AI Music Today: Can You Believe What You Hear?

by | Feb 27, 2024 | tech & ai | 0 comments

I’ve been banging on this AI drum for a while, but now AI is banging on that drum as well. A virtual drum. And guitar. And sax. Basically, the whole band, singer and all.

My techno friends are already rolling their eyes because they know what I’m about to mention… Suno. But before I get into the details, let me just say this: the way you consume media is going to change drastically in the next five years. More specifically, WHAT you consume is going to evolve in a massive way.

Most people have already experienced AI-generated images, and quite frankly, that’s old hat at this point. You’re not going to impress anyone with your Midjourney prompts or kitten amalgamations. OpenAI just blew the lid off everyone’s expectations with their latest release, Sora. Haven’t seen it, yet? Well boy do I have something to scare you with.

I firmly believe at some point, and it’s not too far off, music and video will be entirely generated specifically to your preferences, and maybe even your personal history. Imagine popping open Spotify and listening to an entirely unique playlist of music that matches your tastes exactly, with no song ever heard before by any other person.

Or maybe you have an entire Netflix drama generated that features you, your friends, and family members. Hopefully at Thanksgiving and Christmas, when the true drama is destined to manifest.

We can already clone voices with only 20 seconds of recording needed to make those voice models fairly accurate. If you don’t believe me, visit ElevenLabs and message me back. It’s only a matter of time that mimicking a whole person will be child’s play.

I do not think any of this is good or bad. It’s a murky grey area that we’ll just have to work through together. Some nefarious things can definitely be done with this technology, but also some really, really, friggin’ cool things as well.

But back to music.

I’ve been meddling with an AI platform called Suno for several months now. If you visit my Discord channel, you can see all the people annoyed with me about it. Suno is an AI application that can either generate lyrics, or allow you to add your own, and then add those lyrics to real music in any genre you want. This platform has become so popular that Microsoft has partnered with Suno to include it in its Copilot product.

For a while now, this technology sounded robotic and manufactured, so much so, that it has become a meme. However, the AI tech is evolving. Suno released in alpha their new V3 algorithm, and I can say that we are closely approaching the point where completely generated voices sound close, or identical, to Auto-tuned singing.

What’s Autotune? Only the most overused technology by the music industry, which enables tone-deaf musicians to reach pitch perfect tone, or alternatively, yodel like Mariah Carey. We, as a society, have grown so accustomed to Autotune that I argue if AI can match that style, it will be highly difficult for normal ears to differentiate.

The only gap in automating custom playlists of fully AI-generated beats for you is some integration app that submits songs based on your preferences, then auto-publishes them to your music platform of choice. That is a tiny gap that could probably be built right now. Using ChatGPT.

Right now, for the most part, I’ve only been leveraging Suno to create absurd music that would never exist in the normal music industry. At some point, though, I will start making real music that I’d actually want to listen to over and over again. If this tool can provide that for me, then the music industry is doomed. Well, not doomed, but not in good shape either.

I actually love the idea that music can be accessible to anyone, regardless of the skill gap. I just hope relying on AI tools such as Suno doesn’t stop someone from learning the technical expertise to also make their own works.

You’ll hear me say it a lot but I also believe in “show, don’t tell,” so I’ve generated several music clips for you. You’ll notice these are actually the exact same song, with lyrics generated by ChatGPT, then set to different music styles. While silly, you can definitely see where the technology is going. Suno will let you generate up to two minutes of audio, but then you can always extend each clip indefinitely with new lyrics or even instrumentals.

Buckle up.

Other stuff worth looking at….