MSST-WebUI

MSST-WebUI

Runs music source separation models from a browser page, with UVR support, preset processing flows and extra tools. For producers and audio researchers who split recordings they own.

Made by Sucial on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: MSST-WebUI is a browser interface for Music-Source-Separation-Training, a project for building models that split music into its sources. You use the page to run MSST models and VR models. This Hugging Face repository holds the official models and packaged installers for the interface.

Who it suits: Producers, audio engineers and researchers who split recordings into parts are the main fit. Hobbyists who work with singing can use the bundled MIDI tool too.

Industry | Who uses it | How they use it | Example
Music production | Producers | Split a recording into its sources | Separate a mix you own into its parts
Audio research | Researchers | Run several models and compare the results | Combine model outputs in the advanced ensemble mode
Music tech | Hobbyists | Pull MIDI notes out of singing | Run the bundled SOME tool on a vocal line

If you need a hosted service or plan to use it commercially, read its license terms first, or skip it.

Where to use it: You work in a browser tab, and you fetch the files with Python.

• A browser, where the WebUI shows its pages
• Windows, macOS or Linux, with Python for the download
• A Google Colab notebook, which the card links to

Use case | Who | What they get back
Source separation | Producers | Separated audio from an MSST model or a VR model
Custom processing | Audio engineers | A processing flow you define yourself on the preset process page
Reusing UVR models | Existing UVR users | No second download of VR models when you point Settings at your UVR5 folder
Singing to MIDI | Hobbyists | MIDI extracted from singing with the SOME tool

How to set it up: You need Python for the download and a place to run the WebUI. The card gives no setup commands beyond the download.

• Python with pip
• A terminal on Windows, macOS or Linux
• Free disk space for the models and installers

1. Install or upgrade the Hugging Face hub package.
2. Download the repository files, which hold the models and packaged installers.
3. Run the packaged installation program, or follow the project's own code repository for details.
4. In the WebUI, open the Install Models interface to add the models you want.

pip install -U huggingface_hub
hf download Sucial/MSST-WebUI

How to use it: Open the WebUI in your browser, add a model on the Install Models page, and run it on an audio file. Use Settings to choose an existing UVR5 model folder.

Tool | What it does
Install Models | Adds MSST models and VR models to the WebUI
Settings | Lets you pick your UVR5 model folder so VR models are not downloaded again
Preset process page | Lets you customise the processing flow yourself
SOME | The Singing-Oriented MIDI Extractor, built into the WebUI as a tool
Advanced ensemble mode | An ensemble mode for combining the results of several models

Example: You open the Install Models page and add an MSST model. Then you load a song you have the right to use and run the model on it. The WebUI returns the separated audio.

For newcomers and experts: This is a page in your browser that splits a song into separate parts using AI models. A good first try is to download the files, run the packaged installer and process one short recording that you have the right to use.

Under the hood, the WebUI wraps the MSST training project's models together with the VR models from UVR. It adds preset flows, an advanced ensemble mode and the SOME tool. The trade-off is documentation, because the card gives no model list or benchmark and sends you to the project's code repository.

Install the WebUI and its models
Pick an MSST model or a VR model
Load your audio file
Run a preset flow or a single model
Collect the separated audio

Good to know: The card is short, and the project's code repository holds the fuller notes.

• Reuse: if you already have UVR, point Settings at its model folder and skip the VR downloads
• Rights: use only recordings you own or have permission to process
• License: cc-by-nc-sa-4.0, so read its terms before any commercial use
• Cost to run: none named
• Data it reaches: the audio you load, the models you install and, if you choose it, your UVR5 model folder
• Not for: the card names no exclusions, so check the license terms before commercial use

On Hugging Face, 13 November 2026

• Downloads in the last 30 days: 0
• Likes: 50
• License: cc-by-nc-sa-4.0

Links

• Model: https://huggingface.co/Sucial/MSST-WebUI
• Files: https://huggingface.co/Sucial/MSST-WebUI/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download Sucial/MSST-WebUI

Then follow the usage in its model card. Nothing in this kit runs by itself.