0.1.6

• Added some optimization flags in meson script. Needed for noise-repellent 0.2.3 release.

0.1.5

• Fixed memory corruption due to lower than 44100 samplerates. Special thanks to @SimonG4ELI that helped quite a bit with it!

0.1.4

• Fixes for many invalid read and write issues reported using valgrind.
• Fixes for some memory leaks
• Better documentation in header files.
• Correction of some typos in code comments. Thanks @luzpaz

0.1.3

• More memory leaks fixes

v0.1.2

• Fixed not used dependency check in meson build file
• Make adenoiser_demo example installable
• Fix wrong parameter in adenoiser_demo

v0.1.1

• Fixed a bunch of memory leaks
• Correctly configure library to be installed

First Release

This is the first release of the library. This library condensates all the lessons learned through noise repellent development plus a good amount of refactoring and better code practices. It implements a modular way of developing STFT based processing algorithms using what's called internally as spectral processors. This allows to extend the library in the future with more algorithms. The library its generic enough that can be used in any type of applications, including RT safe processing. The algorithms were refined and some of them even improved in terms of quality of reduction.
After noise-repellent was release many developers had forked the plugin and adapt it to use the algorithms in other contexts. This library effectively removes the lv2 dependency that noise repellent has so it allows the algorithms to be integrated in other type of apps.

Provided in this release there are two spectral processors

• An adaptive denoiser that learns the noise profile automatically in each processing frame. It uses a 20 ms frame so latency is appropriate to be used in a live situation.
• A more intricate denoiser that allow to learn a noise profile and set it externally. This manual denoiser offers much more precise and higher resolution reduction. It's intended to be used as a post processing effect.