Google Gemini Can Now Generate Music
Google has updated its Gemini AI platform to generate 30-second music tracks from text prompts or photos, including custom lyrics and cover art. The feature, available on the Gemini app and web, allows creative teams to rapidly prototype audio for campaigns or social content. Early agency feedback suggests it is an efficient tool for creating demo reels and scratch tracks for pitches, lowering the barrier for audio experimentation in creative workflows.
- The music generation is powered by Google DeepMind's Lyria 3 model, an advancement over the previous Lyria 2. This new version now automatically generates lyrics, a feature not present in its predecessor. - To address concerns about AI-generated content, every track created with Lyria 3 is imperceptibly watermarked with SynthID technology. Users can upload an audio file to Gemini to verify if it was created by Google's AI. - The feature is rolling out in beta for users 18 and older on both desktop and mobile versions of the Gemini app across several languages, including English, German, Spanish, and French. While available to all users, those with paid Google AI Plus, Pro, and Ultra subscriptions will have higher usage limits. - Currently, the tool is limited to creating 30-second tracks, which is suitable for social media clips and short-form video content but not for producing full-length songs. This length limitation is seen by some as a strategic move to minimize copyright concerns. - The generated music is intended for "original expression" and not for mimicking existing artists; Google has implemented filters to prevent the replication of specific artists' styles. If a famous musician is named in a prompt, Gemini is designed to use it as broad stylistic inspiration rather than a command to copy their work. - In addition to text prompts, users can upload images or videos, and Gemini will compose a track with a mood and lyrics to match the visual content. The accompanying cover art for the tracks is created by Google's Nano Banana image generation model. - Competitors like Suno and Udio have been in the AI music generation space longer and offer the ability to create longer tracks, with Suno's v5 being a popular choice for producing full-length songs. These platforms have also faced and settled copyright issues with major music labels. - Beyond the consumer-facing app, Google is integrating Lyria 3 into its Vertex AI platform for enterprise use and into YouTube's Dream Track tool for Shorts creators. This allows for the creation of custom soundtracks for marketing campaigns and video content.
