I was playing a bit with some options for audio generation other than randomly flicking sliders in some DAW and found a couple that seems quite fun to try and piece something together.
Most of them do seem to lock you in when it comes to downloads which is a bummer. I am sure there are probably music generators out with very little to no limits on downloading the music but the main ones search turns up know they own the space so they are cashing in while they can.
Boomy is nice but gives very little true control over the generation. you can't really get a feel for what you might get without changing every setting over and over.
It still produces pretty clear albeit bit off-tune style sounds and for the most part, you can make as many as you like but can only save to their library 5.
On the free account, you can't download anything you generate but I am sure there are extensions able to capture the music as it streams which I suspect will become the norm if many allow for free tiers that are purely to "train".
Another one I looked at which I thought was cool albeit they lulled me into thinking I might use the file it gives is:
I like how they generate multiple variations of a song and then you can tweak the segments inside to add variance. Aside from that their generation is very similar to Boomy.
Most of these popular ones have the same trend really, use a base template it seems and then tweak some of the characteristics in that template of sound.
I am sure this kind of auto DJ'ing gets the job done but being spoilt in the image generation world I was hoping there is something similar for Audio.
I also tried Soundful and Loudly and well.... They are pretty horrible, although it must be said Soundful at least allows 10 downloads even if the AI Creation is hardly more than just that a template.
Loudly never ran a single generation because they were too busy for poor people. So fair enough.
I did not try it yet but the Hugging Face space MusicGen - https://huggingface.co/spaces/facebook/MusicGen - looks more akin to something a person might want to see given the image gen standard practice.
For something quick and dirty you can also try https://www.riffusion.com/ which at least although no download does give you the fun of typing genres and descriptions to mix as it plays.
It is not all doom and gloom and I think we will likely see more of this, similar to how canva were the first to truly deliver value to people with actual utility and not gimmicks. It seems VoiceMod is doing the same, by the looks of it this company is already established and that is why you get immediate utility via their Text to Voice offering... which is my favourite: FREE
With their https://tuna.voicemod.net/text-to-song beta you get to add lyrics of your own and the voice will sing it ontop of a predefined music template. This sure is what I am bitching about with the options above but VoiceMod is not advertising themselves as a music generator. They are a voice generator.
The coolest thing I found though, and I say find as though I spent more than 20minutes on it.
OpenSource is and will be driving the AI production stack the same way as it does across the board where people want to produce and not fluff about with gimmicks.
TuneFlow looks like a solid DAW built to be improved by AI features. This will give you full creative control but also the ability to quickly automate things for more creative inspiration.
I am no musician but I know about clicking buttons and it seems to me that even if I click the wrong buttons on TuneFlow I might just come away with a banger.
