Remix and mashup
Drop the vocal from one song over the instrumental of another, match the tempo and key, and build a mashup in your DAW. Isolated stems give you the freedom to rearrange, re-pitch and re-time every part of the track.
Find BPM & key →All-in-One AI Audio Studio
AI Stem Splitter — Separate Vocals, Drums, Bass & Instruments
Agent mode
Chat with the AI producer to shape style and lyrics — full renders from 15 credits each.
Open the full AI music generator →Pro voices · 1 credit per tag
Open the producer tag generator →- Vocal remover
- Acapella extractor
- Karaoke maker
- Instrumental maker
- Drum stem isolation
- Bass line extraction
- Remix & mashup prep
- Sample chopping
- Practice tracks
- Cover backing tracks
How it works
Split a song into stems in three steps
Sign in from any browser — no plugins or DAW required.
- 01
Upload your track
Sign in, then drag in an MP3, WAV, FLAC, M4A, AAC or OGG file up to 50 MB, or pick one from your phone.
- 02
AI separates the stems
Chorumi splits the song into vocals, drums, bass and other instruments — or a vocal and instrumental pair in 2-stem mode.
- 03
Preview & download
Listen to every stem in the browser, then download each one as MP3 or WAV.
What you can do with stems
One song in. Six new ways to use it.
An acapella extractor, a vocal remover and a practice tool — all from the same AI stem split.
Make an instrumental or backing track
Use 2-stem mode to separate vocals from instrumental and get a clean backing track for live performances, videos, dance routines or any content that needs the music without the lead vocal.
Practise drums and bass
Mute the drum stem and play along with the rest of the band, or solo it to study fills, ghost notes and groove. The same approach works for learning a bass line note by note.
Sample and chop
Producers extract a drum break, a bass riff or a vocal phrase and chop it into new beats. Remember that sampling copyrighted music requires clearance before you release the result.
Karaoke nights
Remove the lead vocal and keep the music to create karaoke tracks of your favourite songs. Keep the vocal stem handy so you can check pitch and lyrics while you rehearse.
Change the key →Covers and learning
Singers and instrumentalists use stems to hear exactly how a part is performed, then record their own cover over the instrumental. Slowing down an isolated part also makes it much easier to transcribe.
Slow it down →Why AI stem separation
A real split, not an EQ trick
Old vocal removers subtract frequencies. An AI stem splitter rebuilds each part, so the instrumental keeps its punch and the acapella stays clean.
| Feature | Karaoke tracks | EQ & phase tricks | Best for stemsChorumi AI stem splitter |
|---|---|---|---|
| Works on any song you upload | — | ✓ | ✓ |
| Uses the original recording | — | ✓ | ✓ |
| Gives you the vocal as its own file | — | — | ✓ |
| Keeps full drum and bass energy | ✓ | — | ✓ |
| Separate drums, bass and other | — | — | ✓ |
The Chorumi studio
Free music tools for every step
Stem separation is just the start. Generate songs, brand your beats and reshape any track.
Create
AI Music Generator
Text prompt to full song
Open AI Music Generator →Tags
Producer Tag Generator
Voice tags for your beats
Open Producer Tag Generator →Lyrics GeneratorVerses, hooks and bridgesSoon
AI Producer TagTags from your producer nameVoice CloneCustom voice for your tagsStem SplitterYou are hereHereBPM Key FinderDetect tempo and keyPitch ChangerShift the key of any songBPM ChangerChange tempo of any trackSlowed ReverbMake slowed and reverb editsOne credit system. Two ways to pay.
One balance for voice tags, AI songs, sound FX, and stem splits — subscribe monthly or buy credits once.
Flexible monthly billing. Upgrade, downgrade or cancel anytime.
Starter
Taste the full studio.
$9.99/ mo
Billed monthly
500 / month
≈ 33 full songs · 41 stem splits · 500 Pro tags
Up to 2 concurrent generations
- All AI voices, including Pro voices
- Up to 2 concurrent generations
- Full AI songs
- Stem splitter
- ai producer tag
- WAV + 320kbps MP3 export
- Full commercial license
- No daily generation limits
Pro
MOST POPULARFor beatmakers in the lab every day.
$24.99/ mo
Billed monthly
1,500 / month
≈ 100 full songs · 125 stem splits · 1,500 Pro tags
Up to 5 concurrent generations
- Everything in Starter
- Up to 5 concurrent generations
Studio
For heavy users and small teams.
$49.99/ mo
Billed monthly
4,000 / month
≈ 266 full songs · 333 stem splits · 4,000 Pro tags
Up to 10 concurrent generations
- Everything in Pro
- Up to 10 concurrent generations
What Is Stem Separation?
A finished song is one blended file. Stem separation pulls it back apart into vocals, drums, bass and everything else.
A stem splitter (source separation) turns that single mix into separate files — vocal, drums, bass, and other parts — from a normal stereo upload in your browser after you sign in. Compare AI stems vs karaoke and EQ tricks.
Supported Formats
Chorumi accepts the most common audio formats and lets you export your stems in the format that suits your workflow. The maximum upload size is 50 MB per file. For step-by-step upload and download, see how it works.
| Format | Upload | Download | Best for |
|---|---|---|---|
| MP3 | Yes | Yes | Small files and quick previews |
| WAV | Yes | Yes | Lossless stems for production |
| FLAC | Yes | No | Lossless source for the cleanest split |
FAQ
Stem splitter questions, answered
Credits, formats, and modes — before you split your first song.
Split a song ↑How much does stem separation cost?
Each split uses music credits from your Chorumi balance — the same pool as AI song generation. Pricing is the same for 2-, 4-, and 6-stem modes. New accounts receive signup credits; you can buy more on the pricing page. Failed jobs are refunded automatically.
Do I need an account?
Yes. Sign in with Google, Discord, or email before you upload or split. Your jobs, downloads, and credit balance stay tied to your account, including splits started from AI Music Generator tracks.
What audio formats are supported?
You can upload MP3, WAV, FLAC, M4A, AAC and OGG files up to 50 MB. Stems can be downloaded as MP3 or lossless WAV, individually or as a ZIP.
How long does it take?
Most songs finish within seconds to a few minutes, depending on the length of the track and how busy the service is. Keep the page open and your stems will appear in the same view.
How many stems can I get?
Choose 2 stems (vocals and instrumental), 4 stems (vocals, drums, bass and other), or 6 stems (adds separate guitar and piano). Four stems is the default for remix and practice workflows.
Can I make an acapella?
Yes. The vocals stem is an acapella of the song. Both the 2-stem and 4-stem modes give you a vocal-only file that you can download as MP3 or WAV and use as an acapella extractor result.
Can I use the stems commercially?
The stems are derived from your original recording, so the rights stay with whoever owns the song. If you own the rights or hold a licence, you can use the stems commercially. If not, get permission from the rights holder before you release remixes, samples or covers.
What's the max file size and length?
Files can be up to 50 MB per upload. That covers most radio-length MP3s; trim very long DJ sets or export a shorter section from your DAW if needed.
Does it work on mobile?
Yes. The stem splitter runs in the browser on iOS and Android. Choose a file from your phone, split it and download the stems without installing an app.
2-stem vs 4-stem vs 6-stem — which should I pick?
Pick 2 stems if you only need an acapella or an instrumental — karaoke, backing tracks, or a quick vocal remover job. Pick 4 stems when you want drums and bass on their own for remixing, practice, or sampling. Pick 6 stems when you also need separate guitar and piano tracks.
Are my files kept private?
Your jobs and downloads stay tied to your account. Uploaded audio and finished stems are removed from our servers after about 24 hours as part of routine cleanup. We do not run a public gallery of user splits.
What if the result sounds bad?
Try a higher-quality source such as WAV or FLAC, switch between 2-stem and 4-stem mode, and avoid tracks that were already processed by another vocal remover. Mixes with heavy reverb or overlapping sounds are naturally harder to separate, and a faint trace of one stem inside another is normal.
Agent mode · Chat free · Credits on render
Describe it. Hear it.
Tell the AI producer what you want — mood, tempo, vocals or instrumental — refine the plan together, then render and download your track.