RELEASE_NOTES

openSMILE 2.3 (beta)

Major new features:

* Android JNI integration.  
While version 2.2. did provide first support for compilation as static binary for Android,
we now offer integration into Apps via the JNI. An example project and a tutorial for 
Android Studio are provided which shows how to build a live audio analysis app that
shows audio parameters in real-time in the App UI.

* Configuration file interface updated
Commandline options for audio input and data output formats have been standardised
through modular config files. This allows all major standard

* Batch feature extraction GUI for Windows and batch extraction scripts for Linux *
These make it easier for beginners to extract audio features from a collection of audio files.

* Improved backwards compatibility *
All standard feature sets are verified to be backwards compatible to Version 2.2 and 2.1.
The original feature sets of openEAR are also backwards compatible with 
the original openEAR models.

* Updated Version of ComParE 2013-2015 baseline acoustic parameter set *
Several optimisations to the feature extraction code were applied and
an updated version of ComParE was released. It has the same features as 
the previous version, but is now numerically optimised. It is referred to
as ComParE_2016, as it is released in 2016. 
It is not the Interspeech ComParE 2016 baseline feature set (this is still the 2013 one)!!

* Lots of various fixes and improvements *
We've added many minor points, fixed bugs, and improved performance.
E.g. we have improved the support for modulation spectra as functionals and optimized 
Jitter computation through a better pitch period detection algorithm.