding probability satisfies a probability threshold, and the associated hypothesis incorrectly represents the proper noun. The method also includes applying a penalty to the minimum word error rate loss function.

Abstract: A method includes receiving a training example for a listen-attend-spell (LAS) decoder of a two-pass streaming neural network model and determining whether the training example corresponds to a supervised audio-text pair or an unpaired text sequence. When the training example corresponds to an unpaired text sequence, the method also includes determining a cross entropy loss based on a log probability associated with a context vector of the training example. The method also includes updating the LAS decoder and the context vector based on the determined cross entropy loss.

Abstract: A method of performing speech recognition using a two-pass deliberation architecture includes receiving a first-pass hypothesis and an encoded acoustic frame and encoding the first-pass hypothesis at a hypothesis encoder. The first-pass hypothesis is generated by a recurrent neural network (RNN) decoder model for the encoded acoustic frame. The method also includes generating, using a first attention mechanism attending to the encoded acoustic frame, a first context vector, and generating, using a second attention mechanism attending to the encoded first-pass hypothesis, a second context vector.

Abstract: 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.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing audio waveforms. In some implementations, a time-frequency feature representation is generated based on audio data. The time-frequency feature representation is input to an acoustic model comprising a trained artificial neural network. The trained artificial neural network comprising a frequency convolution layer, a memory layer, and one or more hidden layers. An output that is based on output of the trained artificial neural network is received. A transcription is provided, where the transcription is determined based on the output of the acoustic model.

Abstract: A method of biasing speech recognition includes receiving audio data encoding an utterance and obtaining a set of one or more biasing phrases corresponding to a context of the utterance. Each biasing phrase in the set of one or more biasing phrases includes one or more words. The method also includes processing, using a speech recognition model, acoustic features derived from the audio data and grapheme and phoneme data derived from the set of one or more biasing phrases to generate an output of the speech recognition model. The method also includes determining a transcription for the utterance based on the output of the speech recognition model.

Abstract: A method of transcribing speech using a multilingual end-to-end (E2E) speech recognition model includes receiving audio data for an utterance spoken in a particular native language, obtaining a language vector identifying the particular language, and processing, using the multilingual E2E speech recognition model, the language vector and acoustic features derived from the audio data to generate a transcription for the utterance. The multilingual E2E speech recognition model includes a plurality of language-specific adaptor modules that include one or more adaptor modules specific to the particular native language and one or more other adaptor modules specific to at least one other native language different than the particular native language. The method also includes providing the transcription for output.

Abstract: 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.

Abstract: A method includes receiving audio data encoding an utterance and obtaining a set of bias phrases corresponding to a context of the utterance. Each bias phrase includes one or more words. The method also includes processing, using a speech recognition model, acoustic features derived from the audio to generate an output from the speech recognition model. The speech recognition model includes a first encoder configured to receive the acoustic features, a first attention module, a bias encoder configured to receive data indicating the obtained set of bias phrases, a bias encoder, and a decoder configured to determine likelihoods of sequences of speech elements based on output of the first attention module and output of the bias attention module. The method also includes determining a transcript for the utterance based on the likelihoods of sequences of speech elements.

Abstract: A method includes receiving audio data encoding an utterance spoken by a native speaker of a first language, and receiving a biasing term list including one or more terms in a second language different than the first language. The method also includes processing, using a speech recognition model, acoustic features derived from the audio data to generate speech recognition scores for both wordpieces and corresponding phoneme sequences in the first language. The method also includes rescoring the speech recognition scores for the phoneme sequences based on the one or more terms in the biasing term list, and executing, using the speech recognition scores for the wordpieces and the rescored speech recognition scores for the phoneme sequences, a decoding graph to generate a transcription for the utterance.

Abstract: A method for generating final transcriptions representing numerical sequences of utterances in a written domain includes receiving audio data for an utterance containing a numeric sequence, and decoding, using a sequence-to-sequence speech recognition model, the audio data for the utterance to generate, as output from the sequence-to-sequence speech recognition model, an intermediate transcription of the utterance. The method also includes processing, using a neural corrector/denormer, the intermediate transcription to generate a final transcription that represents the numeric sequence of the utterance in a written domain. The neural corrector/denormer is trained on a set of training samples, where each training sample includes a speech recognition hypothesis for a training utterance and a ground-truth transcription of the training utterance. The ground-truth transcription of the training utterance is in the written domain.

Abstract: A method includes receiving audio data of an utterance and processing the audio data to obtain, as output from a speech recognition model configured to jointly perform speech decoding and endpointing of utterances: partial speech recognition results for the utterance; and an endpoint indication indicating when the utterance has ended. While processing the audio data, the method also includes detecting, based on the endpoint indication, the end of the utterance. In response to detecting the end of the utterance, the method also includes terminating the processing of any subsequent audio data received after the end of the utterance was detected.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for identifying the language of a spoken utterance. One of the methods includes receiving input features of an utterance; and processing the input features using an acoustic model that comprises one or more convolutional neural network (CNN) layers, one or more long short-term memory network (LSTM) layers, and one or more fully connected neural network layers to generate a transcription for the utterance.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech recognition using complex linear projection are disclosed. In one aspect, a method includes the actions of receiving audio data corresponding to an utterance. The method further includes generating frequency domain data using the audio data. The method further includes processing the frequency domain data using complex linear projection. The method further includes providing the processed frequency domain data to a neural network trained as an acoustic model. The method further includes generating a transcription for the utterance that is determined based at least on output that the neural network provides in response to receiving the processed frequency domain data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for keyword spotting. One of the methods includes training, by a keyword detection system, a convolutional neural network for keyword detection by providing a two-dimensional set of input values to the convolutional neural network, the input values including a first dimension in time and a second dimension in frequency, and performing convolutional multiplication on the two-dimensional set of input values for a filter using a frequency stride greater than one to generate a feature map.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech recognition using complex linear projection are disclosed. In one aspect, a method includes the actions of receiving audio data corresponding to an utterance. The method further includes generating frequency domain data using the audio data. The method further includes processing the frequency domain data using complex linear projection. The method further includes providing the processed frequency domain data to a neural network trained as an acoustic model. The method further includes generating a transcription for the utterance that is determined based at least on output that the neural network provides in response to receiving the processed frequency domain data.

Abstract: 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.