Google Gemini gets a soundtrack button with Lyria 3

What Gemini Lyria 3 does exactly

If you break it down, Gemini Lyria 3 is a kind of music printer in the Gemini app. You input an idea, and a track of about 30 seconds comes out – with instruments, vocals, and automatically generated lyrics.

Technically, Lyria 3 is a generative music model from Google DeepMind, embedded in the Gemini app. Already available on desktop, and gradually in the mobile app as well. You don't need to know a DAW like Ableton or Logic; you just work with prompts.

You can go two ways with Gemini's Lyria 3:

• From text to track: you type what you have in mind: genre, mood, joke, situation. For example:

"Create a cheerful afrobeat about our Friday afternoon drinks with colleagues." Lyria comes up with a melody, beat, vocal line, and lyrics.

• From image to track: you upload photos or a short video and give a prompt, such as:

"Use these photos for a song about our dog during a forest walk." Lyria tries to capture the mood of the image in music and lyrics.

With the 'From image to track' feature, Gemini Lyria 3 tries to directly translate the mood of an uploaded photo into matching music and lyrics.

You have quite a bit of creative control. You can:

• direct style, tempo, and mood ("fast drum 'n bass", "calm piano ballad", "dark trap beat");
• indicate whether you want vocals (male/female, high/low) or just instrumental;
• leave the lyrics entirely to Lyria or provide your own lyrics as a basis.

The output is always compact: a track of up to 30 seconds, plus a piece of cover art that is automatically generated (via Nano Banana) and a link to download or share the file.

There is a threshold: the feature is 18+ and available in countries where Gemini operates. If you have a paid AI subscription from Google (AI Plus, Pro, Ultra), you can generate more tracks per day.

It's important to keep in mind: Lyria 3 is not a replacement for a producer or a full-fledged studio, but a tool to quickly drop TikTok-length soundtracks and ideas. It's more of a sketchpad and snack content music than a final mix for Spotify.

Want to know the broader context of Google's Gemini as a work assistant? Also read: New Gemini 3: AI that truly takes over work processes.

Five practical use cases for creators, marketers, and SMEs

From TikTok intro to birthday song: what can you actually do with it?

The temptation is great to see Gemini Lyria 3 only as a gimmick ("haha, a song about the coffee machine"). Nothing wrong with that, but as an organization, you can get a bit more out of it. Five scenarios where AI music turns out to be surprisingly handy.

From social media to internal fun: AI music shifts from gimmick to a useful tool for quick content.

1. Quick soundtrack for social posts & Shorts

Small brands rarely have the budget for their own jingle or composer. Yet a social video without sound immediately feels poorer. With Gemini Lyria 3, you can generate your own audio hook per campaign or product in a minute.

Think of:

"Create an energetic electro-pop track of 30 seconds about our new spring campaign, cheerful and optimistic."

You can use it under an Instagram Reel, TikTok, or YouTube Short. No hassle with rights to popular music, but a recognizable sound that fits your content. And if the vibe isn't right? Just adjust the prompt, generate another version.

2. Mini-intros for podcasts, webinars, and demos

Many organizations experiment with podcasts, webinars, or recurring demos, but remain stuck with generic stock music or silence. Lyria 3 is ideal for temporary intros and bumpers:

• a short leader for an internal podcast;
• a recognizable tune for your monthly webinar;
• a "sting" between sections in a product demo.

With a prompt like:

"Calm, modern electronic intro of 30 seconds, without vocals, suitable as an opening for a B2B webinar on cybersecurity."

you'll have a usable base within a minute. If it's a success and you want a definitive tune? Then a real producer can later build a professional final result on this direction.

3. Internal communication that feels less boring

Not everything has to be marketing. Internally, AI music can bring some lightness. A few examples you can throw into your Teams or Slack channel:

• A birthday song for a colleague, with their name and a few internal jokes.
• An "end of sprint" anthem for your dev team.
• A short song about the new password policy or clean desk policy.

For example, prompt:

"Create a comedic pop track in Dutch about our support team that always takes lunch breaks too late, with plenty of relatable humor."

The result is rarely radio-quality, but it does break the ice and makes dry topics a bit more fun.

4. Concept tool for agencies and musicians

For agencies and musicians, Lyria 3 is not a threat, but a sketchpad. Instead of just describing a direction ("more upbeat, something between pop and disco"), you can play a rough AI track:

• Creative agency: propose different soundscapes for a campaign.
• Musician/producer: explore what mood the client is looking for in the preliminary stage.
• Video editor: lay a temp track under a rough cut.

Important: don't sell this as a final product, but as a reference. The conversation becomes more concrete ("yes, this tempo, but with real guitars" or "less busy, more ambient").

5. Education and training with a wink

Another unexpected use case: training and education. A song often sticks better than a slide full of bullets. With the use of Gemini Lyria 3 (5), you can quickly create a rap about safe working or a tune about the new core values. Trainers, teachers, and SMEs can quickly:

• create a children's song about the times table of 7;
• a rap about cleaning up or working safely;
• a cheerful tune about the new core values.

For example:

"Write a cheerful Dutch children's song in bubblegum pop style about the safety rules in the warehouse, with a chorus that sticks easily."

There's a good chance your employees will still hum the chorus as they walk into the warehouse. Is it high art? No. Does it work to get a message to stick? More often than not.

Fun, but what about rights and misuse?

AI music immediately raises questions: Who owns that song then? Can I use this in a campaign? And how do you prevent misuse? Google builds a few clear boundaries around this.

First of all, every Lyria track is invisibly labeled. With SynthID, the audio receives an inaudible watermark, so it can later be recognized as AI-generated music. In Gemini, you can even upload an audio file and check if it was made with Google AI. Handy for platforms, but also for those who don't want to accidentally present AI audio as "real" work.

Then the question: "Can I just have a song made in the style of [insert famous artist]?" Officially, Lyria 3 is built for originality, not to replicate artists 1-to-1. If you mention an artist in your prompt, it will at most be used as broad inspiration for mood or genre. Additionally, there are filters in place to prevent you from directly recreating an existing track or infringing copyrights.

Still, it's a young and shifting domain. If you want to use AI music for serious commercial applications – campaigns, TV spots, larger brand videos – it's wise to:

• check the current terms of use from Google;
• have your lawyer or agency take a look;
• make agreements about how you label AI audio to the public.

For internal content, socials, and fun (Reels, Shorts, internal videos, birthday songs), the risk is much smaller, as long as you don't include trade secrets or sensitive data in your prompts. Think of it as working with a mix of stock music and AI: very useful, but you remain responsible for context and rights.

In short: Lyria 3 gives you a lot of creative freedom, but it is not a magical free pass. Use it, experiment with it, but maintain the same legal and practical common sense as with any other content you send out into the world as an organization.

Want to try it yourself? Here's how to get your first track in 3 minutes

Eager to hear it yourself instead of just reading about it? Here's how to roll out your first result and track from Gemini Lyria 3 in a few minutes.

1. Open Gemini and choose "Make Music"

• Go to gemini.google.com on desktop and log in with your Google account (you must be 18+).
• In the interface, choose Make Music. Lyria 3 is under the hood

2. Choose: text or image as a starting point

• Text only: type what you want to hear, for example:

"Create a melancholic synthwave track about the end of the festival season."

• Photo + text: upload a photo or short video and provide context:

"Use this camping photo and create a cheerful campfire song about our annual team trip."

3. Specify style, tempo, and instruments

The more specific you are, the better the result. For example, combine:

• Genre/era: "upbeat 90's eurodance", "lo-fi hiphop", "epic orchestral score"
• Tempo/energy: "slow and dreamy", "fast and energetic"
• Instruments: "with prominent piano and strings", "lots of synths and heavy bass"
• Voice: "cheerful female voice", "low male voice", or just instrumental

Example prompt:

"Upbeat 90's eurodance with heavy bass, fast kick drum, and a cheerful female voice about our new product launch weekend."

Generating a track starts with a simple, descriptive text prompt in the 'Make Music' interface of Gemini – that's where Lyria 3 is activated.

4. Listen, tweak, try again

After a few seconds, you'll receive a track of up to 30 seconds back, complete with automatically created cover art and a download/share link.

If you don't quite like it? Adjust your prompt ("less busy", "no vocals", "faster tempo") and let Gemini generate again. With a few iterations, you'll often find something that hits just the right vibe.

Tip: have Lyria 3 create 2–3 variants on (almost) the same prompt. The differences are surprisingly large – perfect for figuring out what works for your content or audience.

Here are 5 prompts to test Lyria 3 in 10 minutes:

1. "Create a comedic R&B track about our team's Monday morning stand-up."
2. "Cheerful 80's synth-pop about my webshop that has reached its 1000th order."
3. "Calm lo-fi beat with soft piano about focusing on a deadline."
4. "Children's song in bubblegum pop style about cleaning up and going to bed for my kids."
5. "Energetic rock track without vocals for the intro of my product demo video."

AI music as an extra layer on top of your content

Lyria 3 will not replace bands, producers, or composers. What it does ensure is that there is practically no barrier to adding sound to your content.

For students, creators, and professionals, it mainly means that you:

• can experiment faster with sound for campaigns, Reels, and Shorts;
• make small productions (from an internal video, demo, pilot, or business pitch to a video assignment for your studies) feel a bit more professional;
• allow teams that "have nothing to do with music" to create their own jingle, intro, or joke.

Whether you're mapping out a marketing strategy or as a student running your first AI-based productions: see Gemini Lyria 3 as an extra layer on top of what you're already doing with AI text and AI image. And honestly, if you're already playing with prompts, this is a logical next step. Put on headphones, throw a crazy idea into Gemini – and listen to what happens.