I thought some might find this interesting. I've been using SpectraLayers for a while now, and I thought it might be fun to see if you could strip out the vocals from well known songs, and then do a totally new version of the song, and put back the original vocal. I can't stick this on youtube for obvious reasons, so it's sitting on my server at the moment.
I thought I'd do a lounge jazzy style for a Lionel Ritchie track, and I did an Elton and Celine Dion track too. I tried Peter Gabriel's Sledgehammer - but his breathy style produced loads of artifacts. There are some timing issues with fitting their vocals in, but with more work it could be sorted.
See what you think - in the busier tracks the vocals sit quite well. When it's quiet and exposed, it stands out I think - and in some the backing vocals were 'glued' to the main vocal and I couldn't separate them.
Despite the artifacts, this…
Despite the artifacts, this is impressive. Well done.
How long does it take to strip out a vocal? Can you do it in one step and do you have to keep zooming in etc?