TLDR:
AI music software has blown my mind just like ChatGPT did originally.
Udio and Suno are the main AI music softwares, and below I show how simple and powerful they are.
AI is moving super rapidly, it’s hard to predict what’s to come, but huge productivity gains have historically always improved our lives.
My posts are predominantly focused on Web3 and Crypto, however from time to time the latest developments in AI blow my mind and I decide to discuss them too.
I also feel that Web3 and AI are inately linked as they have many of the same entrepreneurs and early adopters navigating them, so I believe readers of my Substack will be interested in AI developments too.
Therefore this week I wanted to look at a couple of Music making AI platforms that I came across them a few weeks ago.
AI in Music
When I first tried ChatGPT in late 2022 my mind was absolutely blown.
It was my first “wow AI can already do this?” moment. I could tell something incredible had just been created and the world was about to change massively.
My use of ChatGPT has grown and grown since and I use it pretty much every day of my life now, it’s only on a rare day here or there that I don’t use it.
A few weeks ago I had my second “wow AI can already do this?” moment when I came across AI music software. My mind was totally blown at just how capable and powerful the technology already is!
I don’t feel like I’ll use this form of AI as often as I use ChatGPT because I’m not a musician, but if I was I’d likely use it every single day.
More specifically the two AI software solutions for music that I came across that blew my mind were Udio and Suno. I have very little musical flare or talent, but with these softwares I can make an orchestral symphony that would make Mozart proud, all with just a single text prompt like with ChatGPT.
Below I’ll show how both of them work and their individual nuances.
Udio
Udio was the first AI music generator I came across. When you open it up you immediately see a set of “staff picks” that can give you a great taste of what’s possible with it.
Clicking through all the staff picks allows you to see just how varied Udio’s generation can be. Although there’s an unendingly large library of songs that other people have made, the magic moment happens when you make something yourself.
To do this sign up by linking your Google, Discord, X or Apple account, and then just get started by typing a prompt in the box at the top. To get started you can write a topic and a genre next to it, for example I wrote “a song about the myth of an uncharted land, opera”, and hit create.
Udio’s AI will then do it’s thing, and assuming there’s not a high number of people using it you’ll get a couple of new tracks in under 2 minutes!
Udio will only make these initial generations 30 seconds long, but you can actually continue developing the song from there onwards. For example in this case I liked the “Realm of Dreams” version more and chose to “Extend” it further.
Here you can add extensions either before or after the song, or add an intro or outro, and then expand on it with more text input, or crop what you have, or even add your own custom lyrics that you want the next 30 seconds to have.
In this simple example I just gave it an outro, and it generated a couple new versions. By clicking the share button you can get something like this, have a listen:
The first time I did this my mind was truly blown!
Clearly Udio is very powerful and also malleable as you can keep tweaking and generating along the way until you’ve generated your ideal song.
As a bonus suggestion I recommend trying to add custom lyrics that you generated with ChatGPT, turning yourself into both a professional musican and lyricist!
Suno
Suno was the second AI music generator that I tried, and to be honest it’s probably the more user-friendly of the two because it can generate full songs for you in one go.
While Udio is more malleable and that helps to give you more control, it’s also more painful to do each section at a time. Meanwhile Suno just does everything for you and that makes it easier for non-professional musicans to just make something quick.
Just like Udio all you need to do is write a text prompt with a theme for your song and a genre and it will generate a couple of songs in under a couple of minutes for you.
The output genreated are full songs with a lot going on musically, a beginning, middle and end, and I’ve always been incredibly impressed with the results. You can customise with a few options and add lyrics, but it’s less customisable than Udio.
Selecting “search” will allow you to browse Suno’s extensive music library, although I admit this is much easier to do in Udio, which has already broken it all down into categories with a UI more reminiscent of Spotify.
There’s also a very interesting feature called “Explore” where you just select a genre in a big visual list of genres, and then it will show 5 pre-generated songs for you to flick through for that genre.
Explore is really cool to just see what’s possible with the technology and it helps you to be creative in what you might want to write as a prompt for your next generation.
AI is evolving rapidly!
Udio and Suno are both truly incredible! The songs generated often seem too good to believe that they are made by an AI because of how well they are produced and how human-like the singing voices are.
AI is evolving incredibly rapidly and to be quite honest keeping up is in equal parts fascinating as it is scary. At each new step you can see more and more of human creativity and work being done by AI at insane speeds.
ChatGPT can make an image in 20 seconds that would take an artist 2 days to make, and now the likes of Udio and Suno can make something in under 2 minutes that would take a professional musician 2 weeks!
What’s most curious is just how you’d imagine AI would take humanity’s most boring and repetitive jobs and do them for us, but in reality AI’s evolving faster on our more creative pursuits like art and music.
As we look around it’s clear that we’re going throug a rapid evolution and it’s hard to imagine what the world will look like 2 years from now let alone 20.
But just like the farmers in the 17-1800s before the industrial revolution could not imagine what their life would be like and now we live exponentially better lives, I’m sure in a couple of decades from now we’ll have a similar feeling of gratitude with AI as we live in ever wealthier and more productive societies.