Google Gemini will now generate a song for you. It handles both lyrics and music, and Czech is no problem for it

  • Google has integrated the Lyria 3 model into the Gemini app for generating music from text descriptions or photos
  • The tool creates 30-second tracks including vocals and automatically generated lyrics
  • The feature is also available in Czech, so you can have songs generated with Czech lyrics

Sdílejte:
Adam Kurfürst
Adam Kurfürst
19. 2. 2026 09:30
Advertisement

After generating images and videos, Google is introducing another creative tool for its Gemini app. Users can now create musical tracks using Google DeepMind’s Lyria 3 model by simply describing their idea. Just write what kind of song you want, and within seconds you’ll have a finished result, including vocals and lyrics.

How to generate your own song

Music generation in Gemini works in two ways. The first is a text description, where you simply write what the track should be about. You can specify the genre, mood, tempo, or a specific theme. Google gives the example prompt “create a funny R&B slow jam about a sock looking for its other half” – and Gemini will indeed create such a track for you.

The second option is uploading a photo or video. Gemini analyzes the content and creates a track that matches the mood of the image. For example, you can upload photos from a trip with your dog and have a song composed about your shared wanderings.

Compared to previous versions, Lyria 3 brings several improvements. Most notably, you no longer have to come up with lyrics – the model generates them itself based on your input. You also have greater control over details like style, vocals, or tempo, and the resulting recordings sound more realistic.

Also works in Czech

As you may have already noticed in the example above, the tool supports music generation in Czech as well, so you can have songs created with Czech lyrics. Google’s local branch even sent journalists a sample where an AI-generated song announces the introduction of this feature. This is an interesting demonstration of what the model can do.

Each generated track is 30 seconds long, and Gemini automatically creates cover art for it using the Nano Banana model. You can download or share the finished recording via a link.

Protection against misuse

Google embeds an invisible SynthID watermark into all generated tracks, which allows identifying AI-created content. If your prompt mentions a specific artist, Gemini takes it only as inspiration and does not create copies of existing works. The tool also checks outputs against existing tracks.

Music generation is available for users over 18 years old in both the mobile and desktop Gemini applications. Subscribers to Google AI Plus, Pro, and Ultra plans have higher creation limits.

Have you already tried generating music in Gemini?

Sources: Google, PR

About the author

Adam Kurfürst

Adam studuje na gymnáziu a technologické žurnalistice se věnuje od svých 14 let. Pakliže pomineme jeho vášeň pro chytré telefony, tablety a příslušenství, rád se… More about the author

Adam Kurfürst
Sdílejte: