Extract vocals or instruments from tracks with AI!
I just came across a really interesting AI application that allows people to upload audio tracks and choose certain elements from them, like vocals, percussion, etc. I just tried it with Getz/Gilberto’s "Girl from Ipanema" and it works surprisingly well! There was one snippet of Joao’s voice in the first free minute preview that was not extracted, but lasts only a split second. It was compelling enough for me to opt in for one of the paid packages. For now, they’re offering 300 minutes (5 hours) of extraction for $25, and you can upload a variety of lossless formats and also download in lossless.
I have no affiliation with this service but thought the audiophile community might find it interesting. Enjoy!
As much as AI scares me, it also intrigues me with the innovations it affords.
What tracks have you tried, and what are your thoughts?
Hi, Bliss Thank you for posting this! Sent me down a most helpful set of rabbit holes. Turns out (of course) that there are other alternatives. A kwik google surfaces, among other links: Top LALAL AI Alternatives for Audio Extraction in 2024 (rushtechhub.com) Then one might ask: What sources might one use as the root files for the extractions? Easier peasier, I suppose, if you do your own recordings or have a server full of music. In my case, it will be more a case of downloading from streaming sites, in particular Tidal, which has its own set of indirect protocols, time sinks and learning curves. Useful in my case for vocals and, one hopes, better than Karaoke. I will follow up with my voice teacher as this process may be well understood and established in that world. Thanks again, Ken |
For an example of what is possible please see my older post from March of last year: https://forum.audiogon.com/discussions/eva-cassidy-remastered about Eve Cassidy digitally paired with the LSO./\ |
