matchering

Matchering 2.0 is an open-source audio matching and mastering solution that takes two audio files — a TARGET track you want to master and a REFERENCE track you want it to sound like — and processes the target to match the reference's RMS level, frequency response, peak amplitude, and stereo width. The project is a complete rewrite in Python 3 with an open-source tech stack, replacing the previous MATLAB dependency. It includes the Hyrax brickwall limiter, developed as part of the project. Matchering can be used in several ways: as a containerized web application via Docker (with platform-specific guides for Windows, macOS, and Linux), as a Python library installable via pip (requiring Python 3.8+ and libsndfile), as a ComfyUI node, and integrated into the UVR5 Desktop App. Online demos are hosted by Songmastr, MVSEP, and Moises. A command-line interface (matchering-cli) and an enhanced fork are also available. The algorithm has been benchmarked in a study by Benn Jordan, ranking third out of twelve mastering solutions, behind two professional mastering engineers. The project emphasizes "Your References, Your Rules" — giving users full control over the reference material. Documentation includes a Habr article (in Russian), press coverage, supporter list, limiter quality tests, and a JSFX clone of the Hyrax limiter by Tokyo Dawn Labs.