Abstract: Techniques are provided for early processing of a part of a user input to produce a response to the entire or final user input. While the user input is being received, a partial user input, which is a part of the final user input, is processed to produce a response. The response is a candidate response for the final user input. After the final user input is received, and if the partial user input is determined to match or be equivalent to the final user input, the first response, which is already available, is provided to one or more output devices for presentation. If the final user input is determined to differ from the partial user input, the final user input is processed to produce a second response to the final user input, and the second response is provided for presentation. In some instances, multiple partial user inputs are received and processed.

Abstract: Systems and methods are provided for smart audio segmentation using look-ahead based acousto-linguistic features. For example, systems and methods are provided for obtaining audio, processing the audio, identifying a potential segmentation boundary within the audio, and determining whether to generate a segment break at the potential segmentation boundary. One or more look-ahead words occurring after the potential segmentation boundary are identified, wherein an acoustic segmentation score and a language segmentation score associated with the potential segmentation boundary and the one or more look-ahead words are generated. Systems then either refrain from generating a segment break at the potential segmentation boundary or generate the segment break at the potential segmentation boundary based on the acoustic and/or language segmentation score at least meeting or exceeding a segmentation score threshold.

Abstract: Techniques are provided for early processing of a part of a user input to produce a response to the entire or final user input. While the user input is being received, a partial user input, which is a part of the final user input, is processed to produce a response. The response is a candidate response for the final user input. After the final user input is received, and if the partial user input is determined to match or be equivalent to the final user input, the first response, which is already available, is provided to one or more output devices for presentation. If the final user input is determined to differ from the partial user input, the final user input is processed to produce a second response to the final user input, and the second response is provided for presentation. In some instances, multiple partial user inputs are received and processed.

Abstract: A method for seismic exploration using a full waveform inversion, FWI, the method including receiving an initial velocity model V of the subsurface, receiving recorded data d related to the subsurface, generating synthetic data u related to the subsurface, using the initial velocity model V and a source signature of a source S, transforming the recorded data d and the synthetic data u, with a complex wavelet transform, into complex wavelet transformed recorded data d? and complex wavelet transformed synthetic data u?, respectively, updating the initial velocity model V using the FWI to generate an updated velocity model, based on a cost function J which depends on the complex wavelet transformed recorded data d? and the complex wavelet transformed synthetic data u?, and generating an image of a surveyed subsurface formation.

Abstract: A joint timelapse full waveform inversion (FWI) method for estimating physical properties of a subsurface includes receiving seismic data related to the subsurface, wherein the seismic data includes a baseline dataset dB and a monitor dataset dM, defining an objective function of the FWI method, calculating a baseline gradient gB of the objective function for the baseline dataset dB, and a monitor gradient gM of the objective function for the monitor dataset dM, computing a baseline preconditioner P?B for the baseline dataset dB and a monitor preconditioner P?M for the monitor dataset dM so that each of the baseline preconditioner P?B and the monitor preconditioner P?M reflects similarities and/or differences of geometrical features of the baseline and monitor acquisition surveys, and determining physical properties of the subsurface based on a baseline physical properties update and a monitor physical properties update.

Abstract: A least-square migration, LSM, based method for generating a 4D image of a subsurface, the method including receiving seismic data d related to the subsurface, the seismic data d including a baseline dataset dB and a monitor dataset dM, calculating a baseline filter B and a monitor filter M based on a same common reflectivity r of the subsurface and corresponding remigrated baseline data mB1 and remigrated monitor data mM1 so that the base filter B applied to the remigrated baseline data mB1 equals the monitor filter M applied to the remigrated monitor data mM1, applying the baseline filter B to raw migrated baseline data mB0 and applying the monitor filter M to raw migrated monitor data mM0 to generate LSM baseline data mB and LSM monitor data mM, and generating the 4D image of the subsurface based on the LSM baseline data mB and the LSM monitor data mM.

Abstract: Systems and methods for obtaining a shape of a component within a subterranean well with an impression tool includes the impression tool having a top sub. The top sub is an elongated member having a central axis and a downhole facing surface. A threaded recess of the top sub extends axially from the downhole facing surface. A bottom body section has a block body, a lead impression block positioned at a downhole side of the block body, and a threaded member located at an uphole side of the block body. The threaded member of the bottom body section threadingly engages the threaded recess of the top sub. The bottom body section is operable to be selectively threaded to, and unthreaded from, the top sub.

Abstract: Disclosed speech recognition techniques improve user-perceived latency while maintaining accuracy by: receiving an audio stream, in parallel, by a primary (e.g., accurate) speech recognition engine (SRE) and a secondary (e.g., fast) SRE; generating, with the primary SRE, a primary result; generating, with the secondary SRE, a secondary result; appending the secondary result to a word list; and merging the primary result into the secondary result in the word list. Combining output from the primary and secondary SREs into a single decoder as described herein improves user-perceived latency while maintaining or improving accuracy, among other advantages.

Abstract: Generally discussed herein are devices, systems, and methods for on-device detection of a wake word. A device can include a memory including model parameters that define a custom wake word detection model, the wake word detection model including a recurrent neural network transducer (RNNT) and a lookup table (LUT), the LUT indicating a hidden vector to be provided in response to a phoneme of a user-specified wake word, a microphone to capture audio, and processing circuitry to receive the audio from the microphone, determine, using the wake word detection model, whether the audio includes an utterance of the user-specified wake word, and wake up a personal assistant after determining the audio includes the utterance of the user-specified wake word.

Abstract: Generally discussed herein are devices, systems, and methods for custom wake word selection assistance. A method can include receiving, at a device, data indicating a custom wake word provided by a user, determining one or more characteristics of the custom wake word, determining that use of the custom wake word will cause more than a threshold rate of false detections based on the characteristics, rejecting the custom wake word as the wake word for accessing a personal assistant in response to determining that use of the custom wake word will cause more than a threshold rate of false detections, and setting the custom wake word as the wake word in response to determining that use of the custom wake word will not cause more than the threshold rate of false detections.

Abstract: Disclosed speech recognition techniques improve user-perceived latency while maintaining accuracy by: receiving an audio stream, in parallel, by a primary (e.g., accurate) speech recognition engine (SRE) and a secondary (e.g., fast) SRE; generating, with the primary SRE, a primary result; generating, with the secondary SRE, a secondary result; appending the secondary result to a word list; and merging the primary result into the secondary result in the word list. Combining output from the primary and secondary SREs into a single decoder as described herein improves user-perceived latency while maintaining or improving accuracy, among other advantages.

Abstract: Systems and methods for obtaining a shape of a component within a subterranean well with an impression tool includes the impression tool having a top sub. The top sub is an elongated member having a central axis and a downhole facing surface. A threaded recess of the top sub extends axially from the downhole facing surface. A bottom body section has a block body, a lead impression block positioned at a downhole side of the block body, and a threaded member located at an uphole side of the block body. The threaded member of the bottom body section threadingly engages the threaded recess of the top sub. The bottom body section is operable to be selectively threaded to, and unthreaded from, the top sub.

Abstract: A non-blended dataset related to a same surveyed area as a blended dataset is used to deblend the blended dataset. The non-blended dataset may be used to calculate a model dataset emulating the blended dataset, or may be transformed in a model domain and used to derive sparseness weights, model domain masking, scaling or shaping functions used to deblend the blended dataset.

Abstract: Techniques are provided for early processing of a part of a user input to produce a response to the entire or final user input. While the user input is being received, a partial user input, which is a part of the final user input, is processed to produce a response. The response is a candidate response for the final user input. After the final user input is received, and if the partial user input is determined to match or be equivalent to the final user input, the first response, which is already available, is provided to one or more output devices for presentation. If the final user input is determined to differ from the partial user input, the final user input is processed to produce a second response to the final user input, and the second response is provided for presentation. In some instances, multiple partial user inputs are received and processed.

Abstract: Disclosed speech recognition techniques improve user-perceived latency while maintaining accuracy by: receiving an audio stream, in parallel, by a primary (e.g., accurate) speech recognition engine (SRE) and a secondary (e.g., fast) SRE; generating, with the primary SRE, a primary result; generating, with the secondary SRE, a secondary result; appending the secondary result to a word list; and merging the primary result into the secondary result in the word list. Combining output from the primary and secondary SREs into a single decoder as described herein improves user-perceived latency while maintaining or improving accuracy, among other advantages.

Abstract: Methods and apparatuses for seismic data processing perform a least-squares reverse time migration method in which surface-attribute-independent coefficients for the surface attribute gathers are demigrated to reduce the computational cost.

Abstract: Seismic exploration of an underground formation uses seismic excitations to probe the formation's properties such as reflectivity that can be imaged using reverse time migration. Using an equal area spherical binning at reflection points improves and simplifies RTM imaging together with adaptability to the data acquisition geometry, while overcoming drawbacks of conventional cylindrical binning.

Abstract: Generally discussed herein are devices, systems, and methods for custom wake word selection assistance. A method can include receiving, at a device, data indicating a custom wake word provided by a user, determining one or more characteristics of the custom wake word, determining that use of the custom wake word will cause more than a threshold rate of false detections based on the characteristics, rejecting the custom wake word as the wake word for accessing a personal assistant in response to determining that use of the custom wake word will cause more than a threshold rate of false detections, and setting the custom wake word as the wake word in response to determining that use of the custom wake word will not cause more than the threshold rate of false detections.

Abstract: Disclosed speech recognition techniques improve user-perceived latency while maintaining accuracy by: receiving an audio stream, in parallel, by a primary (e.g., accurate) speech recognition engine (SRE) and a secondary (e.g., fast) SRE; generating, with the primary SRE, a primary result; generating, with the secondary SRE, a secondary result; appending the secondary result to a word list; and merging the primary result into the secondary result in the word list. Combining output from the primary and secondary SREs into a single decoder as described herein improves user-perceived latency while maintaining or improving accuracy, among other advantages.

Abstract: Generally discussed herein are devices, systems, and methods for custom wake word selection assistance. A method can include receiving, at a device, data indicating a custom wake word provided by a user, determining one or more characteristics of the custom wake word, determining that use of the custom wake word will cause more than a threshold rate of false detections based on the characteristics, rejecting the custom wake word as the wake word for accessing a personal assistant in response to determining that use of the custom wake word will cause more than a threshold rate of false detections, and setting the custom wake word as the wake word in response to determining that use of the custom wake word will not cause more than the threshold rate of false detections.