Bot Eat Brain is back! 🤖😋🧠

Meet your new favorite AI vocalist 👩‍🎤

What's up bots? After a brief (months-long) hiatus, your favorite AI newsletter is back and better than ever.

We're here to keep you up to date on all things AI and give you ideas for your next game-changing business.

Today we're talking about the intersection of AI and music.

Just like 2D and 3D art is being eaten alive by stable-diffusion, DreamBooth, and the full plethora of generative art technology, so too is music undergoing a major disruption.

If you like this newsletter, share it with your friends.

In this issue:

  • Meet our favorite AI singer 👩‍🎤

  • Create music from prompts 🎙️

  • Create music from images 📻

  • Projects on the frontier of AI x music 🪗

This issue is sponsored by Elephas AI

Save time and improve your writing with Elephas, your personal AI writing assistant.

Answer emails in one click, instantly adjust the tone of your writing, or let Elephas finish your sentence.

Elephas integrates seamlessly with your Mac, iPhone, and iPad so you never have to leave your favorite apps. An AI product that you will actually use every day.

Interested in sponsoring the next issue? Click here to book a slot.

Today's Vibe: Why create what you can generate?

Let's dig in.

Meet our favorite AI singer 👩‍🎤

Can't sing for shit? 💩 In our new reality it doesn't matter:

  1. Pick the voice you wish you had.

  2. Upload your best musical effort.

  3. Chill out as our new AI pal makes you sound like a rockstar.

Holly+ is an AI singer trained on hundreds of hours of singing data from vocalist and Stanford Ph.D. holder Holly Herndon, and it's just one example of this new tech in action. Some people can really do it all, huh?

Holly collaborated with the team at Never Before Heard Sounds to create a reusable singing model anyone can use to personify their inner starlet.

Click on the gif above to listen to the demo. Notice how you can even hear the AI "breathing" and taking dramatic pauses between verses. We are far beyond the days of Microsoft Sam.

This isn't the first time we've seen synthesized human voices. Autotune has been around for decades now and you may already be aware of Vocaloids — a product that's been around since 2007 that lets you conduct virtual vocalists.

What's different here is the adaptability and the quality of reproduction. This new technique can be trained on any voice and it sounds much more human.

Who owns a song sung with your virtual voice?

Now you're a singer with a virtual twin anyone can use to sound just like you. That's great for them, but doesn't it suck for you? Who's going to pay you to sing when they can generate something just as good on their laptop?

IP rights in a post-AI world is something we've written about before.

It may be considered IP theft if you replicate someone's voice without a license. In the case of Holly+ above, Herndon has elected to license the rights to use her voice via a crypto-powered mechanism called a DAO  or "Decentralized Autonomous Organization".

A DAO is basically an organization anyone can pay to become a part of and can then vote on how the assets of the organization are used. In this case, you could vote on who gets to license the Holly+ vocal model.

What if I don't want to write the song either?

Sure, having AI sing your song for you is great, but someone still has to write the tune. What about those of us who are both untalented AND uninspired?

Well, my lazy brethren, you're in luck. From just a short prompt, this AI can generate a whole song for you.

Here's an instrumental one of our writers produced using just the prompt "Bot Eat Brain":.

The software is named "Mubert" and you can try it yourself here with a free Hugging Face account:

From Image to Music 📻

But wait, there's more! If a picture is worth a thousand words we really shouldn't need to bother with writing at all, right?

We've already waxed lyrical on tools like stable-diffusion, which can generate images from text, so the idea of using images to generate music should not be a complete shock to your system.

Here, our brave writer has offered their own profile image as food to an Img To Music model. Listen to the result:

And play with it yourself for free here:

Your billion-dollar ideas 🤑

Art and expression are cool and all, but who's looking out for the money-hungry capitalists among us? You're not here to inspire the soul, you're here to get rich damn it!

Hear you loud and clear bucko, and this is what we predict is coming right around the corner:

Idea #1: Entirely synthetic bands composed and performed by AI.

Gorillaz is a concept band of 3D avatars (singing songs written by humans). K-Pop is basically a soulless corporate creation.

Fully AI-powered bands are a logical next step that will remove the one thing corporate labels currently struggle the most to control: their artists.

Idea #2: On-the-fly ad creatives.

Ads ads ads. The world is loud and it's hard to stick out. But you know what always sticks with you? A catchy jingle. The perfect song.

Once reserved for only the biggest brands, AI will let every mom-and-pop afford a full-time (simulated) orchestra.

Idea #3: Hyper-individualized music.

Spotify, Apple, and YouTube invest billions trying to figure out what music to suggest to you based on what they know you already like, but in a world where the cost to produce music drops to nearly free and nearly instant, why limit yourself to what already exists?

End-to-end algorithms like we've described above will only get better and faster, to the point that Spotify (or your own billion-dollar unicorn startup) could generate the perfect song for you just as you hit play.

    More musical quick-bites 🫦

    If you're excited about the future of AI in music and want to dig more, then you're in luck. There's a bevy of active research in the field and new projects coming out constantly.

    Here are just a few more to get you started:

    Found something in this list (or missing from it) particularly cool? Hit reply, let us know, and we might do a full write-up on it in the near future.

    What'd you think of today's newsletter?

    Login or Subscribe to participate in polls.

    Today's Featured AI-Generated Art 

    Until next time ✌️


    P.S. Apologies for the long delay between issues! Editing staff got busy trying to make a living. Want to help us keep this thing more consistent in the future? Click here to join our writing squad.