Kevin Davis wrote:tragabigzanda wrote:Yeah, I’d fundamentally disagree on that last point. For a band that made their career on amazing live shows fronted by a talented singer, the hard lean into Autotune to cover the effects of that singer’s aging — rather than writing and performing songs more suited to his current vocal limitations (or at the very least, publishing live material WITHOUT the use of Autotune) — is decidedly on the inauthentic end of the spectrum.
Hard agree. Autotune came up in one of the Dark Matter song threads earlier this year, where it was theorized that Eddie was using (or could someday use, I don't remember) it to keep his voice from going off-key in concert. This is what I had to say about it then:
I don’t oppose Autotune on principle, but I would not want to see PJ’s live shows move the direction of artificially “fixing” their performances in real time. I expect some sweetening on a studio record, but in concert, the fact that the band is onstage playing with live ammo is part of what makes the performances exciting — knowing that they are humans working within the realm of human limitations is what makes it feel amazing when they occasionally do things that seem superhuman. To me having an autotune pedal (or whatever) on standby would feel a bit like an MLB hitter using steroids — sure, it “improves” performance, but it would strip the thing I am watching of some of what excites me about it to begin with.
I would argue that the same is true tenfold of AI. So much of what makes art impressive, expressive, and emotionally resonant hinges upon how human beings navigate human limits. I realize that technology is one of the ways we do that, and that some artists more seamlessly incorporate that technology into their work than others. But Pearl Jam have never been that band, and I do not think anything about their work will be well-served by moving further into that space.
Sorry for the delayed response.
There are a few different dimensions here, and I disagree less on some than others. I also thought Trag's post was about studio autotuning - that because of their live reputation it made using it in the studio worse. I think I misread that, so apologies.
In terms of live performances, I hear you. The idea of a bootleg is to capture the live experience, or at least the sound, and so at that point excessive post audio correcting/sweetening does undermine the authenticity that is presumably an important element of this. However, if what you want is something that 'sounds' live and you aren't concerned about the historically accuracy of the document this would bother me less. If I am listening to a show I was at I want it to sound like that show. But live Pearl Jam sounds different than studio and if what I want is to maximize how much I enjoy listening to those live songs as a sort of remix/alternate version it bothers me less. But I get why this would be hard for some folks.
If there was some AI modulation on some of the Howard Stern performances that feels like an in between space. It's cooler in some ways if it's not there, but I understand wanting something you are sharing for promotion purposes to sound as good as possible. I am a lot more likely to revisit a performance that sounds good than one that was maybe a little rougher but true to that exact moment.
Studio is, I think, a fairly different beast. There is a reason that I just couldn't use AI filters to reproduce the experience of a pearl jam song. There is an ineffable quality that any good artist would bring to their work that can't really be captured. Even with autotune. Either the song works as a song or it doesn't. You're captured by it or you aren't. If you had that experience listening to the music that was real. Discovering after the fact that it was an enhanced performance using studio tools shouldn't make too much of a different. Autotune, stitching together best takes, overdubs, effects pedals - all of these things are part of the technologically assisted construction of a song.
It's possible that pearl jam might not use those tools effectively, to your point. But that's not really about the tools then, it's about how well you utilize them. And if it was used on an album like Dark Matter, it was used to incredible effect as these songs are immensely engaging both as craft and in the execution of an incredible listening experience. I am swept away by them, and don't much care how I got there.
Now if you were going to use a heavily autotuned/sweetened performance of an example of, say Eddie at the peak of his powers that is a little disingenuous. It can still be an incredible performance in total (and there are a few on Dark Matter), but if I had to highlight it as the most authentic vocal performance he's ever delivered fully under the power of his own instrument I probably wouldn't choose this.
Then again, I think Barry Bonds should be in the hall of fame, and see steroids and eyeglasses on a continuum - in both cases the athlete is not just working with their native talents. The difference is that steroids are harmful.
Kevin Davis wrote:stip wrote:I don't worry about the authenticity of it as much as the impact it has on the ecosystem of creative work. I don't like that this technology will make it that much harder for artists to make a living when bland mediocre AI generated slop will do. Artists incorporating AI into their work are just utilizing a new tool
Here's a hypothetical for you: On one of the Dylan boards I visit, someone recently used AI to "complete" a famously unfinished outtake. This song ("She's Your Lover Now") has several takes: a solo piano demo that is complete, and several full-band takes that are incomplete (the song just never jelled in the studio, and Dylan abandoned it). The forum user took the isolated vocal track from the finished piano demo, and through some combination of meticulous manual editing and AI fakery, "completed" the unfinished studio take. People on the board are losing their minds about it: "It's the lost song we've been waiting for all these years!" One fan suggested that Sony hire this message board user to apply similar AI trickery to other unfinished Dylan works and include them on official archival releases.
Would you want Pearl Jam to do this?
Say you're in charge of producing Lost Dogs II, and on the tracklist are songs like "Anything In Between," "Puzzles and Games," etc. You have two choices: You can include the unfinished versions as they exist, or you can feed them into an AI algorithm that will spit them out on the other side as "finished"-sounding songs. The latter will sound "better," and will be more palatable for release, but will be comprised approximately 25% of sounds that were never made by the band and were generated entirely by the computer program. What do you put on the record: the as-is "real" versions, or the finished AI versions?
I would be interested in the 'finished' versions as a curio, and may even really enjoy them, but I would choose the 'real' versions, I believe. But autotune again feels different to me than machine generated songwriting and performance in the style of person X. That's what scares me, and I where I worry about the impact it will have on art broadly - when it moves from being an assistive tool to one that replaces the artist.