Exhuming McCartney
The future of AI "music archeology."
We love a good portmanteau around here — strativation, bloghaus, Dime Square — but we’re struggling to find the right combination of music and archeology (marsicology?) to pithily describe the emerging opportunities for abandoned music to be dusted off via AI.
The prime example is obviously the “new” Beatles song that, in case you missed it, involved using machine learning to isolate John Lennon’s voice from a poorly recorded 1970s demo, then adding parts recorded by George and Ringo during the 80s sessions when the first “last” Beatles songs, “Free as a Bird” and “Real Love,” were brought back to life, and finally Sir Paul adding his own music and arrangement this summer.
This multi-decade collaboration is no doubt a technological feat. One that was unimaginable a year ago. And it created much excitement for what other discarded gems might be out there.
Being natural contrarians (as well as portmanteauologist), our gut immediately worries that we’ll be drowned in an endless stream of discarded demos brought back to life. In truth, this seems inevitable as older music dominates more and more of our listening time. The increase has been happening even without the help of AI reanimation. In a time of endless new music, we’ve ended up in an era of re-releases, re-recordings, re-remixes and whatever the heck is happening on TikTok.
The evidence is everywhere. In addition to the Beatles (presumably) one-and-done career coda (really a coda to a coda), we are now deep into the "Taylor’s Versions” gambit, which is paying off more than anyone could have imagined. 1989 dropped last week and broke streaming records despite being a nearly note-for-note duplicate of the beloved 2014 original.
Of course, when you’re talking about Swift and The Beatles, you’re dealing with two of the highest-selling artists of ALL TIME. Not to mention two who have remained highly relevant in recent years via a Peter Jackson documentary and record-breaking world tour, respectively. In other words, audiences are primed for curios from these two acts.
But there’s also room for niche artist to enter the redux world — prime example, The Replacements. The band’s classic album Tim got a new box set treatment last month centered around a new mix by legendary producer/engineer Ed Stasium. The justification for this revisionist history was that Stasium was initially meant to engineer the album but backed out due to scheduling. And the notoriously self-defeating Mats had wanted Tim to sound more like the new version in the first place.
Not that justification is needed. Music has always morphed in ways that weren’t obvious when the originators put it to tape. The Beatles were never overly precious about their work, even as it evolved from pop product to artistic statements. How else can we explain them signing off on the Bee Gee’s Sgt. Pepper movie, which looks like it was conjured up by hallucinating AI despite being created by real humans in 1978.
So back to our original point about AI being used to exhume unreleased recordings long left for dead. Will there come a time when Taylor Swift and Travis Kelce (T. Swells?) step away from the spotlight, leaving AI to fill the gaps by finishing up old demos? Are we in for versions of later Replacements albums with new guitar solos by Bob Stinson, who left the band mid-career and left this mortal coil in 1995?
The absolute biggest music news of the week — for The Cadence at least — was the revelation that there exists a 1983 studio recording of a five-piece Minor Threat redoing two songs, “In My Eyes” and “Filler,” off their first 1981 single that was tracked when the band was a four-piece combo.
Dischord label owner and Minor Threat singer Ian MacKaye is the last person we can imagine using AI on his old songs. But if the famously dogmatic figurehead of the 80s and 90s underground scene is starting to crack open the vault, that is a good benchmark for where the bar is set when it comes to disentombing old tunes for new listens.
TAKEAWAYS
Salient statements from this week’s music news.
1. They Said the Album Was Dying. They Were Wrong
For years, executives have claimed the music industry is a singles economy. But it's hard to find an artist who is having sustained impact solely with singles.
Takeaway: Much of the industry bent itself out of shape trying to get singles to pop on TikTok because it seemed like a cheat code — overnight virality as a substitute for the yearslong, painstaking work of building a fan base. But after all those contortions, many of the same old principles still apply.
2. AI Music Pioneer Quits After Disagreement Over 'Fair Use'
Vice president of audio at Stability AI, Ed Newton-Rex, has resigned due to his belief that training generative AI models using copyrighted content doesn't qualify as "fair use.
Takeaway: "I disagree because one of the factors affecting whether the act of copying is fair use, according to Congress, is 'the effect of the use upon the potential market for or value of the copyrighted work,'" Newton-Rex stated. "Today's generative AI models can clearly be used to create works that compete with the copyrighted works they are trained on. So I don't see how using copyrighted works to train generative AI models of this nature can be considered fair use."
3. Spotify to Charge $10 Per Song for Fake Streams
Under the new rules, Spotify will charge when more than 90% of streams on a song are fake. Staying under the 90% threshold will be easy for major artists with significant streams.
Takeaway: Large self-service D.I.Y. distributors like DistroKid, TuneCore, and CD Baby will likely be the most affected since the sheer volume of their unvetted uploads will be difficult to police. With entry-level distribution often priced at $10 or less, these fines could prove problematic.
