Spleeter
Advanced audio source separation with pre-trained deep learning models
About Spleeter
Spleeter is an open-source Python library developed by Deezer for music source separation. It provides pre-trained models to split audio tracks into vocals, drums, bass, piano, and other stems. Spleeter supports command-line and Python library usage, with GPU acceleration for fast processing. It's widely adopted in professional audio software and research.
Pricing
Full pricing page Free option
Free
$0 per month
The basics for individuals and organizations
- Unlimited public/private repositories
- Host open source projects in public GitHub repositories
- Dependabot security and version updates
- 2,000 CI/CD minutes/month
- 500MB of Packages storage
- Issues & Projects
- Community support
Popular
Team
$4 per month
Advanced collaboration for individuals and organizations
- Everything included in Free
- Access to GitHub Codespaces
- Repository rules
- Multiple reviewers in pull requests
- Draft pull requests
- Code owners
- Required reviewers
- Pages and Wikis
- Environment deployment branches and secrets
- 3,000 CI/CD minutes/month
- 2GB of Packages storage
- Web-based support
Enterprise
$21 per month
Security, compliance, and flexible deployment
- Everything included in Team
- Data residency
- Enterprise Managed Users
- User provisioning through SCIM
- Enterprise Account to centrally manage multiple organizations
- Environment protection rules
- Repository rules
- Audit Log API
- SOC1, SOC2, type 2 reports annually
- FedRAMP Tailored Authority to Operate (ATO)
- SAML single sign-on
- Advanced auditing
- GitHub Connect
- 50,000 CI/CD minutes/month
- 50GB of Packages storage
FAQ
Yes, you can try Spleeter without installing it by using the dedicated notebook through Google Colab.
The released models were trained on spectrograms up to 11kHz, so frequencies above 11kHz are discarded. However, you can use alternate config files like spleeter:2stems-16kHz to perform separation up to 16kHz, or modify the F parameter in the config file to go up to 22kHz.
If no output is produced, it usually means the separation process crashed, often due to lack of memory. You can try processing smaller segments of your input file or using the -d option to set a duration for processing.
Some users have reported issues running Spleeter on Apple M1 chips. A workaround is available, but you may need to check the documentation for specific instructions.
This error occurs if the pretrained model files are not correctly downloaded or found. To fix it, remove the pretrained_models folder completely and run Spleeter again. It will download the models again.
If the stems are identical to the input, it means no separation was performed because the library didn't find a model for separation. This could happen if the model download step failed. You can manually download the model archive from the release page as a workaround.
Alternatives to consider
Community ratings & full list
Pricing summary
Free tier available
Starting from
Free plan available, Team plan starts at $4 USD per user/month
Categories
Claim this tool
Are you the founder? Claim your profile to update details and track views.
Claim tool