Matchering
An open-source reference mastering tool. You give it your mix and a reference track, and it matches loudness, tone, peak level, and stereo width to that reference.
- Type Mastering
- Difficulty Easy
- Hardware The Python library asks for a machine with at least 4 GB of RAM and Python 3.8 or newer. No GPU is required. Windows, macOS, and Linux are supported. On Linux you may need to install libsndfile, and FFmpeg if you want MP3 loading. Installing the library or Docker image yourself is Medium; the UVR desktop integration is Easy.
- License GPL-3.0
What it does
Matchering, by Sergree, takes two files: the target (your song) and the reference (a record you want to aim at). It then processes the target so its loudness, frequency response, peak amplitude, and stereo width line up with the reference, and it writes a mastered WAV.
It is not a full mastering engineer. It will not fix a messy arrangement, clipping drums, or a vocal that is too quiet in the mix. It will get your bounce in the same loudness and tonal neighborhood as a record you already like, which is often the missing last step on a home mix.
The project ships as a Python library, a Docker-based web app, a ComfyUI node, and, according to the README, an integration inside the Ultimate Vocal Remover desktop app. The README also lists third-party hosted demos if you want to try it before installing.
Why it matters for hip-hop
Rap mixes live and die on low end and vocal presence. A common problem is a beat that sounds huge in the headphones and small next to a commercial record, or an album where song three is much quieter than song one. Matchering is built for that: pick a reference in the same lane and run your mix against it.
Use a reference that actually fits the song. Matching a quiet record to a maxed streaming single, or the other way around, will not help.
Getting it running
The easiest official paths, for a producer, are the UVR desktop integration or the Docker app. Developers can install the Python package and call it from a script or a command-line tool. None of that requires a GPU. If you are not used to Python or Docker, start inside UVR.
License notes
Matchering is licensed under GPL-3.0. You can use the program, including on commercial releases. If you distribute a modified version of the software itself, or build it into a product you ship, GPL-3.0's share-alike rules apply — that is a software-licensing issue, not a tax on your masters. The tool does not give you rights to the reference track. Pick a reference you have the right to use as a sonic target.