Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for speech synthesis. The methods, systems, and apparatus include actions of obtaining an audio representation of speech of a target speaker, obtaining input text for which speech is to be synthesized in a voice of the target speaker, generating a speaker vector by providing the audio representation to a speaker encoder engine that is trained to distinguish speakers from one another, generating an audio representation of the input text spoken in the voice of the target speaker by providing the input text and the speaker vector to a spectrogram generation engine that is trained using voices of reference speakers to generate audio representations, and providing the audio representation of the input text spoken in the voice of the target speaker for output.

Abstract: Implementations are directed to updating a trained voice bot that is deployed for conducting conversations on behalf of a third-party. A third-party developer can interact with a voice bot development system that enables the third-party developer to train, update, validate, and monitor performance of the trained voice bot. In various implementations, the trained voice bot can be updated by updating a corpus of training instances that was initially utilized to train the voice bot, and updating the trained voice bot based on the updated corpus. In some implementations, the corpus of training instances may be updated in response to identifying occurrence(s) of behavioral error(s) of the trained voice bot while the conversations are being conducted on behalf of the third-party. In additional or alternative implementations, the corpus of training instances may be updated in response to determining the trained voice bot does not include a desired behavior.

Abstract: To reduce the storage size of weight tensors and speed up loading of weight tensors from system memory, a compression technique can be employed to remove zero values from a weight tensor before storing the weight tensor in system memory. A sparsity threshold can be enforced to achieve a compression ratio target by forcing small weight values to zero during training. When the weight tensor is loaded from system memory, a direct memory access (DMA) engine with an in-line decompression unit can decompress the weight tensor on-the-fly. By performing the decompression in the DMA engine, expansion of the weight values back to the original weight tensor size can be carried out in parallel while other neural network computations are being performed by the processing unit.

Abstract: A system may include a turbine and a recuperative heat exchanger system. The recuperative heat exchanger system is configured to receive exhaust gases from the turbine. The recuperative heat exchanger system may include a precool section to cool the exhaust gases, a major heating section to receive the cooled the exhaust gases, and a minor heating section to receive the cooled the exhaust gases.

Abstract: The present disclosure discloses a garment for being used as a nursing bra and a suitable breast pad that is adapted to be fitted into pad pockets of the nursing bra. The nursing bra is basically made of a seamless knitted fabric and with an attachment of additional second fabric layer to the seamless knitted fabric pad pockets are being formed between the seamless knitted fabric and the second fabric. These pad pockets are accessible from respective openings that are formed in the second fabric at a position suitable for receiving a nipple of a wearer such. Furthermore, the openings are designed so as to allow relatively easy insertion and removal of the breast pad into and from the pad pockets. Thus, when the breast pads are accommodated within the pad pockets, a nipple wearer engages the breast pads such that any leakage of breastmilk is absorbed by the breast pad and the rest of the bra maintains dry.

Abstract: An offshore drilling vessel includes a hull and at least one opening in the hull arranged to receive an end of at least one submersible cable. The offshore drilling vessel also comprises a cable capture mechanism configured to lift the end of the at least one submersible cable through the at least one opening from a submersed position to a raised position. A cable connector is configured to couple to the end of the at least one submersible cable when the end of the at least one submersible cable is in the raised position such that the offshore drilling vessel comprises an external data and/or power connection via the submersible cable.

Abstract: Navigation systems and methods for magnetic interference correction involve antennae that generate magnetic fields at different frequencies, sensors that measure the magnetic fields, and a computing device that uses sensor measurements to determine the magnetic interference and produce accurate sensor position and orientation information.

Abstract: The volume, content, and void volume of a box are determined. Using those determinations, a fill ratio calculation is used to establish the packing element requirement of the box.

Abstract: The present invention provides a crystalline form and stable salts of 6-fluoro-7-(2-fluoro-6-hydroxyphenyl)-1-(4-methyl-2-(2-propanyl)-3-pyridinyl)-4-((2S)-2-methyl-4-(2-propenoyl)-1-piperazinyl)pyrido[2,3-d]pyrimidin-2(1H)-one, including several hydrochloride salt forms, phosphate salt form, mesylate salt form, and solid state forms thereof, pharmaceutical compositions, and methods of treating a disease mediated by KRAS G12C inhibition.

Abstract: A method comprising: receiving a first workload data set, the first workload data set specifying a cache hit outcome distribution that is associated with a plurality of input-output (I/O) operations; identifying a plurality of workload portions of the first workload data set, each of the workload portions identifying: (i) a rate of a cache hit outcome that is associated with a respective I/O operation, and (ii) a data size that is associated with the respective I/O operations; generating a plurality of initial vectors, each of the initial vectors being generated based on a different one of the plurality of workload portions, each of the initial vectors being generated by a different sub-network of a correlation neural network; generating a context vector based on the plurality of initial vectors; processing the context vector with a decoder to generate a plurality of data points in a response curve of a storage system.

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.

Abstract: Apparatus, consisting of a probe with a probe distal end having an external surface that contacts tissue when the distal end is inserted into a patient. A heater and a temperature sensor are in the distal end, and a gas inlet is coupled to the distal end to provide gas. While the heater is activated, a controller registers a first time when a temperature measured by the sensor equals or exceeds a preset temperature, and, in response to the temperature equaling or exceeding the preset temperature, delivers a gas pulse through the gas inlet so as to cool the probe distal end. The controller then registers a second time when the temperature measured by the sensor equals or exceeds the preset temperature, and, in response to a time difference between the second and first times, estimates a temperature of the external surface of the probe distal end.

Abstract: A method is disclosed. A data set including: (a) identifiers of a set of incidents occurring within a defined geographic region to which at least one service provider responded during a first time period and (b) address data identifying a location within the geographic region of each said incident of the set is retrieved over a network. An instruction to generate a heat map of the incidents occurring within the geographic region during the first time period is received from a user via a user interface generated to a display device. In response to the instruction to generate the heat map, the address data is converted to GPS data. A heat map of an aerial view of the geographic region based on the GPS data is generated. The heat map is displayed to the display device in a user interface.

Abstract: Methods of sequencing and resequencing nucleic acids using incorporation and binding of nucleotides, including terminated nucleotides, are provided. Also provided herein are methods using combinations of terminated nucleotides and non-terminated nucleotides or combinations of terminated nucleotides and non-incorporable nucleotides. These methods allow for determination of the length of homopolymer sequences and increased accuracy of the sequencing reads. The compositions, reagents, and kits for practicing the methods are also provided.

Abstract: A plasma source is configured to be in fluid communication with the interior of a vacuum vessel, a vacuum chamber, such as by being installed in a wall of a vacuum vessel that encloses the chamber, or in a pipe that is connected to the chamber. The plasma source may produce plasma in a line of sight of a surface to be cleaned, such as an internal surface of the vessel, which may be used for processes such as deposition or etching.

Abstract: An apparatus may include an interface system and a first local control system. The first local control system may be configured to: receive first sensor data from a first preview environment while a content stream is being presented in the first preview environment; generate, based at least in part on the first sensor data, first user engagement data corresponding to one or more people in the first preview environment, the first user engagement data indicating estimated engagement with presented content of the content stream; output, via the interface system, either the first user engagement data, the first sensor data, or both, to a data aggregation device; and determine, based at least in part on user preference data, whether to provide at least some of the first user engagement data, at least some of the first sensor data, or both, to one or more machine learning (ML) models.

Abstract: Examples described herein involve switching between two user accounts of a streaming media service. In an example implementation, a playback device stores data representing credentials for multiple user accounts including a first user account of a first streaming audio service and a second user account of the first streaming audio service. The playback device receives instructions to play first audio content from the first streaming audio service and plays back the first audio content by streaming data representing the first audio content from one or more servers of the first streaming audio service using credentials of the first user account. The playback device detects a trigger associated with the first user account of the first streaming audio service and switches from streaming data representing the first audio content using credentials of the first user account to streaming data representing the first audio content using credentials of the second user account.

Abstract: A method of estimating dimensions of vehicle's exterior defects based on reference dimensions comprising analyzing one or more images captured by one or more image sensors deployed to depict a vehicle to identify one or more reference feature relating to the vehicle, obtaining real-world size of the reference feature(s), computing a pixel to real-world size ratio for the image sensor(s) based on the real-world size of the reference feature(s) and a size in pixels of the reference feature(s) in the image(s), analyzing one or more images captured by the image sensor(s) to identify one or more defects in an exterior of the vehicle, computing a size in pixels of one or more dimensions of each defect; and computing a real-world size of the dimension(s) based on the size in pixels of the respective dimension and the pixel to real-world size ratio.

Abstract: Systems and methods for determining predictive model types are provided. A method may include generating a predictive model for a web page of a web site, wherein the web page includes a configuration defining one or more objects presented with the web page, and wherein each object is associated with a predictive model. The method may include determining one or more predictive model types that are associated with the predictive model, determining one or more performance indicators that correspond to each determined predictive model type, wherein performance indicators represent one or more benefits to a website, selecting a predictive model type of the predictive model out of the one or more predictive model types, wherein the predictive model type is selected based on a performance indicator corresponding to the selected predictive model type, and determining a configuration of the web page using the selected predictive model type of the predictive model.