Google has launched Lyria 3, a fresh artificial intelligence model capable of generating original 30-second songs from text prompts, images, or videos, beginning February 18, 2026. The feature is now available within the Gemini app to users aged 18 and older, initially on desktop platforms, with a planned rollout to mobile devices in the coming days.
Lyria 3, developed by Google DeepMind, allows users to create musical tracks by simply describing a desired theme, mood, or concept. The system autonomously generates the audio, lyrics and accompanying cover art, utilizing Google’s Nano Banana image generation tool for the artwork. Users are not required to provide any pre-existing musical input or lyrics; the model creates all content based on the user’s prompt.
The launch of Lyria 3 represents an expansion of Google’s generative AI capabilities into the realm of music creation, a field increasingly competitive among AI platforms. Google previously experimented with music generation through its Dream Track project on YouTube, allowing creators to generate short background music clips for YouTube Shorts. Lyria 3 builds upon this earlier perform, offering improvements including automatic lyric generation, greater control over musical style and vocals, and more complex, realistic audio output.
Dream Track will also be updated to utilize the Lyria 3 model, though Google has not yet specified whether the update will allow for the creation of tracks exceeding the current 30-second duration. The initial release of Lyria 3 supports multiple languages, including English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese, with plans to add further language support in the future.
Google emphasizes that Lyria 3 is designed to foster original musical creation and avoid direct replication of existing copyrighted works. The system is designed to generate music *inspired* by a referenced artist or style, rather than producing identifiable copies of existing songs. Filtering systems are in place to prevent outputs from closely resembling copyrighted material, though Google acknowledges these safeguards are not foolproof and encourages users to report any potential copyright concerns.
The integration of Lyria 3 into the Gemini app aims to lower the barrier to music creation, enabling users to generate short audio tracks without requiring specialized music production software or expertise. The current 30-second limitation positions the feature primarily for use in short-form content creation, such as social media clips and video projects. The rollout is consistent with a broader industry trend of embedding generative AI tools directly into consumer platforms, expanding creative possibilities beyond text and image generation.
Google has not announced a timeline for the potential addition of support for longer track lengths or further feature enhancements.
