It’s been a whirlwind of experimentation lately with AI music and I’d like to share some of my experiences with SunoAI because I think, at the very least, you’ll have fun with it and maybe be inspired to pick up that axe.
In the ever-evolving world of music, computerized innovation continues to shape how artists and creators bring their visions to life. Enter SunoAI, a fairly groundbreaking AI tool designed to revolutionize the way we create music. In some ways, EDM artists have been doing this with success for some time, but lets get out of the dance and electronic realm and see what the rest of us are getting from it. Let’s look at what SunoAI is, how it works, and the myriad benefits it offers to musicians and lyricists alike. I’ll even give you a sample or two of stuff I’ve been doing with it.
What is SunoAI?
Found online at https://Suno.com, SunoAI is an advanced artificial intelligence platform that empowers users to create melodies, harmonies, and even full musical compositions with ease. I’m not saying they don’t need icing on them after you make them, but they can occasionally come out pretty unique. Whether you’re an aspiring songwriter, a seasoned musician, or someone looking to explore the joy of music creation, SunoAI offers a powerful and intuitive playground to maybe enhance or stimulate your creative process.
How SunoAI Works
At its core, SunoAI leverages some cutting-edge machine learning algorithms trained on a vast and diverse dataset of musical compositions. This machine training allows the AI to generate melodies and harmonies in a wide range of styles, from classical to EDM to pop, rock, metal, regional styles and more. Really, virtually every style of music, and believe it or not, things you wouldn’t expect – like Trap Flamenco with Doom Rhythm Undertones. This may have never existed, but like a chef in the kitchen, you can literally throw anything in the bowl and discover it’s flavor – good or bad. What’s more, and this is something the engineers at Suno have stated so it’s not from this horses mouth, “The process is designed to ensure that the AI does not replicate existing works but instead creates unique compositions inspired by its training data.” More on this based in reality as we go along.
Ease of Use
One of the standout features of SunoAI is its user-friendly interface. Even those with minimal musical training can quickly grasp how to use the platform. It’s actually nothing musical at all. It’s more of a Willy Wonka creating thing – if you can dream it, it will make it. By simply inputting a few parameters, such as the desired mood, genre, and key, you can generate your desired style of music effortlessly. This ease of use democratizes music creation, making it accessible to a broader audience. Understand though, you aren’t choosing more than the style or the lyrics you may want – and yes, instrumentals are a thing, it’s got settings for that. Users can experiment with different styles and moods, discovering new sounds and combinations that they might not have considered otherwise. By inputting specific prompts related to genre, mood, tempo, and instrumentation, users can tailor the generated music to fit their creative vision, opening up a world of endless musical possibilities. The prompt thing is more detailed but you’ll figure it out fast. I’ll explain further in a minute.
Supporting Lyrics with Melodies
For lyricists, SunoAI offers an invaluable resource. Where creating melodies that complement and enhance lyrics can be a challenging task, SunoAI simplifies this process by generating melodies that fit the lyrical content, to what it’s been trained on. You just want to hear that song you penned with music behind it, this is a good way to do that. Again, if you want it in G/D/E/B – you might not get it, but maybe – I’m not sure I’ve experimented that far, but keep reading.
Live Performance Potential
SunoAI holds potential for live performers. This might sound cheesy, but in Los Angeles where bands spend three and a half years in a rehearsal room repeating the same 9 songs and never actually get on stage, you could theoretically knock out a set list of tracks here, learn them over a couple weeks and play a dive bar like the Universal Bar and Grill while you refine the tunes, heck, play The Whisky. Musicians can theoretically use the platform to create an entire consistent setlist of songs never heard before, complete with diverse styles and moods, ready to be performed on stage. This sort of opens up exciting possibilities for live shows, allowing artists to blend AI-generated compositions with their live performances seamlessly. I mean, you’ve still got to learn them and adapt them to your style, but hey – here’s a set you didn’t have to spend time writing so now go out and pay Lance Sterling $500 and open up for Berlin at the Canyon. Congratulations – you’re a band.
Ethical Considerations
SunoAI claims a commitment to ethical music creation. I didn’t invent it so I have to go by what you read about it online. The AI is claimed to be trained to avoid directly replicating existing works, ensuring that each composition is original and free from plagiarism. Again, more on this below. Their claimed approach addresses concerns about the potential for AI to infringe on the creative rights of existing artists and maintains the integrity of the music industry. ASCAP has been sending me stories that dispute this, but hey – are you Michael Buble? No – just go make some songs and play them somewhere.
Okay – let’s get into the meat and potatoes here. You’re going to take 5 minutes to learn how the prompts work. You’ve got two main ways to describe your song – the first is to either give it lyrics, or direction – and the second is to define the style.
Lets start without lyrics.
In the segment that describes the music style, you can be more explanative but it’s going to address your request to conform to it’s learning. If you simply type “jazz” – you might not get Chick Corea, you might get Frank Sinatra. Be more descriptive. If you want Fusion, try “Fusion, Jazz, Modern.” Understand that it’s trained to stop you from using actual names, though people have tried work-arounds for that. If you put Metallica, it’s going to tell you that it’s a band name and it can’t do that. if you put Metallic A, it will likely give you an aggressive metal song and name it as such. It’s AI, it’s got rules but it isn’t dumb.
Now lets consider lyrical content.
You’ll find success doing this:
[intro] [verse]
Add your lyrics here.
[pre-chorus] Add your lyrics here
[Chorus] Add your lyrics here
… and so on. Finish your song with [outro]. Putting directives in brackets is the key. If they aren’t bracketed, the voice will actually sing them. Play with it and you’ll understand.
This is a song where I asked it to create – “syncopated jazz fusion, modern“. I then took the song it gave me and used Moises to separate it into stems. I loaded each individual track into Reaper, used ToonTrack’s EXMix to improve the audio quality of each instrument set and recorded my own bass on it. After all, I’m a bassist, I’m making jam tracks for myself here. Check out the end result.
It’s not bad. Truthfully, instrument tones may be lacking in some departments, to this date no one has really created a realistic and non-robotic distorted guitar – but in the clean jazz realm, this is alright. With EZMix I was able to make it sound pretty top notch.
Now let’s look at what happens when you add lyrics. For the next example, I decided to create a thrash metal song about Howard Stern. Yes, you can virtually do anything in your creative musical mind with SunoAI.
I fed it the lyrics and asked for the style to literally be “Aggressive Thrash Metal.” Is it Nuclear Assault? Maybe, but how cool is this?
There have been some developments lately at SunoAI that have thoroughly enhanced song creation since it’s inception. At the start, you could only make 2 minutes of music and you could “Extend” the song in 1 minute increments – that’s how the Howard Stern Show song was made. Ultimately, it was a costly pain in the ass to get parts I could extend and make into a decent sounding track – but it’s got a lot better. Each time you hit the button to create, you’re spending 10 credits and getting two results. Credits add up when every song isn’t a winner – but it’s improved in version 3.5 a whole lot. We’ll get into the cost in a minute.
SunoAI now has evolved into Version 3.5, and by the time you’ve read this I’m sure it will have continued to evolve. Version 3.5 creates tracks up to 4 minutes in length and adds increments of 2 minutes, so maybe you can be the new Return to Forever or Mahavishnu orchestra. Not likely, but consider the dream – did anyone over 40 ever see this being a thing? We have no idea of the potential at this time, but it’s there and it’s looking good.
So, here’s a question: How many different tracks can actually be made before it starts repeating itself? In late May 2024, the company announced that 10 million people had signed up to make music with SunoAI. Ten million is a lot of people! Are there even 10 million unique songs in the world? Well, frankly, no.
For kicks, let’s look back at an article written by Rhett Allain, a brilliant physics professor, who penned some interesting notes in Wired Magazine in 2015. In his article titled “How Many Different Songs Can There Be?” he essentially pondered how many unique songs can actually exist. Music is math with feeling, but math has limitations. Only so many configurations can arise from a limited field of numerical values. He delves into the physics in his article ( https://www.wired.com/2015/03/many-different-songs-can/), and to summarize—no, there aren’t that many songs.
So what is everyone paying for? Well, I don’t know all those people, so even if they made a track virtually identical to mine, no one is going to notice. There are enough different genres, tempos, key changes, and structural enhancements that the same four-chord progression can sound distinct enough each time. However, here’s something to note with SunoAI: you can hear the repetition, almost immediately, but again, this might not matter to you, depending on your reason for using SunoAI.
Here are 4 tracks created at different times by different users on SunaAI, I’ll just give you the start of them. You’ll hear the similarities.
I guess, in the end – who cares? It’s like comparing two MTV acts circa 1988, they’re in the same genre, using the same chords to write the same style of music. Have fun with this thing and be inspired.
Lastly, about the Cost
The good news is you can try it free! Just sign up and get 500 credits, it’s good for making 50 songs and just enough to get you hooked. I think they even add 50 credits a day for free to keep you motivated. When it comes to actually buying credits, you’ve got options ranging from something like $30 – $50 with it automatically refreshing, recharging and refilling your credits monthly, and you get a boatload of credits with the ability to “top-up” when you run out. Remember, each time you hit the button to create a track, it’s taking 10 credits from your till and giving you two results. Ultimately, you’ll end up with a big-ass folder of concepts and inspirations on your desktop in the first month and probably still have credits left at the time it refreshes. Decide how committed you are to becoming a daily user when you buy – I recommend starting low and adding if you need it.
Warning: Much like GTA-V and Tik-Tok, the dopamine rush of creating tracks is a bit addictive and you’ll likely find in the first month that you’re going to spend a lot of time playing with this. Just make sure you’re having fun with it and not looking to make a career out of the results.
In the end, for how I use it, I highly recommend it. Sure, as a bass player I can get my reps in by playing to Zeppelin tracks on Spotify all day, but the fun of making something new, writing out a bass line that has never been recorded by anyone and creating something fresh from my bag of tricks is a wonderful experience. As for SunoAI, I highly recommend trying it out for yourself.
