YouTube Shorts creators are in for a treat as they’ll soon have access to some cool tools powered by fancy AI tech. These tools, backed by Google’s DeepMind, will let them whip up music for their super short videos in the style of famous artists.
In this musical collaboration between YouTube and DeepMind, they’ve rolled out an advanced AI model called Lyria. This nifty tool is designed to get creative with music, handling everything from instrumentals to vocals. It’s not just a one-trick pony either; Lyria can pick up on an existing piece of music and seamlessly continue it.
Explaining the complexity of the task, Google shared, “Music contains huge amounts of information — consider every beat, note, and vocal harmony in every second.” Lyria, however, is up for the challenge, excelling in maintaining musical flow across different sections and passages.
The initial test run for Lyria will be in the Shorts realm, those snappy videos maxing out at 60 seconds. It’s like a playground for experimenting with AI-generated music, even though the trial will yield a 30-second soundtrack. A select group of users will get their hands on this generative AI model, allowing them to input a text prompt describing the kind of music they’re after. Lyria then works its magic, producing a complete soundtrack with harmony, melody, and vocals for their video.
But here’s the cool part: Some big-name artists like John Legend, Sia, Charli XCX, Louis Bell, and T-Pain have pitched in, lending their unique styles and voices to DeepMind. So, users can pick their favorite artist, describe the vibe they’re going for, and let the AI generate a soundtrack that fits the genre and style. It’s like having a virtual collaboration with your musical idols for that perfect background tune to your Shorts video.
“The first time YouTube presented me with a proposal, I was quite skeptical and remain so today. “AI is going to change our world, including that of the music culture,” said Chari XCX who had her track speed drive adopted in this summer’s Barbie movie. This experiment, though, provides only a minor glimpse at the potential creativity involved. It will be exciting to see what results emerge from this.
In addition, Google showed its AI music tools that would be rolled out within few months to the creators that would help them during the imaginative limits of their creativity.
During a demonstration, Google showed how one can convert melody into the brass line using Lyria by singing or humming a part taken from music. For instance, the other part of the song created by the MIDI keyboard is changeable and can take the form of authentic choir or imitating musical accompaniments.
According to the company, lyria music tools could be used for creating fresh music genres and even a whole instrumental track using only music from one genre or one instrument. In this, artists could decide to go from folky music to metal in order to understand its impact on their music at once.
Adding a touch of detective work to the mix, the Lyria model team spilled the beans about a little something called SynthID. This system, crafted by the brilliant minds at DeepMind, acts like an invisible signature for content generated by Google’s Imagen image-generating AI model in Vertex AI, the company’s go-to platform for AI building. Essentially, it’s a nifty watermarking tool that leaves an indelible trace on the images, making it possible for users to track down artwork created by this AI wizardry. And guess what? This same SynthID comes in handy for Lyria’s musical creations too, letting users verify if a track is indeed a brainchild of Lyria.
Now, this revelation unfolds against the backdrop of some stormy weather in the realm of AI and entertainment relations. Universal Music Group threw down the gauntlet in October, slapping a lawsuit on AI startup Anthropic for copyright infringement. Their bone of contention? Anthropic’s chatbot Claude allegedly scraped song lyrics from the Universal Music Group’s clients to fuel its AI training, pumping out reproductions and even churning out new song lyrics and poems mimicking the artists’ styles.
And if that wasn’t enough drama, the Screen Actors Guild – American Federation of Television and Radio Artists recently wrapped up a strike, and guess what played a part? You got it – AI. One of the major wins from the negotiations involved setting up some AI guardrails. Now, film and TV producers have to get the green light from actors before using their likenesses in generative AI replicas. Plus, there’s compensation in the mix for putting those AI-created doppelgängers to work. Looks like the AI entertainment saga is getting more intricate by the day!
Leave a Reply