Free online vocal remover

Remove vocals from any song online, free

Upload a file, split vocals and accompaniment in the browser, and hear the result before spending money on another tool. No signup, no credits, no catch.

  • free vocal remover
  • no signup needed
  • karaoke and remix prep
  • vocals + accompaniment

Drop a song here — or tap to try it on your track

Free, in your browser. No signup. MP3, WAV, FLAC, M4A, OGG, or video.

Choose a file
Want 5-stem (drums, bass, piano)? iOS App Android App

What to expect from the output

Separation quality depends on the song. Knowing the limits upfront saves time.

Clean mixes work best

Centered, dry vocals in simple arrangements separate the cleanest. Pop, R&B, and acoustic tracks with clear vocal placement give the most reliable results.

Dense tracks are harder

Heavy reverb, layered production, and 808-heavy mixes leave more bleed in the output. Live recordings with room ambience are the toughest.

Some artifacts are normal

Warbling and underwater texture show up in difficult passages with every tool, not just Unmix. These are most noticeable on sustained vowels and reverb tails.

How to remove vocals in 3 steps

1. Upload

Drop an MP3, WAV, or supported video/audio file. Processing happens locally in your browser.

2. Separate

The browser splits your track into vocals and accompaniment. Takes seconds on most files.

3. Preview and export

Listen before downloading so you only keep usable results. Skip to the chorus first — that is usually the hardest section.

Supported formats and quality tips

The format of your source file affects separation quality. Here is what to know.

MP3 — most common

Works fine for most use cases. 320kbps MP3 separates noticeably cleaner than 128kbps. Check sibilant sounds (S and T) first — they reveal warbling artifacts fast.

WAV — cleanest input

Lossless audio gives the AI more detail to work with. If quality matters for your project, start with WAV. Better input means cleaner stems and more headroom for post-processing.

Video files (MP4, etc.)

You can upload video files directly. The audio track gets extracted and separated. Audio quality inside the video matters more than video resolution.

Will this work on your song? A quick reality check by genre

Before uploading, know what to expect. Separation quality varies by how a track was recorded and mixed — not by the tool.

Works well: ballads, acoustic, clean pop

Songs with one centered vocal, sparse arrangement, and little reverb come out clean. Think Adele-style ballads, James Bay acoustic sessions, or most singer-songwriter material. Expect minor warbling only on long sustained notes and reverb tails. These are the tracks the browser model was built to handle first.

Mixed results: hip-hop, EDM, heavy production

Modern pop with autotune stacks, trap with 808s, and dense EDM produce usable but imperfect stems. Listen for warbling on sustained notes, musical noise (a swirly metallic shimmer), and 808-bleed where sub-bass merges with kicks. Pre-echo — a faint ghost of the vocal in the instrumental — is normal on heavily compressed masters.

When to use something else

For dense live recordings, 128 kbps MP3s, or songs where you need vocals, drums, and bass separated individually, the browser 2-stem model isn't the right tool. The Unmix mobile app does 5-stem locally. For release-quality masters, a dedicated desktop tool like LALAL.AI trades speed for fewer artifacts.

Looking for something else?

Test it on your own track

Upload any song and hear the separated stems in seconds. Free, no account needed.

Tips for better results

Upgrade from MP3 to WAV

If your MP3 result has artifacts, try the same track as WAV. The lossless detail gives the AI more to work with.

Check the chorus first

The densest section reveals problems fastest. If the chorus is clean, the rest is almost certainly usable.

Bitrate matters for MP3

320kbps MP3 separates much cleaner than 128kbps. If you only have a low-bitrate MP3, expect more warbling.

FAQ

Can I remove vocals perfectly from any song?

No. AI separation is useful but not perfect. Results vary by song, genre, and how the track was mixed.

What file formats work?

MP3, WAV, M4A, OGG, FLAC, and most common audio and video formats. WAV and FLAC give the cleanest results because there are no compression artifacts.

Does MP3 quality affect the result?

Yes. Higher bitrate MP3s (320kbps) separate cleaner than low-bitrate files (128kbps) because the AI has more detail to work with.

Can I remove vocals from video files?

Yes. Upload a video file directly and the audio will be extracted and separated automatically.

Why does my output sound underwater?

Warbling artifacts are common on dense mixes and reverb-heavy tracks. Try a cleaner source file, a higher bitrate version, or test a different section of the song.

How long can my track be?

Length caps depend on the quality preset and your device's memory. On Balanced in a desktop browser, aim for under 6 minutes (max 9). Fast handles up to 12 minutes. Mobile browsers are tighter — usually 2–3 minutes before the browser runs out of memory. If your song is longer, trim to the section you actually need or switch to the Unmix mobile app.

Will YouTube or Spotify detect the instrumental as copyrighted?

Very likely, yes. Content ID systems fingerprint the underlying musical composition — melody, chord progression, production — not just the vocal. Removing vocals doesn't strip the fingerprint. Treat any separated stem from a released track as copyrighted material. Personal practice, karaoke, and private use are generally fine; public uploads, monetization, or commercial release need a license from the rights holders.

Can I process multiple songs at once?

No — the browser tool handles one file at a time. Running multiple separations in parallel would compete for WebAssembly memory and freeze the page. If you need to process a batch, queue them manually (start the next one when the current export finishes), or use the Unmix mobile app which is built for a full library workflow.

Explore more tools