Hi-Fi stems available on Pro plan
Stem separation for professional work
Stem Separation:
Isolate any part. Keep every detail.
Separate vocals, drums, bass, guitar, keys, and more from any audio file. Clean, isolated stems with the accuracy producers and engineers depend on.


How it works
From upload to isolated stems, in minutes.
Upload your audio
Drag in a file, paste a link, or pull one from your device or cloud storage. Audio or video: MP3, WAV, FLAC, AIFF, MP4, MOV, and WMA.
Choose your model and stems
Standard for speed, or Hi-Fi Models when accuracy is non-negotiable. Separate into 27 stem types, including drums, bass, guitar, keys, strings, and male and female vocals.
Download session-ready stems
Export individual stems or the full mix in MP3, WAV, FLAC, M4A, or OGG.
Studio-grade stem separation
World-class stem separation, for every part of your mix.
Over 27 stems to mute or solo. Hi-Fi models available for when every frequency matters. The result is clarity in the vocal, tighter instrument isolation, and fewer artifacts, even on dense mixes. Stems that hold up when you process them.
+27 stems available:
- Lead/Backing Vocals
- Male/Female Vocals
- Acoustic/Electric Guitar
- Rhythm/Solo Guitars
- 킥
- 스네어
- 하이햇
- 탐탐
- 심벌
- 현악기
- 퍼커션
- 건반악기
- 피아노
- Grand Piano
- Woodwind
- 관악
- 브라스
- Reeds
- 기타
Every instrument, isolated.
Vocal removal is one job. This does all of them.

Separate the full mix, part by part.
Reach for any of 27 stem types when you want something specific, down to a single snare or a backing vocal. You can even split vocals further, into male and female.

Automatic separation finds the parts for you.
Choose Automatic when you're not quite sure what you want to isolate. It listens to the track, then separates up to 7 top-level parts on its own.

Export exactly what you produced.
Take the stems individually or the whole mix at once, in lossless quality at up to 192 kHz. Which means what you hear in the session is what you hear after export.
Separate your first track free.
Upload a song, pick the parts you need, and hear them on their own.
Why Moises
Quality that holds up in production.
Built for the work you put your name on.

Stems that don't need a cleanup pass.
Minimal bleed, minimal artifacts, and Hi-Fi Models when the mix demands more. Clean enough to go straight into the work.

Separation is the start, not the export.
Stems load directly into Moises Studio: edit, arrange, generate new parts, master, and hand off to your DAW. Other separators end at the download. Moises is where the work continues.

Used by working professionals.
Musicians who make their living on the output, from bedroom producers to 5x Grammy winner Cory Henry.

Your audio stays yours.
We never train our models on your uploads. Your audio is processed for your session only.
전문가들이 신뢰하고, 8,000만 명 이상의 뮤지션이 사랑해요.

Charlie Puth
"저는 수년간 저만의 창작 과정에서 Moises를 사용해 왔습니다. Moises는 아티스트들이 배우고, 탐구하고, 아이디어를 실현할 수 있도록 돕기 위해 만들어졌습니다."

Eloy Casagrande
"평생을 기다려 온 제품입니다. Moises가 10년, 15년 더 일찍 나왔더라면, 제 삶이 훨씬 더 편했을 거예요."

Cory Henry
"곡이 막 시작되는 단계에서도 Moises로 스템을 생성하고 베이스 파트를 만드는 등 다양한 창의적인 방식으로 활용하고 있어요."
Only need to remove vocals from a song?
That's a simpler job, and we built a dedicated tool for it.
자주 묻는 질문
Stem separation splits a finished mix into parts, or “stems.” Instead of isolating frequencies by hand, you can let AI detect the spectral and timing patterns that define each instrument or voice. The outcome is a set of standalone audio files, so you can remix, master, or repurpose material without endless trial and error.
Hi-Fi separation uses larger, more powerful models to deliver the absolute best quality available. This can take a little longer, so use it for workflows where quality is a priority over speed.
Moises offers 27 different separation targets at our highest subscription tier. More recent separation targets include Male Vocals, Female Vocals, Brass, and Reeds.
The biggest difference is quality. Moises offers world-class stem separation trusted by Grammy-winning artists, professional studios, and over 80 million musicians worldwide. Another major difference is ethics. Our AI models are trained exclusively on licensed material. We never train on music we do not have permission to use.
Some level of impact is expected with any stem separation. The difference depends on the tool you use. High-quality models deliver the best possible result with fewer artifacts.
The automatic stem detection feature uses AI to find up to 7 separation targets on its own and then separate them automatically.