Matchering

Input—per 1M tokens

Output—per 1M tokens

Context—tokens

WeightsClosed

About

Matchering is ranked #10 of 50 in AI music mastering software on Inferse. It runs on Linux, macOS, Self-hosted, Web, Windows. There is a free plan.

Compared on AI music mastering software

Free plan: Yesgithub.com
Reference track matching: Yesgithub.com
WAV export: Yesgithub.com
Input formats: WAV, MP3github.com

Facts

What it does: Matchering is an open-source automated audio matching and mastering algorithm.github.com · 1 Oct 2026
Matching method: It takes a TARGET track and a REFERENCE track and produces a mastered TARGET matching the reference’s RMS, frequency response, peak amplitude and stereo width.github.com · 1 Oct 2026
Software forms: Matchering 2.0 is provided as a containerized web application, Python library and ComfyUI node.github.com · 1 Oct 2026
Desktop integration: Matchering is integrated into the UVR5 Desktop App.github.com · 1 Oct 2026
Hosted integrations: The project says users can try it without installation through Songmastr, MVSEP and Moises hosting.github.com · 1 Oct 2026
Developer integration: The Python library can be connected to everything in the Python ecosystem and has a command-line application.github.com · 1 Oct 2026
Runtime requirement: The Python library requires a machine with 4 GB of RAM and Python 3.8.0 or higher.github.com · 1 Oct 2026
Audio-format limit: MP3 loading requires installing FFmpeg separately.github.com · 1 Oct 2026
Genre guidance: The algorithm is described as working well with almost all genres, especially EDM, except experimental music with a very specific musical form.github.com · 1 Oct 2026
Neural-network status: The FAQ says Matchering does not use a neural network and instead uses digital signal processing.github.com · 1 Oct 2026
Usage rights: The FAQ says users may use their Matchered tracks wherever they want.github.com · 1 Oct 2026
Security warning: The maker says the web application is designed for home and in-house use and is not suitable for public Internet hosting without security and scalability changes.github.com · 1 Oct 2026
Public-hosting risks: The privacy page lists Django, SQLite, Redis and the Matchering worker in one container, non-scalable SQLite, DEBUG enabled and no production web server as reasons not to expose it publicly.github.com · 1 Oct 2026
License: The Python package is distributed under the GNU General Public License v3 (GPLv3).pypi.org · 1 Oct 2026