Gemini Now Enables Music Creation With Lyria 3: Transform Your Tunes Today!

Google has added a new feature to the creative capabilities of its AI assistant. Following the generation of images and videos, the Gemini app now incorporates music creation, powered by DeepMind’s latest audio generation model, Lyria 3. This feature, launched in beta on February 18, 2026, is available to all Gemini users.

Users can simply describe the desired music piece (genre, mood, theme) and within seconds receive a 30-second track complete with automatically generated lyrics and a cover designed by the Nano Banana model.

Capabilities of Lyria 3 in Gemini

Within the Gemini application, a Create Music button has been added, similar to the Create Image CTA. There are two creation modes available:

• Text to Music: This mode allows users to write a prompt describing the desired music piece. Users can specify five categories of creative parameters: genre and era, tempo and rhythm, desired instruments, type of voice including gender, timbre, and range, and finally, lyrics, either by simply describing a theme to let the automatic generation take over, or by providing their own structured text with tags like [Verse 1] or [Chorus]. A gallery of predefined templates is also available as a starting point.
• Image and Video to Music: This second mode allows users to upload a photo or video. Gemini then analyzes the visual content and composes a song with lyrics that match the atmosphere of the image or video.

For each generated track, Gemini produces a 30-second title accompanied by lyrics and a cover generated by the Nano Banana image model. All of this is downloadable and shareable via a link.

Compared to previous iterations of the Lyria model, this third version can automatically generate lyrics from the prompt, without the need for users to write them themselves. Google states that the tracks are also more realistic and musically complex.

We had the opportunity to test the feature with automatic lyric generation (also available in French):

Availability and Safeguards

The music generation feature is now available on desktop and will be rolled out on mobile in the coming days, according to Google. At launch, the feature supports eight languages: English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. Subscribers of Google AI Plus, Pro, and Ultra have higher usage limits. Additionally, Lyria 3 is also integrated into Dream Track on YouTube, a tool for creating soundtracks for Shorts, which is now available beyond the United States.

As for safeguards, Google assures that the system is designed for original creation and not for imitating existing artists. Mentioning an artist in the prompt will only serve as stylistic inspiration, according to the company. Filters compare the generated tracks to the existing catalog to detect any potential similarities. All created titles are marked with a SynthID watermark, DeepMind’s technology for identifying AI-generated content. Google has also added a new verification feature to Gemini, which allows users to upload an audio file and ask the assistant if it was generated by a Google AI.

Music generation with Lyria 3 is intended for original expression, not for imitating existing artists, Google asserts.