Separating a lead vocal from its harmonies takes two passes, not one. The first pass splits the song into vocals and instrumental. The second takes just that vocals file and splits it into the main vocal and the backing vocals. A normal vocal remover cannot do it, because it has a single vocals stem and a harmony stack lives inside it.
Drop a song in and get three files back: the lead vocal on its own, the backing vocals on their own, and the instrumental. Nothing to install, and the first ten minutes are free.
Stem separator
Split a song into vocals and instrumental
Start from a song, come away with the voice on its own and the instrumental on its own.
Previous runs
Noise remover
Take hiss, hum and room noise off a recording
Hiss, hum and room noise come off. What was said or sung stays exactly as it was.
Previous runs
Reverb remover
Take room and echo off a vocal you want dry
The room's wash comes off and the recording comes back dry, ready to place anywhere.
Most tools that advertise “vocal removal” return two files: vocals and instrumental. That is a genuinely different job from what most people actually want when they say “remove the backing vocals”. Doubles, harmonies, ad-libs and the odd shouted word are all singing, so they all end up in the same file as the main vocal.
Getting the main vocal on its own needs a second, different tool. The first one knows singing from not-singing. The second knows the main vocal from the harmonies — and it only works properly if you give it the vocals on their own. Hand it the whole song and there is far too much going on for it to make sense of.
Pass
Input
What comes out
One
The whole song
Vocals · Instrumental
Two
Just the vocals file
Main vocal · Backing vocals
What you get back
Lead vocal. The main line, with the harmonies taken out from underneath it.
Backing vocals. The harmonies, doubles and ad-libs, on their own.
Instrumental. The song with every voice removed.
Each is a separate download, and taking all three costs no more than taking one — a job is charged once, by the length of the audio you put in.
What people use it for
AI covers. Convert only the lead and keep the original harmonies underneath. This is the big one, and it has its own guide.
Karaoke and backing tracks that keep the harmonies but drop the main vocal, which a plain vocal remover cannot give you.
Training audio. A voice model should learn one voice. Leave harmonies in and it learns those too, not just the singer.
Study and practice. Hearing a harmony arrangement on its own is the fastest way to learn it.
Worth knowing: this is never perfect. Thick piles of harmonies, heavy effects, and harmonies sung very close to the main line are the hard ones, and you may hear a little of each file leaking into the other. Have a listen before you build on the result.
Cleaning up afterwards
Two more tools sit beside the separator, each with its own free allowance:
Noise removal — for hiss, room tone and hum, mainly when you are preparing training audio.
Reverb removal — for the room and the tail, which matter more than most people expect once a voice model is involved.
Common questions
Can you separate lead vocals from backing vocals?
Yes, with the two-pass approach above. One pass gives you vocals and instrumental; the second splits the vocals into lead and harmonies.
What is the difference from a vocal remover?
A vocal remover stops at vocals and instrumental. Every harmony is still inside the vocals file. Lead and backing separation splits that file again.
What audio should I upload?
The finished song is fine — MP3, WAV, FLAC, M4A or OGG. You do not need to separate anything yourself first.
Is it free?
Ten free minutes on the separator, and a separate ten minutes each on the noise remover and reverb remover. They do not share a pool. A subscription removes the limits.
Try it on a song
Lead, backing and instrumental, in your browser. The first ten minutes are free.