- AI For Humans - The Newsletter
- Posts
- Using AI To Remaster Your Greatest Hits đź‘‚
Using AI To Remaster Your Greatest Hits đź‘‚
And extremely impressive updates to the fast-moving world of generative talking head videos
Today on AI For Humans The Newsletter!
Slogging through today’s AI gray areas
Live performance-driven AI video just got better
And our favorite music generation model’s due for an upgrade 🎵
Plus, our can’t-miss AI feature of the week!
Welcome back to the AI For Humans Newsletter!
Remember Suno? That AI music generation app we’ve raved about on AI for Humans? Well, they’re back, and V4 is here to blow our minds… again. This latest version is a real sonic upgrade which turns a stale “Hot Dog City” into something far beefier — it’s like the Kirkland Signature of music remastering.
Expectations Don’t Ruin the Experience
So here’s the thing: you expect tech to get better; you assume it’s going to improve. But we got a preview of the soon-to-be-released Suno V4, and I’m giddy over what it does to sound quality. With V4, you can remaster your old tracks with a click. Which means you can take older songs and ideas you’ve created on the platform, and instantly breathe new (cleaner/clearer) life into them!
Songs that I created months back are suddenly so crisp that I’m picking out instruments I didn’t even know were there. Clarinets are finally doing their thing, drums snap, and vocals come through clean and clear.
🎧 Listen for Yourself: Just a Little Thing 🎧
To give you a taste of what Suno V4 can do, here’s a little experiment I ran with one of my old GarageBand files. I had this bluesy, country-western-ish riff—a quick melody I abandoned eons ago, complete with a basic drumbeat, some tambourine, and a hint of guitar twang.
I liked the vibe but left it in the “maybe someday” pile.
With just one pass through the remaster tool, Suno transformed it into a full-bodied, polished track. It brought out nuances that made it feel real, adding performative touches that I’d only dreamed about. Tremolos, guitar string bends and little accents were bringing the piece to “life”.
I took things a step further. I bounced out stems (separate tracks for each instrument), so I could isolate the drums, guitar, and vocals, rearrange them, and send it all back into Suno. And each time I did this, it generated something fresh and exciting; more like collaborating with a whole band than just remixing a track.
This process took minutes, not hours, and now the result is something I can see performing live. Which is, something I can’t wait to get back to my drums and accomplish with some very real, very human pals…
Feel free to loop this lil’ thang while you read the rest of the newsletter. Hope you dig it.
-Kevin & Gavin
3 Things To Know
Jazza Explores A Constructive Path Forward for Creatives
Wading into the AI debate is dangerous territory for creatives, but we doff our digital caps to Jazza for this thoughtful video. He does a nice job of acknowledging the inevitability of change, without steamrolling the real human impact it’s already having on professionals. Gavin shared a few of his own thoughts yesterday for the larger impact on Hollywood professionals, suggesting some changes may be overdue.
A New State Of The Art Model for Talking Heads?
TikTok parent company ByteDance is flexing a new X-Portrait 2 model, and initial impressions are it may be even better than Runway’s Act-One. It’s just a research paper (for now), but the previous X-Portrait 1 is available on GitHub for advanced computer users. It makes you wonder if social media platforms are in the driver’s seat to deliver the best models, given they have access to the best training data available.
Kling Launches Custom Face Models
Similarly, it’s about to get much easier to generate convincing videos of yourself (or anyone else). Kling has given access to a limited set of creators and the results look impressive. Remember that uncomfortable gray area we mentioned? Chinese AI companies, arguably taking the most liberties with training data, continue to push the boundaries of both what’s technically possible and ethically justifiable.
We đź’› This - Descript
We love flashy AI solutions as much as the next Twitter thinkperson, but Descript has built AI into nearly every aspect of its product in a way few others have.
What started as a children’s video editor has sneakily grown into a powerhouse editing suite (one we use every week) offering a host of features you can’t even get from pro tools professional tools. The best part is they’re integrated in a way that makes you forget you’re using AI, which, isn’t that the point?
Some of our favorite AI features:
Remove retakes - Analyzes the script to identify additional attempts, removing all but the last one
Overdub - Replace a word or phrase with something different in your own voice
Regenerate - Replace with the same thing in your own voice, because you said it weird or bumped the mic
Eye contact - Forces your eyes to be locked on the camera because millennials make videos too
It’s among our favorite examples of AI done 🤌🤌
Are you a creative or brand looking to go deeper with AI?
Join our community of collaborative creators on the AI4H Discord
Get exclusive access to all things AI4H on our Patreon
If you’re an org, consider booking Kevin & Gavin for your next event!