I’ve been using the same song on this channel for ages to demonstrate mixing techniques, and there’s one sneaky thing in the vocal stack that basically nobody ever noticed: all those different “people” you hear in the harmonies… were actually sung by one person. The problem is, when you stack your own voice for harmonies, it can get that weird, samey character where even an untrained ear can tell it’s just one singer at different pitches—and the illusion of a real vocal group falls apart. In this video I show you my exact workflow for fixing that using Sonarworks SoundID VoiceAI. I start by building stereo unison vocals from an unused lead take (make sure your track is set to stereo), capture and analyse once, then swap voices without re-capturing. I create two-voice unisons, deal with phasey-ness by adding timing variation (my sweet spot is around 30–40), and then I add a lower octave male layer for depth (VoiceAI even auto-suggested -12 transpose in my case). From there I repeat the process across multiple harmonies, ending up with a 12-voice stack that sounds like a bunch of different singers—wider, deeper, and far less “same person” obvious in the mix. I also show how I use it for vocal stand-ins: I’ll record a quick guide vocal with my voice and replace it with a female voice so I can build the arrangement before Suz comes in to record the real thing. You still have to write and sing the parts—this just handles the voice replacement, and it does it surprisingly well.