iJustine just uploaded a video and it seems decent, better than the humane pin. This thing is exactly what I want my phone to do in the future.
This might be my favorite AI tune I've heard so far. As someone who's not as familiar with AI-generated music: what portion of this (if any) is human-made? Specifically the lyrics--they're funny and they're good at sticking to a theme. I'm wary of stuff that presents as fully AI to get clicks, but in reality had a heavy human hand behind the scenes.This one is a pretty decent country song, like I honestly kind of like it:
Udio | Wow..... I Didn't Know That [Full Track], Americana, Country by BobbyB
Make your musicwww.udio.com
First time seeing AI music in action on that Suno site.Okay. I wanted to run an experiment with this Suno thing out of pure curiosity. I took the lyrics and soundscape ideas of an actual song I'm working on and shoved them in. So working with my own music. I got this
Nurgle's Garden | Suno
death metal, melodic death metal, synths, square wave synths, growling vocals, gutteral sounds song. Listen and make your own with Suno.suno.com
It's a bit messy, but I appreciate the ideas. I think some of the vocal patterning on the verse is cool, though not what I went with. It got close with parts of the chorus. Instrumentally it's chaos in a way I don't like but I could see being appealing. I've listened to plenty of deathmetal tha'ts just wall-of-sound.
For proper reference (and sorry if this breaks rules for advertising, but it's a soundcloud link lol) here's the actual song in it's first-mix form
Cauldron Dance first mix
Listen to Cauldron Dance first mix by Inonum #np on #SoundCloudsoundcloud.com
I don't know what to make of this thing yet. My friend has used it to make some real fun shitposts that you listen to once and then forget. Where it goes from there? Not a clue.
In its current state, it is a very remarkable toyFirst time seeing AI music in action on that Suno site.
I of course immediately gave in and used it to make a shit post country song about Fallout with the joke prompt "A western song about how Fallout New Vegas is better than Fallout 3 and I won't shut up about it".
feeling the need for an extended cut but might not be as good lolFirst time Suno was able to spit me a synthwave song with saxophone and I really like it.
Midnight Shadows | Suno
atmospheric synthwave saxophone song. Listen and make your own with Suno.
It's pretty scary how good those things are right now, it's gonna be crazy in a couple of years.
I think there is a huge difference between a computerized workflow replacing physical props and a black box replacing CG. CG can have varying levels of quality but there is still someone deciding what camera angle to use, controlling the lighting, tweaking the assets as needed, etc. As soon as a significant number of major productions that are actually trying to create something artistically beautiful, like Masters of the Air, settle for stuff like a plaintext prompt of "a squadron of bombers flying back to base at sunset" and just run with whatever their model spits out, we truly will have lost something. Maybe the model has some guardrails and knows what specific planes to use based on previous scenes, maybe you can adjust a slider to tweak the exact time of day on the final output, but the idea of it is still depressing and disgusting to me.
I hesitated to make this post because I know how this community feels about generative AI, but after finally sitting down with Udio as a veteran composer and musician with over 30 years of experience, and publishing a song that is currently trending/charting on the Udio platform, I think it's a good opportunity to demonstrate the usefulness of generative AI for artists like myself and the positive impact it can have on the entertainment industry.
Here is the song:
Udio | Transfiguratio Animae (Full Version) by Blambda | AI Music Generator - Official Website
Listen to Transfiguratio Animae (Full Version) by Blambda on Udio. Discover, create, and share music with the world. Use the latest technology to create AI music in seconds.
What is noteworthy here is that this was merely an experiment and the first [full] song that I published on this platform (I published a snippet of the song first to gauge appetite and encourage the community to riff off it). I knew going in that my options would be limited since my ability to control the outputs were limited to prompts, but I saw the potential in the technology and understood that people with strong musicality and good prompt engineering skills (both are requirements for high quality outputs) would likely rise above the rest, and that is exactly what happened. The point here is that artistic intentionality is something you can imbue in AI generated content, but the degree of intent will always be limited by the level of control that you have over the outputs. A basic prompt and a sophisticated, well-informed, well-structured prompt are going to result in widely different levels of quality, and iterative prompts exponentially boost that quality. And in this case, I simply used my musical sensibilities and experience to guide the model the best I could. You simply cannot get this level of precise intentionality without understanding what you're doing, full stop. And that's important, because it means that even when using generative AI (even with only prompts!), artists still provide significant value compared to the average consumer that lazily requests a random song. I don't think artists are going anywhere and our talents will always be valued, even when AI has far exceeded our ability to produce high quality content.
If you're interested in my process for generating this song, you can read about it my reddit comment here:
Having said all that, though I was able to create something I was eventually happy with on the platform, I cannot seriously use this as part of my workflow until I have much greater control over the output. The model was doing way too much heavy lifting for my tastes, and I had to fight with it a bit to align with my intent, and no artists wants that. Once there are some real editing options, I see this as a very potent tool in the toolkit and I'm very much looking forward to when that day comes. What's most important for me is the creative aspect, the production aspects are less important and I would actually prefer for AI to rid me of a lot of that tedium.
EDIT:
I should add that I did not intend for this post to be a slight against laypeople using Udio. There is absolutely value in that and I fully encourage everyone to try it out. I just wanted to provide a little optimism for artists who are worried that AI-generated content getting better will mean that people will no longer be interested in their content, when the opposite seems to be true.
I understand that analogy, but I think what people feared most is that... well, not everybody can be a 5 star gourmet chef. Only a small amount of people working in the industry possess the incredible talent compared to the amount of average folks working in "fast food" content. As well, there are also people's career growths to consider where they may start working in "fast food" to gain experience and then become a master gourmet chef. Point is, there's a lot of artists working in the areas that could be hugely impacted by these AI. Not only can that be disruptive of people's career now but also future careers of people working their way up the food chain. That is a serious issue to consider.
As someone who spent a few hours trying to remix a Suno song, yeah. it's just better to make your own from scratch.
