Strange New World of AI Vocalists!

My son Josh has been showing me the amazing and scary impact that AI tools are having on the recording industry. We’ve known for a while that we can’t really trust images that we see any more, because they can so easily be faked and doctored. Well, now we can’t trust our ears either.

There are now a wide variety of AI tools available online that can extract the vocal line from any recording, and then substitue another voice for the orignal. There are even vast libraries of real-sounding voices to choose from that people have made, adjusting the vocal characteristics to be ‘ideal’ for certain kinds of genres, styles and songs. And you can ‘train’ the AI model to imitate a real voice by upload samples of their singing. Consequently there are fake versions of a whole legion of famous singers,

Josh used it to make a synthetic version of me, based on over a dozen songs I recorded over the years, and then he used that synthtic voice to dub me onto an old studio demo sung by someone else of a song I wrote many years ago, but somehow never recorded here at home. He also used it to make a new recording of a song that I wrote for a childrens record that sounded great, with a terrrific band, but I had written a new lyric for a ‘grownup’ version, which I never recorded. Try listening to I Wonder, below. Its not perfect, but it isn’t bad!

Then for fun, he used the tools to make versions of my songs sung by people who I wish had sung my work. Please enjoy Fake Sinatra singing Blue Notes. Sure, there’s some noise from digitizing the ancient late ‘50’s vinyl that we’ll have to clean up too, but I like it!