How to Extract Vocals from Any Song Free (AI Acapella Maker)

Drop a song into ClipCraft's Vocal Extractor and about a minute later you have the vocal by itself, with the instruments and backing track stripped out by AI. It runs in a browser tab, the free plan covers your first run, and it accepts video files too. This post walks through the whole thing: the steps, the real cost in plain numbers, and the places where AI separation still gets weird.

What vocal extraction actually does

Under the hood the tool runs Demucs, a music source separation model that listens to a finished mix and pulls it apart into stems. We run it on a GPU, so a typical song processes in well under a minute once the model is warm. You get two output choices: Vocals only, which is the clean acapella, or Vocals + faint backing, which keeps a little of the instrumental underneath for context.

It takes audio in most common formats (MP3, WAV, FLAC, M4A) and it takes video (MP4, MOV). Give it a video and you get the cleaned vocals back as an audio file; putting that audio back into the video is a quick job for a video editor, and it means the tool works on footage you filmed over background music you no longer want.

Step by step

1. Create a free account. No card needed; the free plan includes enough tokens for an extraction every month.
2. Open Tools → Vocal Extractor.
3. Drop your song (or video) onto the big drop zone, or click to pick a file.
4. Check the file card. It shows the length and the exact token cost up front.
5. Pick your output mode and hit the button. Then wait a bit.

Here's a 2:47 track loaded and ready. The card reads 30 tokens, because pricing is 10 tokens per started minute: a 2:47 song counts as three minutes. You see the number before anything is charged, and if a job fails to start the tokens come back automatically.

What it costs, in actual numbers

The free plan gives you 100 tokens and one tool run per month, which covers one song. The Saver plan at $1.99/month raises that to 300 tokens and five tool runs, and the tiers go up from there. If you only need a burst of extractions, token packs start at $0.99 and never expire.

One limit worth knowing: a single run caps at 3 minutes of audio on every plan except Studio. For a longer track, trim it first or split it in two.

What to expect from the quality

On a modern pop or hip-hop track with an up-front vocal, the acapella comes out surprisingly clean. Where it struggles is the same place every separation model struggles: heavy reverb smears into the instruments and comes back with artifacts, thick stacked harmonies can lose their edges, and a noisy low-bitrate rip gives the AI less to work with. Feed it the best copy of the song you have. A WAV or FLAC beats a 128 kbps MP3 every time, and the difference shows up in the stem. If you're weighing this against other separators, the vocal remover comparison lines them up side by side.

What to do with the vocal once you have it

• Open it in SoundCraft, our free browser DAW, and practice mixing a real vocal: EQ, compression, reverb, even Auto-Tune.
• Chop it into a sampler and build something new around a phrase you love.
• Solo the vocal to study how a singer phrases against the beat.
• Lay it over your own instrumental, then post the result to the ClipCraft music feed.

Is extracting vocals legal?

For private use, like practice, study, or messing around at home, you're generally fine. Publishing or monetizing someone else's vocal is a different story: that needs permission from whoever owns the recording. Your own songs are yours to split however you want, and that includes tracks you made with AI tools like Suno. When in doubt about a release, ask the rights holder, not a blog post.

That's the whole workflow. One upload, one button, and the vocal is yours to work with. The free plan covers your first song, so test it on something you know well and judge the stem with your own ears.