You can now create music with Google's AI

After text, image, and video, Google adds a new string to its assistant's bow.

With Lyria 3, its latest model from Google DeepMind, Google Gemini can now generate music tracks from a simple description. This development confirms Google's ambition to transform its chatbot into a true multimodal creative studio.

Google makes no secret of its intention and wants to offer a fun tool, designed to enrich content or illustrate ideas, but without aiming for the production of commercial tracks… for now?

Short, but highly customizable tracks

While music generation was previously confined to specialized platforms, integrated directly into Gemini, it is now accessible to millions of users, including in a free version.

Lyria 3 allows the creation of musical excerpts of up to 30 seconds, with a limit which is a given. Indeed, for Google, it's more about generating an atmosphere, a ringtone, or a soundtrack for a short video than composing the next summer hit. In concrete terms, the user can describe a genre such as rap, hard rock, or pop, a time period, a tempo, specific instruments, a vocal style, and even a lyrical theme. The model is capable of automatically writing the lyrics, leveraging Gemini's text capabilities. Google also recommends a specific formula to optimize results, combining genre, mood, instruments, voice type, and lyrical subject.

From image to music…

Another advantage of the tool is its ability to generate music from an imported image or video.

The model natively supports several languages, including French, English, German, Spanish, and Japanese, but still contains some translation errors.

A Response to Platforms Like Suno

With this announcement, Google is encroaching on the territory of Suno and other players specializing in AI-generated music.

But while competitors offer longer tracks and advanced editing tools, Lyria 3 prioritizes immediacy and integration into a broader ecosystem.

Furthermore, it should be noted that all generated tracks include an invisible SynthID watermark, intended to to identify their origin and limit fraudulent use. And to go further, it is also possible to import an audio file into Gemini to check if it was generated by AI.

Finally, Google is already preparing the integration of Lyria into YouTube to allow creators to easily produce soundtracks for their videos, especially short videos. In the battle for generative creativity, music is thus becoming the new pillar of Gemini's strategy…