VEIRA Vocal Remover
What is VEIRA Vocal Remover for?
It splits a finished, mixed stereo song into the vocal and everything else (the music), and two faders, VOCAL and MUSIC, set how loud each comes back. Pull VOCAL down for a karaoke track or instrumental, pull MUSIC down for an a cappella, or set anything in between to rebalance a vocal in a finished mix. It is made for sung music; a spoken voice separates far less, and the Speech presets are built for that.
How do I make a karaoke track or an a cappella quickly?
Click REMOVE VOCAL (VOCAL Off, MUSIC 0 dB) for the music without the voice, or VOCAL ONLY (VOCAL 0 dB, MUSIC Off) for the voice alone; ORIGINAL puts both at 0 dB so you can compare. For a tuned starting point, load Karaoke ▸ Karaoke or A Cappella ▸ A Cappella; there are 45 factory presets in 8 folders, sorted by job.
Does it upload my audio or need the internet?
No. The AI model (Spleeter, by Deezer Research) is installed with the plugin and runs on your own computer, in real time. Like every VEIRA plugin, it only needs the internet through VEIRA Manager, to install and to refresh the licence.
I can still hear the singer faintly. How do I get rid of it?
Removal bottoms out about 21 dB down, so in a quiet, sparse passage a ghost can remain. Raise STRENGTH to 70–85 %, or load Karaoke Strong or Instrumental Max Removal. Vocal reverb, doubled vocals and harmonies are the hardest parts to remove.
What does STRENGTH do, and why does the music sound watery?
STRENGTH (0 to 100 %, default 50 %) sets how much of the sound counts as voice: higher removes more of the vocal but takes more of the music with it and can sound watery. If the music sounds watery, swirly or dull, lower it towards 50 % or below. From 35 to 70 % the overall quality hardly changes, so in that range it’s a matter of taste.
What are the VOCAL RANGE LOW and HIGH knobs for?
Outside the range they set, everything counts as music and is left exactly as it was. LOW keeps the kick and bass clean (up to about 90 Hz it costs the voice nothing), and HIGH at 10 to 14 kHz keeps the cymbals, hi-hats and air. You can also drag the LO and HI grips on the right edge of the SEPARATION display; double-click a grip to switch it off.
I turned VOCAL down 6 dB but it sounds like less. Why?
About 0.8 dB of a typical vocal is never separated, so the real change is a little smaller than the fader’s: −6 dB on the fader is about −5.3 dB. Move the fader further; the Vocal Balance presets already do this. When turning a vocal up there is no peak protection, so lower OUTPUT until the OUT peak stays under 0 dBFS.
Why is there so much latency, and why do I hear myself late when recording?
The AI needs to hear each sound in context, so the plugin holds the song back by about 1.18 seconds (52,224 samples at 44.1 kHz). Your DAW compensates on playback and in bounces, but not for a live signal. Put the plugin on the song’s track only and monitor your microphone on its own track, or bounce the karaoke track first and record against that.
The LATE lamp lights and the vocal leaks for a moment. What should I do?
Your computer didn’t finish the AI’s work in time. Watch AI LOAD: if it’s near or over 100 %, use fewer instances at once, close other heavy plugins, or bounce the track offline (a bounce never runs late). Apple silicon Macs need much less processing than Intel or AMD processors.