How to extract an acapella from a song
Isolate clean vocals with AI stem separation, tidy the result and export it for remixes and mashups.
Last updated · 6 min read
An acapella is the vocal of a song on its own, without instruments. Producers use acapellas for remixes and mashups, singers use them to study phrasing, and DJs layer them over other tracks. Official acapellas are rare. Here is how to extract one yourself with AI in MixSongs, right in your browser.
Extracting a vocal from a commercial recording does not give you any rights to it. Remixes and mashups you publish usually need permission from the rights holders. Use audio you own or are licensed to use. See our Terms.
Step 1: Start from the best source you have
Separation quality depends on the input. In order of preference: the original WAV or FLAC master, then a 320 kbps MP3 or 256 kbps AAC, then anything lower. Avoid files ripped from videos with heavy compression; artefacts in the source become artefacts in the acapella.
Step 2: Separate the song
Drag the song into MixSongs and click AI on its track. The vocal model is Mel-Band RoFormer, which handles breaths, reverb tails and harmonies far better than older tools. After about a minute you get four stems: vocals, drums, bass, music. Separation needs an account and one credit per song (15 songs for $2.99). If it fails, the credit is refunded automatically.
Step 3: Solo the vocals
Click S (solo) on the vocals stem and listen through the whole song. What you’re hearing is your acapella.
Step 4: Clean it up
A few quick edits make an acapella much more usable:
- Remove the empty parts. Long instrumental sections leave near-silence with faint artefacts. Split them out with S at the playhead and delete them, or draw the vocals’ volume line down to zero there.
- Tame leftover instrument bleed. If a loud snare peeks through during a quiet line, add two volume checkpoints around it and dip the level briefly.
- Fade the edges. Drag the fade handles at the start and end of each phrase so nothing starts with a click.
Step 5: Export the acapella
With the vocals soloed, click Export and choose WAV 24-bit or FLAC 24-bit. These keep every detail for further processing in your DAW. Solo state is respected, so only the vocals are rendered.
Tip: to export the instrumental of the same song afterwards, un-solo the vocals and mute them instead. See how to make a karaoke version.
Using the acapella in a mashup
You can build the mashup right in MixSongs: add the instrumental of another song as a second track, drag the acapella to line up its first downbeat with the instrumental’s, and set the BPM grid to check that the phrases line up. MixSongs doesn’t change tempo or key, so pick two songs with similar tempo and a compatible key for the best result.
Why AI acapellas sound better today
Older acapella extractors relied on phase cancellation and needed an instrumental version of the exact same mix. Modern separation models learn the sound of a voice from huge amounts of training data, so they work from the full song alone and keep the reverb and breaths that make a vocal sound natural.