TEMPO.CO, Jakarta - Google introduces new capabilities to Gemini by integrating its latest music model, Lyria 3. With this update, Gemini users can create 30-second music tracks from a prompt or remix existing songs according to their preferences.
The Lyria 3 model expands Gemini's capabilities, which previously supported the creation of text, images, and videos. This feature will also be available on YouTube's Dream Track, allowing users to create detailed "backing tracks" for Shorts content.
Like other AI-based music creation tools, Gemini does not require overly detailed descriptions to produce listenable songs. Google provides an example prompt: "a cute, slow jam R&B song about a sock finding its match." However, users can also provide more specific instructions, such as setting the tempo or changing the drum style.
As quoted from Engadget, Google mentions that the Lyria 3 is an improvement over previous audio models, especially in creating more realistic and musically complex tracks. Lyria 3 also provides greater control over song components and can automatically generate lyrics.
Currently, Gemini's creations are limited to 30-second clips. However, it's possible that the duration will increase or that it will be integrated into other applications in the future. Like other Gemini AI content, songs created through Lyria 3 will be watermarked using Google's SynthID to prevent them from being easily misconstrued as human creations. Google has also launched the SynthID Detector to help identify AI-generated content.
Starting today, Google allows users aged 18 and older who speak English, Spanish, German, French, Hindi, Japanese, Korean, or Portuguese to use the music-making feature with Lyria 3.
Read: Google Releases Gemini 3.1 Pro, Records Highest Benchmark Score