StemRoller
stemroller.com
- Free plan
- Windows
- macOS
What StemRoller is
StemRoller is a desktop app that separates songs into vocals, drums, bass, and other instrumental parts, then combines the non-vocal stems into an instrumental track. Search by song title and StemRoller finds results online, using YouTube search according to its project README. Its separation process uses Demucs, a machine-learning project, and the README says processing takes several minutes. The maker describes StemRoller as free and open source; the README offers either the Unlicense or MIT No Attribution license. It runs on macOS and Windows, with the homepage linking to a macOS ZIP and a Windows CUDA executable, version 3.1.1. The Free plan is free and provides four stem outputs and an instrumental track. There is no batch processing. Linux is not officially supported, though the README gives manual dependency setup instructions. The maker directs people with questions or issues to Discord.
Who it suits
StemRoller suits people who want to make vocal, karaoke, or instrumental tracks from songs on macOS or Windows. It may also suit users comfortable with manually setting up Linux dependencies.
Good to know, and what to watch
- Separates vocals, drums, bass, and other stems
- Searches for songs by title
- Creates an instrumental from non-vocal stems
- Free to use and described as open source
- Offers help through Discord
- Splitting takes several minutes
- Batch processing is not available
- Linux is not officially supported
Plans
Every plan and priceFree
Free plan
The verdict
StemRoller combines song search with four-part stem separation and instrumental creation. Its free desktop offering has a clear platform limitation: Linux requires manual setup and is not officially supported.
What the maker publishes
- Algorithm
- StemRoller uses Demucs, a machine learning project, to separate stems. source
- Downloads
- The homepage links to a macOS ZIP and a Windows CUDA executable, version 3.1.1. source
- Instrumental
- It creates an instrumental by mixing together all non-vocal stems. source
- Integrations
- StemRoller uses YouTube search results to find songs, according to its GitHub README. source
- License
- The maker describes StemRoller as open source; its README offers the Unlicense or MIT No Attribution license. source
- License
- The repository offers a choice of Public Domain (Unlicense) or MIT No Attribution licensing. source
- Linux support
- Linux is not officially supported; the README gives manual dependency setup instructions for Linux. source
- Machine learning
- The product is based on Facebook's Demucs machine learning research project. source
- Open source
- StemRoller describes itself as open source. source
- Platforms
- The maker's releases page provides downloads for Windows and macOS. source
- Price
- The maker describes StemRoller as totally free. source
- Processing time
- The project README says splitting takes several minutes. source
- Processing time
- The GitHub README says stem splitting takes several minutes. source
- Purpose
- StemRoller makes karaoke, vocal, and stem tracks from songs. source
- Purpose
- StemRoller creates karaoke, vocal, and stem tracks from songs. source
- Song search
- Users can search for a song by title, and StemRoller finds it online. source
- Song search
- Users can search for a song by title and StemRoller finds it online. source
- Stem separation
- It splits songs into vocals, drums, bass, and everything else, and creates an instrumental mix from the non-vocal stems. source
- Stem separation
- It splits songs into vocals, drums, bass, and everything else. source
- Support
- The maker invites users with questions or issues to get help on Discord. source
- Support
- The maker offers help through its Discord server. source
- YouTube
- The project README says StemRoller integrates search results from YouTube. source
Checked against the maker's pages on 5 Oct 2026.