Free workflow · new

Vocal Split — a voice sample from any song

Pull the lead vocal out of a finished song.

Demucs separates the vocal and keeps the stretch with the most singing, ready to use as a Voice Clone sample. Liked the singer on a song you generated? Keep that voice for your next tracks.

Model: Voice
VRAM: 8 GB+
Level: Beginner

Download workflow — free

Drag the .json onto ComfyUI. Runs best on the Gmanski RunPod template.

You get

Vocal Split workflow

You need

The Gmanski RunPod template (Nov 2026 image) or the voice engines installed once
8 GB VRAM

Pairs well with.

ACE-Stepnew

Song from an idea — ACE-Step

A second song engine. Apache 2.0, 8 steps.

Beginner · 16 GB+Free

Voicenew

Voice Clone — speak in your voice

Ten seconds of your voice. Any text, 23 languages.

Beginner · 12 GB+Free

Voicenew

Voice Clone — any song in your voice

Same song, same backing track, your voice singing it.

Beginner · 16 GB+Free