Vocal removal
One click and you have a clean instrumental. Good for karaoke night, covers and practice tracks.
v1.1.0 · Now with YouTube support
Pull any song apart with AI.
Turn any track into a clean instrumental, or split it into vocals, drums, bass, piano and more. A free Windows app that runs on your PC. Nothing is uploaded, and you don't need Python, ffmpeg or any setup.
- No account
- No uploads
- No admin prompt
Spleeter · on-device 0 bytes uploaded
- 2 · 4 · 5
- stem modes
- 3
- quality presets
- WAV · MP3 · FLAC
- export formats
- 0
- things to configure
FeaturesStudio-grade separation.
Studio-grade separation.
Zero studio required.
The settings stay hidden until you want them. Beginners get one big button, and power users get every control under Advanced options.
Multi-stem separation
Split into 2, 4 or 5 stems: vocals, drums, bass, piano and everything else.
Batch processing
Queue a whole album and leave it running. Each song runs in its own process, so if one crashes the rest still finish.
Quality presets
Fast preview, Standard or Best quality. Two engine settings combined into one simple choice.
Your format
Export WAV, FLAC or MP3 (128–320 kbps). Imports MP3, WAV, FLAC, M4A, AAC, OGG and WMA.
Drag & drop
No command line or setup. Drop songs in, choose an output folder and press go.
YouTube support
Paste a video or playlist link to send its audio straight into the queue, or just save the video or audio.
Offline after first use
Each AI model downloads once, the first time you use its mode. After that it works with no internet.
Three steps
Drag. Drop. Done.
- 01
Drop
Drag songs onto the window, click to browse, or paste a YouTube link.
- 02
Choose
Keep the defaults for a quick instrumental, or open Advanced options for stems, quality and format.
- 03
Done
Your stems are saved to the output folder, one ✅ per song as each one finishes.
Separation modes
Choose your split.
Pick how many stems you want, then choose which ones to save. It works like the "Tracks to save" checkboxes in the app, so try it below.
New in v1.1Paste a link.
Paste a link.
Get the stems.
You don't need the file first. Paste a YouTube video or playlist link and Vocal Remover pulls the audio straight into the separation queue.
Link → Queue
Add a video or a whole playlist to the batch. Every track is downloaded, then separated.
YouTube Downloader tab
Just want the file? Save audio as MP3, WAV, FLAC or M4A, or video in High, 720p or 480p.
The downloader engine updates itself in the background, so you won't need a new app version when YouTube changes something. Please only download content you have the rights to.
Add
0 songs addedClear all
Remove Vocals
The app
Clean, dark and focused.
One window: drop zone, queue, a collapsible Advanced panel and a big button.
Under the hoodPrivate by design.
Private by design.
Robust by architecture.
The app does its AI work on your own machine. A lightweight UI process hands each song to an isolated worker that loads the model, separates the song, reports progress and exits.
Your music never leaves your PC
Separation runs locally on your CPU. No uploads, no accounts, no cloud queue and no per-song fees.
Crash-isolated batches
Every song runs in its own fresh worker process. If one file runs out of memory, the rest of the album still finishes.
Powered by Spleeter
Deezer's open-source separation engine, built on TensorFlow, does the actual splitting.
Instant startup, zero setup
The installer bundles Python, TensorFlow and ffmpeg, and installs per-user with no admin prompt.
# UI process spawns one isolated worker per song
$ VocalRemover.exe --worker-job job-02.json
JOBEVENT {"event": "status", "text": "Loading 5stems model..."}
JOBEVENT {"event": "status", "text": "Separating song.mp3..."}
JOBEVENT {"event": "file_done", "outputs": [
"song_drums.wav", "song_bass.wav",
"song_piano.wav", "song_other.wav"]}
✓ worker exited cleanly → next song
# model cached at %LOCALAPPDATA%\VocalRemover
# network: noneIs it really free?
Yes. No trial, no watermark, no account and no per-song limits. It runs on your own hardware, so there are no server costs to pass on.
Windows says “Windows protected your PC”. Is it safe?
This is Microsoft SmartScreen. It shows for apps that aren’t code-signed (a certificate costs money every year, which doesn’t make sense for a free project). Click More info → Run anyway. The installer is built from the public source on GitHub and installs only for your user account, with no admin prompt.
What are the system requirements?
Windows 10 or 11 (64-bit). About 1.5 GB of disk space for the app plus the models you use. 2-stem vocal removal runs comfortably on most PCs. For 4 and 5-stem modes on long songs, 8 GB of RAM or more is recommended. No GPU is needed.
Why does it need internet the first time?
To keep the installer small, the AI models aren’t bundled. The first time you use a mode (2, 4 or 5 stems), its model downloads once (about 100–200 MB). After that, separation works fully offline. The YouTube features download a small helper tool the same way.
Do my songs get uploaded anywhere?
Never. All processing happens locally on your computer. The only things the app downloads are the AI models and the YouTube helper tool. Your audio never leaves your machine.
Which audio formats are supported?
Input: MP3, WAV, FLAC, M4A, AAC, OGG and WMA. Output: WAV, FLAC, or MP3 at 128, 192, 256 or 320 kbps.
How good is the separation?
Vocal Remover uses Spleeter, Deezer’s well-known separation model. Results are very good on most commercial music. Dense mixes and heavy reverb can leave faint artifacts. Use Best quality for final exports and Fast preview for quick checks.
Can I use the stems commercially?
The app puts no restrictions on its output, but the rights to the original song still belong to its owners. Separating a track doesn’t change who owns it. Use stems for personal practice, study, or with music you have the rights to.
I found a bug / have a feature idea.
Open an issue on GitHub, or contact me directly at voltajeba@gmail.com or on WhatsApp at +94 77 463 7185.
Ready to pull your first song apart?
Free for Windows 10 and 11. Installs in a few clicks, with no admin rights needed.
Contact
Let's talk.
Questions, bug reports, feature ideas, collaborations, or a project you'd like built? I'm Volta Jebaprashanth, the developer behind Vocal Remover. Reach out any way you like.
Star the project on GitHub