Mathematica 11.2 is now available! This latest version expands Mathematica's cutting-edge functionality in audio and image processing, mathematical computation and machine learning—plus much more. Here are some of the highlights:

Audio and Image Processing

• Generate spoken audio in different voices with SpeechSynthesize
• The Audio framework now has stereo support, programmatic playback and new AudioStream objects
• Alter one image to look more like another with ImageRestyle
• Use CurrentScreenImage or CurrentNotebookImage to capture the current screen or notebook
• Generate 2D and 3D raster images from geometric regions with RegionImage
• Retrieve satellite images of a geographic area with GeoImage

Mathematical Computation

• Limit can now calculate undirected, nested and multivariate limits
• RSolveValue and DSolveValue can now compute limiting values directly
• Additional functions and operators for calculating limits: MinLimit, MaxLimit, DiscreteLimit, etc.
• Support for 3D Boolean operations through RegionUnion, RegionIntersection, etc.

Machine Learning

• Classify and Predict have new training methods, built-in models, progress reporting and TimeGoal for limiting training times
• New feature types for extracting audio, sequences and word vectors from data
• WMLF and MXNet formats for import/export of neural networks and other machine learning models
• Expanded options and methods for various neural network layer types

Version 11.2 also includes a number of new import/export formats, interpreter types and low-level performance optimizations. Programming is more flexible than ever with new functionality for initialization, background tasks and network programming, as well as new interfaces for external interpreted languages.

For more details about these and other new additions, read through Stephen Wolfram's launch-day blog post and take a look at the Summary of New Features in 11.2.