Remove the music. Keep the voice.

AI voice extraction. Upload any audio and download the isolated vocals with the background music stripped away. Free, no account, no watermark.

Powered by Demucs (htdemucs) on dedicated GPUs · files auto-delete in 2 hours
★ Pro active — unlimited files
Drop your audio file here or click to browse
MP3, WAV, FLAC, M4A, OGG · max 100 MB · max 10 min · 6 files/day free
Uploading…
ad — rectangle (shown while processing)

What people use it for

🎙

Vocals-only nasheeds

Turn a nasheed with instruments into a voice-only version you're comfortable listening to.

📖

Lectures & talks

Strip distracting background music or noise from recorded talks, khutbahs, lessons and interviews.

🎧

Recordings & podcasts

Rescue a voice recording that was made over music — isolate the speech and keep what matters.

How it works

1

Upload your audio

Any common format. Your file goes straight to a GPU queue — no signup, no email.

2

AI isolates the voice

Demucs, the model that tops academic source-separation benchmarks, lifts the human voice out of everything else.

3

Download the result

A clean voice-only file, ready in under a minute for most uploads.

Pricing

Free

$0 forever
  • 6 files per day
  • Files up to 10 minutes / 100 MB
  • Full-quality 192 kbps MP3 output
  • No account, no watermark
Start free

Pro

$5/month
  • Unlimited files, priority queue
  • Files up to 40 minutes / 500 MB
  • Ad-free experience
  • Cancel anytime

API

$0.04/minute
  • REST API for developers
  • Prepaid credits, never expire
  • Volume packs from $0.03/min
  • Failed jobs auto-refunded
API docs →
ad — leaderboard

From the blog

FAQ

How does it work?

We run Demucs (htdemucs), a state-of-the-art open-source source separation model, on dedicated GPU servers. It isolates the human voice from music and background sound with studio-grade quality.

Is it really free?

Yes — the free tier is supported by the ads on this page and allows 6 files per day. Pro removes the cap and the ads for $5/month.

Will it work on nasheeds?

Yes. Nasheeds with a clear lead voice separate very well. Heavy reverb or layered backing voices can leave faint traces, but the instruments themselves are removed cleanly.

What happens to my files?

Files are processed automatically and deleted from our servers within 2 hours. Nothing is shared, published, or used for training.

Why is there a daily limit?

GPU time costs real money. The free cap keeps the service fast and free for everyone; Pro users skip it entirely.