Remove vocals from any song and keep the instrumental, or pull out a clean acapella. Free in your browser, or fast on our servers.

Press play, then solo a track Upload on the left to separate yours
A vocal remover splits a finished song into two files: the instrumental, with the singing taken out, and the vocals on their own (an acapella). This one is an AI vocal remover: a separation model listens to the whole mix, so it keeps the bass, drums and stereo width that older tricks lose.
Both songs were made on MakeASong and separated by the Free mode on this page. Move the faders to build your own mix: drop the vocals for a karaoke track, or drop the music to hear the acapella.
Free mode on an M1 laptop: 43 s for this 2:43 song, nothing uploaded.
Audio editors such as Audacity remove vocals by subtracting the left channel from the right. Anything mixed in the middle cancels out, and that includes the kick, the bass and the lead voice. Listen to the same chorus of Read My Light three ways; the bars are measured levels against the original.
The mix as released.
Mono and thin. The kick and bass go with the voice.
Stereo kept, drums and bass intact.
Take the voice out, keep a 15% guide vocal if you want help with the melody, then add synced lyrics.
Pull the vocal on its own for a remix, a mashup or a study of the phrasing. Reverb stays with the voice.
Play or sing over the real backing track instead of a MIDI version, then hear it in a new style.
Use the instrumental under a voice-over or a reel so the lyrics don't fight the speaker.
| Freein your browser | Faston our servers | |
|---|---|---|
| Price | Free, no account | 8 credits per minute |
| 3-minute song | About 45 s* | About 20 s |
| First use | 180 MB model, once | Nothing to install |
| Devices | Desktop Chrome, Edge | Any, phones too |
| Your file | Never leaves your computer | Uploaded, deleted after 7 days |
| Result | Download it on the page | Kept in your account |
* Measured on an M1 Pro laptop in Chrome: 43-48 s for songs of 2:43 to 3:02. Older or lower-power computers take longer.
Checking your browser…
Drop an MP3, WAV or M4A up to 10 minutes, or pick one you made on MakeASong.
Free runs on your computer at no cost. Fast runs on our servers and works on phones.
Solo each track, compare with the original, then save the instrumental and the vocals.
Mostly, yes. They are different names for the same split. What changes is which half you keep.
Another name for a vocal remover. You keep the instrumental.
Same split, but you keep the vocals. Also called a voice isolator or vocal extractor.
Same split, focused on a clean vocal for remixes and samples.
The instrumental, often with a quiet guide vocal and lyrics on screen.
Splits a song into four or more parts: vocals, drums, bass and other instruments.
Either a vocal remover, or a generator that writes new music. We have both.
| Option | Cost | Quality | Good when |
|---|---|---|---|
| MakeASong · Free | Free, no account | AI separation, full stereo | You're on a desktop and want privacy |
| MakeASong · Fast | 8 credits/min | AI separation, full stereo | You're on a phone or in a hurry |
| Audacity “Vocal Reduction” | Free app | Phase trick, loses bass | The vocal is dead center and dry |
| Paid separation apps | Subscription | AI, more stem types | You need guitar or piano stems often |
| Searching “karaoke version” | Free | Often re-recorded | A popular song already has one (it usually wins) |
Yes. Free mode runs the AI model inside your desktop browser, so it costs us nothing and has no daily limit and no sign-up. Fast mode runs on our servers and costs 8 credits per minute of audio, rounded up; new accounts get 80 free credits.
That is the separation model itself. Your browser keeps it, so the next song starts in a few seconds instead of waiting for the download again. Clearing your browser data removes it.
Yes, with Fast mode. Free mode needs a desktop graphics chip that phone browsers don't let web pages use yet, so on phones and tablets the page offers Fast only.
No. Upload an audio file you have the right to use. Audio ripped from video sites is also low bitrate, which leaves more traces of the voice in the instrumental.
On a studio mix with one lead voice, the voice is gone and the drums, bass and stereo width stay. Heavy reverb, a crowd singing along and stacked harmonies can leave faint traces you hear when the music is quiet.
Yes. Remove the vocals, turn on Keep a guide vocal if you want a quiet hint of the melody, then open the song in the Lyric Video Maker to put synced lyrics on screen.
Yes, and usually very cleanly, because AI songs are dry studio mixes. Songs you made on MakeASong appear under My songs, so you don't need to download and upload them.
MP3 (192 kbps) for listening, sharing and practice. WAV (16-bit, 44.1 kHz) if you will edit, remix or master the result, because every new MP3 encode loses a little more.
Not in Free mode: the file is read and separated on your computer and nothing is sent anywhere. Fast mode uploads it to our servers; uploads are deleted after seven days and the vocals stay in your account so you can open the result again.
Removing vocals doesn't change who owns the song. The separated files are for personal use such as practice, karaoke and study. To publish anything made from a song, you need the rights to the original recording.
Free in a desktop browser. Fast on any device.