Light and fast
Light enough to run at scale on ordinary CPUs, which is what makes catalogue-sized batch jobs affordable.
Spleeter
Deezer
The separation tool that made stem splitting ordinary.
In short
Spleeter is Deezer's open source source-separation library, splitting a mix into two, four or five stems. It is fast and light on resources, and has been superseded on quality by Demucs while remaining useful for batch processing.
Spleeter specifications
- Output
- 2, 4 or 5 stems
- Architecture
- Spectrogram masking
- Maintenance
- No major release since 2022
- Dependency
- TensorFlow 1.x
- Status
- Evaluating
Figures verified 2026-08-10. This field moves quickly — re-check before relying on them.
About Spleeter
Spleeter's historical importance is hard to overstate: it made stem separation a thing anyone could do, and an entire generation of remix and karaoke tools was built on it. Its enduring advantage is efficiency. It is light enough to run on ordinary hardware, fast enough to process a large catalogue without a GPU budget, and simple enough to reason about when something goes wrong. For batch work across thousands of files where good-enough separation is genuinely good enough, that profile still wins. On quality it has been clearly overtaken. Demucs produces cleaner separation with better transient preservation, and the gap is audible on any material more complex than a sparse pop mix. Choose Spleeter for throughput and simplicity, Demucs when the stems have to survive close listening.
Strengths
What Spleeter is good at
Historically important
It made separation ordinary, and the resulting ecosystem of tutorials and wrappers is still the largest of any separation tool.
Simple to reason about
A simple, well-understood pipeline is easy to debug when a file fails, which matters more in production than peak quality does.
Spleeter limitations
- Effectively unmaintained
- Beaten by Demucs on every benchmark
- Ageing dependency stack
How it compares
Choosing Spleeter
What comes back
Spleeter is Deezer's open source source-separation library, splitting a mix into two, four or five stems. It is fast and light on resources, and has been superseded on quality by Demucs while remaining useful for batch processing.
Compare with MusicGenerate
Spleeter review, pricing and alternatives
Searched as spleeter, deezer spleeter, spleeter github and free stem separation. It made separation ordinary, which is why the ecosystem around it is still the largest.
Spleeter — questions people ask
Is Spleeter still worth using?
For batch work across thousands of files where good-enough separation is genuinely good enough, its efficiency still wins. For stems that must survive close listening, use Demucs.
How many stems can Spleeter produce?
Two, four or five, depending on the model you pick. There is very little else to configure, which is precisely the appeal for batch deployment.
Why is Spleeter faster than Demucs?
It is a lighter model that runs happily on ordinary CPUs, which is what makes catalogue-sized jobs affordable without a GPU budget.
Is Spleeter free?
Yes, open source under a permissive licence, so deploying it is unproblematic. The licence covers the software and grants no rights over the recordings you process.
Is Spleeter still maintained?
Development has slowed considerably, and Demucs has taken over as the actively developed reference. Spleeter still runs fine, which is why it persists in batch pipelines.
How many stems should I split into?
Two if you only need vocals and instrumental, four for drums and bass separately. More stems means more chances for the algorithm to misattribute sound.
More models
More AI music models and generators
Models and platforms we track, compared on capability and licence.
