Skip to main content

I just discovered Emeli Sante. I had heard the name but her music had passed me by. On the radio I heard a track, and Shazammed it. I think it will be a great one for a friend of mine to sing - so I jumped into an arrangement of it - perhaps with the intention of doing it live in a theatre idea we have had on the simmer since pre-covid . Did the demo - just piano. Then because I like it, added double bass and then got a bit carried away with a string quartet. Trouble is, my friend always likes a guide vocal and a fella doesn't work - so I wondered if SpectraLayers would do a good job of taking the vocal off the real song, and then I popped it into mine. Slightly different tempo and feel. I think it's done a pretty decent job, and Cubase did the chopping up and stretching and squeezing. I didn't put it in the mp3 area because it's an experiment really - but if I hadn't told you it was an extraction from the released recording, would you have known? 

Audio
Play

Comments

paulears Sun, 08/01/2021 - 23:28

What I tried was to produce my version of the emeli sante track and get my singer to learn it so she can sing it live. So our usual process is to record everything ground up. I recorded the piano, double bass and string quartet then had the problem that it’s just not a fellas voice and I couldn’t record me. I tried but it didn’t work even as a rough quite. 
 

That’s when I wondered if spectral layers would be able to remove everything apart from the vocal line from the commercial track? It did a great job so I just then needed to stretch, shrink and chop it up. If you look on YouTube there are lots of isolated tracks from popular songs. Some are studio tracks other are where people have tried to isolate the vocals in software and lots sound very odd. Spectral layers did a really decent job. Isolating stems from a finished track is very difficult to do and it seems only a couple of products do it well, so far.