Abstract: Disclosed are apparatuses, systems, and techniques that may use machine learning for generating artificial speech. The techniques include obtaining a synthetic embedding using learned embeddings associated with different speakers. At least one learned embedding may be generated using a multi-stage training of a machine learning model (MLM) with progressively increasing quality of training speech utterances. The techniques may further include using the MLM and the synthetic embedding to generate synthetic audio data.

Abstract: In various examples, one or more text-to-speech machine learning models may be customized or adapted to accommodate new or additional speakers or speaker voices without requiring a full re-training of the models. For example, a base model may be trained on a set of one or more speakers and, after training or deployment, the model may be adapted to support one or more other speakers. To do this, one or more additional layers (e.g., adapter layers) may be added to the model, and the model may be re-trained or updated—e.g., by freezing parameters of the base model while updating parameters of the adapter layers—to generate an adapted model that can support the one or more original speakers of the base model in addition to the one or more additional speakers corresponding to the adapter layers.

Abstract: The system and method described herein relate to production of power from the wind farm that incorporate tunable power production forecasts for optimal wind farm performance, where the wind farm power production is controlled at least in part by the power production forecasts. The system and method use a tunable power forecasting model to generate tunable coefficients based on asymmetric loss function applied on actual power production data, along with tuning factor(s) that tune forecast towards under forecasting or over forecasting. The power production forecasts are generated using the tunable coefficients 34 and power characteristic features that are derived from actual power production data. The power production forecasts are monitored for any degradation, and a control action to regenerate the coefficients or retune the model is undertaken if degradation is observed.

Abstract: The present disclosure is directed to a system and method for forecasting a farm-level power output of a wind farm having a plurality of wind turbines. The method includes collecting actual operational data and/or site information for the wind farm. The method also includes predicting operational data for the wind farm for a future time period. Further, the method includes generating a model-based power output forecast based on the actual operational data, the predicted operational data, and/or the site information. In addition, the method includes measuring real-time operational data from the wind farm and adjusting the power output forecast based on the measured real-time operational data. Thus, the method also includes forecasting the farm-level power output of the wind farm based on the adjusted power output forecast.

Abstract: The present disclosure is directed to systems and methods for generating one or more farm-level power curves for a wind farm that can be used to validate an upgrade provided to the wind farm. The method includes operating the wind farm in a first operational mode. Another step includes collecting turbine-level operational data from one or more of the wind turbines in the wind farm during the first operational mode. The method also includes aggregating the turbine-level operational data into a representative farm-level time-series. Another step includes analyzing the operational data collected during the first second operational mode. Thus, the method also includes generating one or more farm-level power curves for the first operational mode based on the analyzed operational data.

Abstract: The present disclosure is directed to a system and method for forecasting a farm-level power output of a wind farm having a plurality of wind turbines. The method includes collecting actual operational data and/or site information for the wind farm. The method also includes predicting operational data for the wind farm for a future time period. Further, the method includes generating a model-based power output forecast based on the actual operational data, the predicted operational data, and/or the site information. In addition, the method includes measuring real-time operational data from the wind farm and adjusting the power output forecast based on the measured real-time operational data. Thus, the method also includes forecasting the farm-level power output of the wind farm based on the adjusted power output forecast.

Abstract: A method for creating predictive damage models includes receiving a first predictive damage model, identifying latent space between a first and a second domain asset, building a regression model from first domain asset projected source data, creating target dependent variables of a second model, applying classification or regression techniques to determine a function expressing the dependent variables, determining data points from the function to develop a second regression model, applying the second regression model to data points to predict target dependent variables, evaluating the second predictive damage model using the predicted target dependent variables, performing a sensitivity study to determine a directionality parameter of the second predictive damage model, and if the results are within an acceptable predetermined range, providing maintenance or servicing recommendations generated by the second predictive model to a user platform display, else repeating the process by rebuilding the regression mo

Abstract: A rail vehicle wheel flat warning system comprising a first sensor, a second sensor and a controller. The first sensor may be located adjacent to a first side of a rail to provide data associated with a rail vehicle wheel passing over the first side of the rail. The second sensor may be located adjacent to the first side of the rail to provide data associated with the rail vehicle wheel passing over the first side of the rail. Furthermore, the controller may be in communication with the first sensor and the second sensor to receive data from the first sensor and the second sensor. The controller may determine a potential wheel deformity based on the data received from the first sensor and the second sensor.

Abstract: In an embodiment task description information for tasks that are to be performed by developers is accessed. Performance information related to the tasks that are to be performed by the developers is accessed. The task description information and the performance information is analyzed to determine current performance estimation parameters that are indicative of how current instances of the tasks should be performed by the developers. The current performance estimation parameters are then output.

Abstract: A rail vehicle wheel flat warning system comprising a first sensor, a second sensor and a controller. The first sensor may be located adjacent to a first side of a rail to provide data associated with a rail vehicle wheel passing over the first side of the rail. The second sensor may be located adjacent to the first side of the rail to provide data associated with the rail vehicle wheel passing over the first side of the rail. Furthermore, the controller may be in communication with the first sensor and the second sensor to receive data from the first sensor and the second sensor. The controller may determine a potential wheel deformity based on the data received from the first sensor and the second sensor.

Abstract: The system and method described herein relate to production of power from the wind farm that incorporate tunable power production forecasts for optimal wind farm performance, where the wind farm power production is controlled at least in part by the power production forecasts. The system and method use a tunable power forecasting model to generate tunable coefficients based on asymmetric loss function applied on actual power production data, along with tuning factor(s) that tune forecast towards under forecasting or over forecasting. The power production forecasts are generated using the tunable coefficients 34 and power characteristic features that are derived from actual power production data. The power production forecasts are monitored for any degradation, and a control action to regenerate the coefficients or retune the model is undertaken if degradation is observed.

Abstract: The present disclosure is directed to systems and methods for generating one or more farm-level power curves for a wind farm that can be used to validate an upgrade provided to the wind farm. The method includes operating the wind farm in a first operational mode. Another step includes collecting turbine-level operational data from one or more of the wind turbines in the wind farm during the first operational mode. The method also includes aggregating the turbine-level operational data into a representative farm-level time-series. Another step includes analyzing the operational data collected during the first second operational mode. Thus, the method also includes generating one or more farm-level power curves for the first operational mode based on the analyzed operational data.

Abstract: A method for creating predictive damage models includes receiving a first predictive damage model, identifying latent space between a first and a second domain asset, building a regression model from first domain asset projected source data, create target dependent variables of a second model, applying classification or regression techniques to determine a function expressing the dependent variables, determining data points from the function to develop a second regression model, applying the second regression model to data points to predict target dependent variables, evaluating the second predictive damage model using the predicted target dependent variables, performing a sensitivity study to determine a directionality parameter of the second predictive damage model, and if the results are within an acceptable predetermined range, providing maintenance or servicing recommendations generated by the second predictive model to a user platform display, else repeating the process by rebuilding the regression mode