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 includes obtaining a graph that includes a plurality of nodes each representing a computational job and connected to one or more other nodes via edges. Each edge represents a dependency between two connected nodes. For each computing device of a distributed computing system, the method includes obtaining resource characteristics. The method includes slicing, based on the dependencies of connected nodes, the graph into a plurality of sub-graphs each including one or more nodes. For each respective sub-graph, the method includes determining a computational cost of the respective sub-graph and distributing, based on the computational cost and the resource characteristics, the respective sub-graph to the respective one of the computing devices. The method includes receiving a respective result from the respective one of the computing devices based on execution of the computational jobs. The method includes determining a total result based on each respective result.

Abstract: Audio rendering devices comprising at least one audio rendering unit for playing audio streams, processing circuit and a wireless communication unit for establishing wireless links and related method are disclosed. One audio rendering device is configured to receive a remote audio stream prioritization policy comprising one or more lists of associations between audio stream attributes and priority values, update a local audio stream prioritization policy based on the remote audio stream prioritization policy, identify audio streams available via wireless links, retrieve audio stream attributes of the available audio streams, select an available audio stream to be played among the available audio streams, based on the updated local audio stream prioritization policy and based on the audio stream attributes of the available audio streams, and play the selected audio stream, by the audio rendering unit of the audio rendering device.

Abstract: A method includes receiving audio data corresponding to a query spoken and processing the audio data to generate multiple candidate hypotheses each represented by a respective sequence of hypothesized terms. For each candidate hypothesis, the method includes determining whether the sequence of hypothesized terms includes a source phrase from a list of phrase correction pairs. Each phrase correction pair includes a corresponding source phrase that was misrecognized and a corresponding target phrase replacing the source phrase. When the respective sequence of hypothesized terms includes the source phrase, the method includes generating a corresponding additional candidate hypothesis that replaces the source phrase.

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: Systems and methods that enable hardware memory error tolerant software systems. For instance, the system may comprise a host device that instantiates a kernel agent in response to one or more requests to access hardware memory, determines, by the kernel agent based on the received information, whether the request to access memory will cause access to a corrupt memory location, and skip an operation associated with the corrupt memory location in response to determining that the request will access a corrupt memory location. The systems may also include a system that detects software vulnerabilities to hardware memory errors.

Abstract: A method includes obtaining a transcript of multiple dialog acts during a multi-party communication session and extracting, from the transcript, a plurality of extractive action items each associated with a task expected to be completed within a timeframe to begin after the multi-party communication session ends. The method also includes generating, using an abstractive action item identification model configured to receive the extractive action items extracted from the transcript, one or more abstractive action items. Each abstractive action item is associated with a respective group of one or more of the extractive action items that are associated with a same respective task. For each of the one or more abstractive action items, the method also includes presenting, in one or more graphical user interfaces, information related to the corresponding abstractive action item.

Abstract: A method for processing data exactly once using transactional stream writes includes receiving, from a client, a batch of data blocks for storage on memory hardware in communication with the data processing hardware. The batch of data blocks is associated with a corresponding sequence number and represents a number of rows of a table stored on the memory hardware. The method also includes partitioning the batch of data blocks into a plurality of sub-batches of data blocks. For each sub-batch of data blocks, the method further includes assigning the sub-batch of data blocks to a buffered stream; writing, using the assigned buffered stream, the sub-batch of data blocks to the memory hardware; updating a storage log with an intent to commit the sub-batch of data blocks using the assigned buffered stream; and committing the sub-batch of data blocks to the memory hardware.

Abstract: A method includes initializing a client state on a client device be executing a private batched sum retrieval instruction to compute c sums O of data blocks from an untrusted storage device. Each computed sum O stored on memory hardware of the client device and including a sum of a corresponding subset S of exactly k data blocks. The method also includes a query instruction to retrieve a query block Bq stored on the untrusted storage device by iterating through each of the c sums O of data blocks to identify one of the c sums O that does not include the query block Bq, instructing a service to pseudorandomly partition the untrusted storage device into partitions and sum the data blocks in each partition to determine a corresponding encrypted data block sum.

Abstract: A system for creating a sync group in a distributed system may include an offline electronic device having a short range communication transmitter and a storage service, and a computer-readable storage medium comprising one or more programming instructions. The system may receive a request from an application to create a sync group, and verify that the application has read access to the storage service. In response to verifying that the application has read access to the storage service, the system may create the sync group by receiving an indication of one or more objects stored in the storage service that are to be shared, identifying one or more members of the sync group, and assigning one or more permissions to the one or more members that are identified. The system may synchronize access to the objects among the electronic device and one or more member electronic devices.

Abstract: A system for generating an output audio signal includes a context encoder, a text-prediction network, and a text-to-speech (TTS) model. The context encoder is configured to receive one or more context features associated with current input text and process the one or more context features to generate a context embedding associated with the current input text. The text-prediction network is configured to process the current input text and the context embedding to predict, as output, a style embedding for the current input text. The style embedding specifies a specific prosody and/or style for synthesizing the current input text into expressive speech. The TTS model is configured to process the current input text and the style embedding to generate an output audio signal of expressive speech of the current input text. The output audio signal has the specific prosody and/or style specified by the style embedding.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for adapting a language model are disclosed. In one aspect, a method includes the actions of receiving transcriptions of utterances that were received by computing devices operating in a domain and that are in a source language. The actions further include generating translated transcriptions of the transcriptions of the utterances in a target language. The actions further include receiving a language model for the target language. The actions further include biasing the language model for the target language by increasing the likelihood of the language model selecting terms included in the translated transcriptions. The actions further include generating a transcription of an utterance in the target language using the biased language model and while operating in the domain.

Abstract: Methods, systems, and apparatus, including an apparatus for analyzing content that is displayed within an application (e.g., a native application) on a client device and making decisions regarding the presentation of content within the application. In one aspect, a process includes identifying an event occurring within the application. The event can be indicative of a display of content by the application. The content to be displayed is received from the application. Information representative of the content is sent to a server. A response is received from the server. The response is indicative of the suitability of the content for display within the application.

Abstract: A method for training a machine learning model, e.g., a neural network, using a regularization scheme is disclosed. The method includes generating regularized partial gradients of losses computed using an objective function for training the machine learning model.

Abstract: A method includes generating an audio signal encoding an utterance captured by a microphone of a user device and transmitting the audio signal encoding the utterance to a server. The server is configured to determine a speaker of the utterance from one of a plurality of different users of the user device based on a comparison between the audio signal encoding the utterance and corresponding speaker verification data, and process the audio signal encoding the utterance using a speech recognition module to identify a particular action. The method also includes executing the particular action identified by the server to cause a particular application to launch on the user device based on user permissions associated with the speaker determined by the server to access the particular data.

Abstract: A method for instantaneous learning in text-to-speech (TTS) during dialog includes receiving a user pronunciation of a particular word present in a query spoken by a user. The method also includes receiving a TTS pronunciation of the same particular word that is present in a TTS input where the TTS pronunciation of the particular word is different than the user pronunciation of the particular word. The method also includes obtaining user pronunciation-related features and TTS pronunciation related features associated with the particular word. The method also includes generating a pronunciation decision selecting one of the user pronunciation or the TTS pronunciation of the particular word that is associated with a highest confidence. The method also include providing the TTS audio that includes a synthesized speech representation of the response to the query using the user pronunciation or the TTS pronunciation for the particular word.

Abstract: This document describes systems and techniques directed at concurrent communication in multiple time-division duplex (TDD) bands. As new industry standards (e.g., the Fifth Generation New Radio (5G NR) standard) are being implemented, more TDD bands are becoming available for wireless communications. Generally, manufacturers will add additional antenna systems for each TDD band, but this method may become costly and need extra space, which is already limited, within the user device In various aspects, the concurrent communication system includes a radio frequency (RF) modem module configured to operate on multiple TDD bands, which may include bands that are located near each other on the RF spectrum. The concurrent communication system further includes transceiver circuitry with at least one transmission chain and at least four reception chains. The architecture of this system offers an efficient and inexpensive way to communicate on at least two TDD bands concurrently with reduced hardware cost.

Abstract: Technology for a user interface that supports an access control mechanism for network sharing. An example method involves providing a user interface comprising media items and media availability indicators, wherein a media availability indicator indicates one or more states of a media item; updating the media availability indicator to represent that an encrypted version of the media item and a wrapped key have been received over a connection, wherein the wrapped key was encrypted using a public key of a server; enabling the user interface to initiate the transmission of the wrapped key to the server, wherein the wrapped key comprises a cryptographic key for the encrypted version of the media item; receiving the cryptographic key that is unwrapped by the server using a private key of the server; and updating the media availability indicator of the user interface to represent the media item is available to be experienced.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating a computer chip floorplan. One of the methods includes obtaining netlist data for a computer chip; and generating a computer chip floorplan, comprising placing a respective node at each time step in a sequence comprising a plurality of time steps, the placing comprising, for each time step: generating an input representation for the time step; processing the input representation using a node placement neural network having a plurality of network parameters, wherein the node placement neural network is configured to process the input representation in accordance with current values of the network parameters to generate a score distribution over a plurality of positions on the surface of the computer chip; and assigning the node to be placed at the time step to a position from the plurality of positions using the score distribution.

Abstract: Systems and methods for matching live media content are disclosed. At a server, obtaining first media content from a client device, herein the first media content corresponds to a portion of media content being played on the client device, and the first media content is associated with a predefined expiration time; obtaining second media content from one or more content feeds, wherein the second media content also corresponds to a portion of the media content being played on the client device; in accordance with a determination that the second media content corresponds to a portion of the media content that has been played on the client device: before the predefined expiration time, obtaining third media content corresponding to the media content being played on the client device, from the one or more content feeds; and comparing the first media content with the third media content.