What are the 4 stems produced?

The song is split into four separate MP3 files: vocals, drums, bass, and other (all remaining instruments).

What is the maximum file size?

You can upload an audio file up to 25MB.

How long does processing take?

A typical 3-minute song can take roughly 3-5 minutes of CPU processing time; longer tracks take proportionally longer.

Which audio formats are supported?

We support MP3, WAV, M4A and FLAC.

How good is the separation quality?

The hybrid transformer model used is one of today's most advanced open-source separation technologies; however, in passages where vocals and instruments overlap heavily, slight bleed between stems can occur.

How do I download the result?

Once processing completes, a single .tar archive containing all 4 MP3 files is ready to download automatically.

Can I separate and redistribute a copyrighted song?

No, this is not recommended. The tool only performs technical separation — it does not transfer copyright. You should only process audio you own or have legal rights to use, and never redistribute copyrighted material without permission.

Yes, the tool is currently offered free of charge.

Can I upload multiple songs at once?

No, the tool processes one file at a time. For multiple songs, repeat the process for each one.

Why does it run on CPU instead of GPU?

Our server infrastructure runs on CPU, which makes processing slower than GPU-based professional studio software, but lets us offer the tool free without extra hardware cost.

Music Source Separation — Free Online Tool

Drop audio file

max 25 MB · MP3/WAV/M4A/FLAC · 3-min song ≈ 3-5 min CPU · Max 25MB

How it works: An AI hybrid transformer model splits the mix. Output: 4 MP3 files (192 kbps) in a .tar archive.
Legal: Use only with audio you own or have rights to. Republishing copyrighted material may infringe.

Auto-deleted in 7 days

256-bit SSL

Support

Your files are encrypted with SSL and automatically deleted from our servers 7 days after processing.

📖 How to Use

Upload audio

MP3/WAV/M4A/OGG.

Pick operation

Format / bitrate / trim / transcript.

Start

Server completes the operation.

Download

Get the result.