Google is integrating Lyria 3 into the Gemini app, allowing anyone to easily generate 30-second music clips.
Google is including Lyria 3 in the Gemini app. Lyria 3 is the latest generative music model from Google DeepMind. Based on a text prompt, a photo, or a video, users can have a short musical track composed.
The feature is available to users aged 18 and older and is first being rolled out in beta on the desktop application. The mobile version will follow in the coming days.
Description, photo, or video
According to Google, the model translates a description into a 30-second track in just a few seconds. Users do not need to provide their own lyrics; Lyria 3 generates lyrics itself based on the prompt. Additionally, the model offers control over style, tempo, and vocals. Lyria 3 is designed to enable more complex and realistic-sounding compositions than previous versions.
read also
Google may not favor Gemini in Android
Users can describe a genre, mood, or personal memory. The app then creates a matching song, either instrumental or with vocals. When you upload a photo, the same thing happens: Lyria 3 analyzes the content and creates a matching soundtrack with generated lyrics.
Illustration and soundtrack
The generated tracks automatically include album art. This is also created with AI, but using Nano Banana.
Lyria 3 is also becoming more widely available for content creators on YouTube. Through Dream Track, creators can provide Shorts with a custom soundtrack. The rollout started in the US and is expanding to other countries.
Check
All music generated in the app receives an invisible SynthID watermark. With this, Google aims to make AI-generated audio identifiable. The verification feature in the Gemini app is being expanded to audio. Users can upload an audio file to check if it was created with Google AI.
Google states that Lyria 3 is intended for simple original creation, allowing you to express yourself in an original way. The company explicitly states that the intent is not to imitate or replace existing artists.
When a user mentions a specific artist, the model uses that as a general style reference. Filters are in place to prevent generated output from looking too much like existing works. Users can report rights infringements through a provided procedure.
Lyria 3 is available in multiple languages, including English, German, French, and Spanish. Subscribers to paid AI plans receive higher usage limits.