Matchering
github.com
- Free plan
- Open source
- Windows
- macOS
- Linux
- Web
- API
- Self-hosted
What Matchering is
Matchering is an open-source audio matching and mastering algorithm. It uses a target track and a reference track to produce a mastered target that matches the reference’s RMS, frequency response, peak amplitude, and stereo width. Matchering 2.0 comes as a containerized web application, a Python library, and a ComfyUI node, and it is integrated into the UVR5 Desktop App. The Python library includes a command-line application and can connect with the Python ecosystem. It requires Python 3.8.0 or higher and a machine with 4 GB of RAM. WAV and MP3 are listed as input formats, but MP3 loading requires a separate FFmpeg installation. The maker says it works well with almost all genres, especially EDM, while experimental music with a very specific form is an exception. The project says users may use their mastered tracks wherever they want. The software is free under the GNU General Public License v3, with no published paid plans. Its web application is designed for home or in-house use, not public Internet hosting without security and scalability changes.
Who it suits
Matchering suits musicians and audio users who want to match a target track’s sound characteristics to a reference. Developers can use its Python library or command-line application.
Good to know, and what to watch
- Matches RMS, frequency response, peak amplitude, and stereo width
- Available as a web app, Python library, and ComfyUI node
- Includes a command-line application
- Free software with no published paid plans
- Python library requires at least 4 GB of RAM
- MP3 loading needs a separate FFmpeg installation
- Web app is not suitable for public hosting without changes
Plans
Every plan and priceOpen-source software
Free plan
Getting started
- Open the Matchering GitHub project
- Choose the containerized web app, Python library, ComfyUI node, or UVR5 Desktop App integration
- Try hosted versions through Songmastr, MVSEP, or Moises without installation
- For the Python library, use Python 3.8.0 or later on a machine with 4 GB of RAM
- Install FFmpeg separately when loading MP3 files
- Provide a TARGET track and a REFERENCE track
Limits to check
The Python library requires Python 3.8.0 or higher and 4 GB of RAM. MP3 input requires a separate FFmpeg installation. The web application is designed for home or in-house use and needs security and scalability changes before public hosting.
The Club's review
Matchering is an open-source mastering tool that uses a reference track to shape a target recording. It is best suited to musicians who want a reference-led result and developers who want to run or integrate the processing themselves. Its no-cost, multi-form approach is compelling, but it is a poor choice for public web hosting without security and scalability work.
Overview
Provide a TARGET and a REFERENCE, and Matchering produces a mastered target intended to match the reference’s RMS, frequency response, peak amplitude and stereo width. That makes it a focused option when a particular recording is the benchmark; it is less suited to users who want a broad mastering environment rather than a reference-driven process.
Matchering 2.0 is available as a containerized web application, Python library and ComfyUI node, and it is integrated into the UVR5 Desktop App. Songmastr, MVSEP and Moises host the project for use without installation. The FAQ describes the algorithm as digital signal processing rather than neural-network based, and says users may use their mastered tracks wherever they want.
Key features
Reference matching is the core capability. It can help bring material toward a chosen track’s loudness, tonal response, peak level and stereo width, but the choice of reference matters: the algorithm follows that target rather than creating a direction from scratch. The project describes it as working well with almost all genres, especially EDM, while cautioning against experimental music with a very specific musical form.
The Python library has a command-line application and can connect to the Python ecosystem, which makes it the most flexible form for developers. That flexibility comes with a setup threshold: the library requires Python 3.8.0 or higher and a machine with 4 GB of RAM. Loading MP3 files also requires installing FFmpeg separately, so WAV is the more direct route for users who do not want an extra dependency.
ComfyUI and UVR5 integrations offer ways to use Matchering alongside those tools, while the hosted integrations provide an installation-free route. Those options broaden access, but the self-hosted web application carries an important caveat: its maker says it is intended for home and in-house use, not public Internet hosting as-is. The cited risks include DEBUG being enabled, no production web server, non-scalable SQLite, and Django, SQLite, Redis and the Matchering worker running in one container. Operators need security and scalability changes before exposing it publicly.
Pricing
The Open-source software plan costs nothing and uses the GPLv3 license. It includes reference-track matching, WAV export, and WAV and MP3 input; MP3 loading depends on separately installed FFmpeg. There are no published paid plans, so Matchering is a strong fit for users comfortable with open-source software and its setup requirements, rather than buyers looking for a supported paid tier or a tiered service quota.
Platforms
Matchering supports Linux, macOS and Windows, as well as web and self-hosted use. Its forms range from the containerized web application to the Python library and ComfyUI node, with UVR5 Desktop App integration and hosted access through Songmastr, MVSEP and Moises. This range accommodates both self-directed technical users and people who prefer to try a hosted version, though the public-hosting warning applies to operating its web application on the Internet.
Who it's for
Choose Matchering if you have a reference recording in mind and want to pursue a similar balance, level and stereo width without paying for the open-source software. It also makes sense for developers who can meet its Python requirements and want a command-line or ecosystem integration. Look elsewhere if your music depends on a highly specific experimental form, you need public web hosting without additional engineering, or you want a broad set of mastering controls rather than automated reference matching.
Pros and cons
- Pro: It targets four concrete properties—RMS, frequency response, peak amplitude and stereo width—making the reference-led workflow well defined.
- Pro: The GPLv3 software is free, with a Python library, command-line application, ComfyUI node and desktop integration to suit different workflows.
- Pro: WAV and MP3 input and WAV export cover common file needs, and the FAQ says users may use mastered tracks wherever they want.
- Con: MP3 loading requires a separate FFmpeg installation, adding setup work.
- Con: The Python route requires Python 3.8.0 or higher and 4 GB of RAM, which may rule it out on constrained or older systems.
- Con: Experimental music with a very specific musical form is outside the genres the project says it handles well, and the web app is not safe for public hosting without changes.
Alternatives
For a different mastering workflow, DSP-Quattro is a macOS freemium option with a free trial; its new-license plan is $99 a month, while its upgrade plan is $49 a month. Pick it instead if a macOS option with a trial and paid license route better fits your buying preference.
IK Multimedia ReSing is a freemium option for macOS and Windows, with a free plan and a one-time paid plan. Consider it instead if you want a different software category and its free or paid plan fits your needs.
Tunr is free for macOS and Windows; its free option allows three masters before a licence key and offers 16-bit and 24-bit output. Choose it instead if you want a free desktop option and that initial master allowance suits you.
FUSER is a paid macOS and Windows option with a free trial and a £59 one-time price. Consider it instead if you prefer a paid desktop product and its host requirements fit your setup.
Sonoris DDP Creator is a paid option for iOS, macOS and Windows with a free trial; its Standard and Pro plans cost €249 and €349, paid once, respectively. Choose it instead if its platform support and license options suit your workflow.
StudioZIO Mastering Suite is a free macOS option that supports AU, VST3, AAX and standalone use, with no licence key or account. It is a reasonable alternative if you want that plugin and standalone format rather than Matchering’s reference-matching forms.
Brainworx bx_masterdesk is a paid macOS and Windows option with a free trial and monthly plans. Consider it instead if you prefer a paid alternative to Matchering.
Voxengo SPAN is a free real-time FFT spectrum analyzer plugin for macOS and Windows. Choose it instead if spectrum analysis, rather than automated mastering, is what you need.
Browse Audio Mastering Software or AI Music Mastering Software for more options.
Verdict
Matchering is an easy recommendation for musicians with a suitable reference and developers comfortable with its open-source setup: it delivers a focused matching workflow at no cost and in several software forms. Its main drawbacks are the extra FFmpeg dependency for MP3 and the web application’s unsuitability for public hosting as-is. If you need reference-led mastering and can work within those limits, start with Matchering; if you need a broadly configurable mastering environment or a public web service without additional engineering, look elsewhere.
The verdict
Matchering offers a focused way to master a target against a reference, with several software forms and no published paid plans. Note the MP3 dependency and the web app’s hosting warning before choosing how to run it.
Questions people ask
Is Matchering free?
Yes. It is distributed as free GPLv3 open-source software, with no published paid plans.
What does Matchering match?
It uses a TARGET and REFERENCE track, then matches RMS, frequency response, peak amplitude, and stereo width.
Which formats can it load?
WAV and MP3 are listed. MP3 loading requires installing FFmpeg separately.
Does Matchering use a neural network?
No. Its FAQ describes digital signal processing instead.
Can Matchering be hosted publicly?
The web application is intended for home or in-house use. Public Internet hosting requires security and scalability changes.
What license covers the Python package?
The Python package is distributed under the GNU General Public License v3.
What the maker publishes
- Album consistency
- The documentation presents using one reference across multiple tracks as a way to make an album sound more cohesive. source
- Audio-format limit
- MP3 loading requires installing FFmpeg separately. source
- Audio formats
- An independent software directory describes WAV and FLAC input/output support and professional WAV output. source
- ComfyUI integration
- A ComfyUI-Matchering node is offered as an integration route. source
- Command line
- The repository points developers to a separate Matchering CLI and an enhanced fork of that CLI. source
- Desktop integration
- Matchering is integrated into the UVR5 Desktop App. source
- Developer integration
- The Python library can be connected to everything in the Python ecosystem and has a command-line application. source
- Docker deployment
- The repository provides separate Docker instructions for Windows, macOS and Linux, plus an update guide. source
- Documentation
- The repository includes a README, FAQ wiki, Docker guides, examples, log-code documentation and limiter testing material. source
- Genre guidance
- The algorithm is described as working well with almost all genres, especially EDM, except experimental music with a very specific musical form. source
- Hosted access
- The project README identifies Songmastr, MVSEP and Moises as services that host Matchering for browser-based trials. source
- Hosted integrations
- The project says users can try it without installation through Songmastr, MVSEP and Moises hosting. source
- Independent assessment
- Hysen Labs describes the project as matching four measurements with a project-specific limiter and notes that the newest tagged release is from 2022. source
- License
- The Python package is distributed under the GNU General Public License v3 (GPLv3). source
- License
- Matchering is distributed under the GNU GPL version 3.0 license. source
- Limiter
- Version 2 introduced the project’s own open-source Hyrax brickwall limiter. source
- Limit: interface and controls
- Beathoven characterizes Matchering as having a less polished interface and fewer adjustable parameters than commercial services such as LANDR or eMastered. source
- Limit: reference dependence
- Beathoven warns that good results depend on choosing a suitable reference track. source
- Limit: service outage
- A Songmastr user reported that the hosted site had gone down, while another later comment said it was back online. source
- Limit: setup knowledge
- ToolStock says self-hosting requires basic technical knowledge for installation. source
- Limit: upload errors
- Reddit commenters reported browser bad-request errors and later said they could not upload songs to the hosted service. source
- Linux dependency
- Linux installations require the system libsndfile library to be installed separately. source
- Maker identity
- The package credits Sergree as author and lists one maintainer on PyPI. source
- Maker location
- The maintainer’s GitHub profile identifies Sergey Grishakov and lists Moscow, Russian Federation. source
Checked against the maker's pages on 21 Oct 2026.