Acapella Extractor

How to extract an acapella from a song

1. Upload the song. MP3, WAV, M4A or FLAC, up to 10 minutes. It becomes the left reel.
2. Press "Extract the vocals". The price shows on the key before it runs. A song usually takes one to three minutes, and longer when the queue is busy.
3. Check the Vocals reel first. Click it to hear the voice alone, then download it as an MP3 at 320 kbps. The Instrumental reel is the leftover music, saved as a second file.

Who needs an isolated vocal

• Remixers and bootleggers want a clean vocal to drop on a new beat or a different key.
• Producers pull a phrase or a single word for a sample.
• Singers and songwriters listen to a vocal alone to hear the breath, the timing and the exact notes, then learn the melody by ear.
• Language learners and lyric writers find that words are far easier to catch with the music out of the way.
• Video editors put a vocal over new footage or a different backing.

Vocal isolator, extractor, separator: same job

You will see this called a vocal isolator, a vocal extractor, or a way to separate vocals from music. They all describe the same thing: one recording goes in, and a voice-only file comes out. Here the work is done by BS Roformer, an AI model trained to tell voices from instruments, so it does not rely on the old stereo-cancelling trick that only worked when the voice was panned dead center.

How clean will the acapella be?

A vocal pulled from a finished song is never quite the same as the original studio take. Keep these in mind before you build a remix on it.

• Reverb and echo that were baked into the mix come with the voice, and some of the instrument tail may too.
• Harmonies and ad-libs are all kept, since everything sung counts as vocals. If you want the lead only, use Separate Lead and Backing Vocals.
• Dense mixes and live recordings leave more bleed, such as a hi-hat tick or a guitar hum under the quiet words.
• Doubled and stacked vocals can sound slightly phasey, because the model has to separate several similar voices at once.

For a remix, listen to the whole file with headphones before you commit. Quiet sections and the ends of phrases are where leftovers show up. If the room sound is too much, run the vocals file through the Reverb Remover.

Acapella maker or instrumental maker?

The same upload gives you both files, so you never need to run the song twice. The vocal comes first here because that is what most people come for. If the instrumental is the main thing you want, the Instrumental Maker is set up around it, and the Vocal Remover covers the general case.

Tips for getting a usable acapella

Start from the best file you have. A WAV or a high-bitrate MP3 gives the model more detail than a low-quality rip, and the vocal inherits that quality. Songs with a dry, upfront voice, such as many pop, R&B and rap tracks, separate best. Older recordings with the voice and band recorded together in one room are harder, and the vocal may carry some of the room.

If you plan to change the tempo or key for a remix, do it after you download, on the vocals file alone.

Turn the vocal into notes

Once you have a clean voice, you can convert the melody to MIDI with the audio to MIDI converter. A vocal on its own gives much more usable notes than a whole mixed song.

Your files

The page works in your web browser, but the splitting does not run on your computer: your song is uploaded to our storage and split in the cloud. The song and its parts are saved in your account for 30 days, then deleted, and removing a song deletes its files straight away. We don't use your songs to train AI models. Only upload music you have the right to use.