The original recordings sound much better and more interesting to me. Way better. The AI generated versions sound slicker in some sense but... like the re-recorded versions of old hit songs from the 60s you hear at the grocery store sometimes. Technically the song is still there, but it blends in with the rest of the muzak.
I'm sorry to be so negative, it's great you're returning to the material after all these years, but the AI versions I've listened to all have the same smoothed-over quality that loses everything interesting and relatable to my ears in the original versions.
I get it. Recorded music is a lot of fun to play and listen to. This still was a fun project for us to work on together after 25-years and with both of us living on opposite ends of the country. Hopefully our next release will be an in-person recording!
I gotta brag on Chuck a bit more though. I do love the depth to his lyrics and the way he puts them to melodies. I think his stuff, at the core, is excellent. Can't wait to work on some new stuff!
Interesting idea, and it sounds like you like the results, so kudos.
My musical taste tends toward less produced, rougher/lo-fi recordings, e.g. Guided By Voices. So, it follows that I prefer the original recordings.
There is one aspect that I find slightly dehumanizing -- the AI changed the singer's voice so drastically that I don't recognize the original singer anymore. The uniqueness of the voice has been smoothed out to be much more generic. For me, music (and art in general) is in part an expression of one's identity, so I'd consider this a negative outcome. I'm sure similar things could be said of the instrument performances as well.
But again, this is all just my two cents, and it's a matter of taste. Ultimately if you're happy with the results, that's what matters!
Completely agree. We tried feeding his voice in and using that as a voice model and it actually sounded even further away from his actual voice so we nixed it. We uploaded our original tracks as a reference, so there were some notes that left me feeling like "that sounds like Chuck", but for the most part, yeah, it's pretty far off.
it's pretty easy to make words that fast once you have the words written, fwiw. mostly breath control and there are things you think you could never do and you'll be doing it after a few times practicing. now writing the lyrics is a bit harder. twista is great
I think you nailed it here. The originality happened 25 years ago. That was fun! Then 25 years passed, life went on, and this project gave two old friends a reason to reconnect and see what the technology could do with our old songs. That was fun too!
Nice - I've done similar things with some of my music [1].
I have a classical piece I wrote over a decade ago for piano [2] (it’s the instrument I play), but it was always intended to be an orchestral work. Using AI allowed me to sonically experiment with a stringed score which was pretty cool.
It’s basically the equivalent of taking a piece you’ve written and running it through an arranger keyboard or Band-in-a-Box on steroids.
That's great. AI is a tricky beast. It can be used for good or evil. I had a lot of convictions while working on this project and my soul is resting easy with how we navigated things!
I think this is a super intriguing project! I’ve been experimenting with some similar work on my old unfinished songs from eons past. It’s been really fulfilling to take that old work and see the original vision spring to life through modern tools.
The web stuff is well done. Now, with this newly created momentum, you should re-record the songs now, without AI. Hell, you could probably afford an awesome studio and equipment too. You could easily blow the AI out of the water.
I'd love that. We hope to. We live on opposite sides of the country now. The AI stuff was a great, nostalgic kick-off for us. We don't plan on using AI like this as our only way of making music, but it was a great project to work on. We'd love to write some new stuff and play live again. Hopefully soon!
I've read the "how revival works" section, but still have no idea "how the revival works".
("We've used ai" is all I got from both this intro on HN and the we pages I read, though possible I missed some section.)
Can you share?
I.e. Did you take original audio recordings and run it though some audio chain that optimizes the mix and volumes? Did you put the sheets and lyrics into ableton and recreate the music? Did you feed audio files into chatgpt and prompt "make it better"? Something else?
In the interest of transparency, understanding what happened here will significantly guide my own emotional response :). I appreciate the details of 5 core principles, but spending so much time on principles without actual detail on what got done makes me skeptical and even cynical, which may not be the intent. For example, I personally distinguish between a raw photo, edited photo, composite image, and AI regenerated image, and one of the things I'm trying to understand is the path / traceability from human to final audio file.
The revived (AI) versions have this...thin and hollow sound to it. It is difficult to explain, most AI-generated songs have this when they're modelling acoustic drums, stringed instruments, etc.
FWIW, I'm (now a hobby) musician and have done studio work. Even the latest and best models have this unmistakable sound.
Totally agree. I guess that's what comes with the AI/robot stuff. Hopefully we'll be able to create an in-person album again, but with a little higher production quality than our dorm room versions.
We're about the same age, I was even in a band at a small liberal arts college in the great lakes area in 2003. AI can't bring it back, and the stuff the music AI has created here sounds terrible to me.
Have you considered processing the original recordings using AI? I've witnessed some truly amazing results. I've had twenty year old Skype call recordings sound like we were sitting in a recording studio.
I'm sorry to be so negative, it's great you're returning to the material after all these years, but the AI versions I've listened to all have the same smoothed-over quality that loses everything interesting and relatable to my ears in the original versions.
I gotta brag on Chuck a bit more though. I do love the depth to his lyrics and the way he puts them to melodies. I think his stuff, at the core, is excellent. Can't wait to work on some new stuff!
My musical taste tends toward less produced, rougher/lo-fi recordings, e.g. Guided By Voices. So, it follows that I prefer the original recordings.
There is one aspect that I find slightly dehumanizing -- the AI changed the singer's voice so drastically that I don't recognize the original singer anymore. The uniqueness of the voice has been smoothed out to be much more generic. For me, music (and art in general) is in part an expression of one's identity, so I'd consider this a negative outcome. I'm sure similar things could be said of the instrument performances as well.
But again, this is all just my two cents, and it's a matter of taste. Ultimately if you're happy with the results, that's what matters!
I want to be able to rap like Twista. If I use AI to change my voice and speed it up, it’s kinda fake.
Where’s the originality in that. I’ll never be *that good*, but I have fun doing it.
Now I guess using AI strictly for mastering is OK , but even then the results haven’t been good for me.
I think you nailed it here. The originality happened 25 years ago. That was fun! Then 25 years passed, life went on, and this project gave two old friends a reason to reconnect and see what the technology could do with our old songs. That was fun too!
I have a classical piece I wrote over a decade ago for piano [2] (it’s the instrument I play), but it was always intended to be an orchestral work. Using AI allowed me to sonically experiment with a stringed score which was pretty cool.
It’s basically the equivalent of taking a piece you’ve written and running it through an arranger keyboard or Band-in-a-Box on steroids.
[1] - https://mordenstar.com/blog/dutyfree-shop
[2] - https://mordenstar.com/blog/screwdriver-sonata
I've read the "how revival works" section, but still have no idea "how the revival works".
("We've used ai" is all I got from both this intro on HN and the we pages I read, though possible I missed some section.)
Can you share?
I.e. Did you take original audio recordings and run it though some audio chain that optimizes the mix and volumes? Did you put the sheets and lyrics into ableton and recreate the music? Did you feed audio files into chatgpt and prompt "make it better"? Something else?
In the interest of transparency, understanding what happened here will significantly guide my own emotional response :). I appreciate the details of 5 core principles, but spending so much time on principles without actual detail on what got done makes me skeptical and even cynical, which may not be the intent. For example, I personally distinguish between a raw photo, edited photo, composite image, and AI regenerated image, and one of the things I'm trying to understand is the path / traceability from human to final audio file.
Thx!
I might have to revisit some of my old songs...
FWIW, I'm (now a hobby) musician and have done studio work. Even the latest and best models have this unmistakable sound.