Mountain View, California — Google’s Gemini app has unveiled its latest innovation, Lyria 3, a cutting-edge music generation model positioned as the most sophisticated of its kind developed by Google DeepMind. Unlike its predecessors, Lyria 3 eliminates the need for users to supply their own lyrics, as the system now automatically crafts lyrics according to user prompts. This new feature is designed to enhance user control, allowing customization in terms of style, vocals, tempo, and overall musical complexity.
The guiding principle behind Lyria 3 is to facilitate original musical expression rather than replicate the sound of existing artists. Google has emphasized its commitment to ethical standards during the development process, ensuring compliance with copyright laws and partner agreements. When users reference a specific artist in their prompts, Gemini interprets this as broad inspiration, aiming to generate music that evokes a similar essence without directly copying.
To further secure intellectual property rights, users have the option to flag any content they believe may infringe upon their rights or the rights of others. This proactive measure highlights Google’s acknowledgment that their content oversight may not be infallible.
Within the app, a new “Music” option is available in the “Tools” menu, allowing users to create original pieces based on specific criteria like genre, mood, or personal memories. For instance, a user might request a nostalgic track that reflects fond childhood memories, resulting in a composition tailored to their request, such as an upbeat afrobeat melody celebrating cherished family moments.
Gemini also supports visual prompts; users can upload images or video clips to inspire track creation. For example, a photograph of a dog on a hiking trip could lead to a uniquely personalized musical tribute that captures the mood of that moment.
The output from Lyria 3 typically results in a 30-second musical piece, complete with custom cover art generated through a feature named Nano Banana. Users can conveniently share their creations by downloading files or generating links for easy access.
In terms of security, each track produced comes embedded with a SynthID watermark, a measure that complements existing image and video verification processes. Users can even upload audio files to ascertain whether they were crafted using Google’s artificial intelligence.
Currently, the Lyria 3 model is accessible to all users aged 18 and older in several languages, including English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese. Enhanced limits will be available to subscribers of Google AI Plus, Pro, and Ultra.
Looking ahead, Google plans to broaden the language support and improve the overall quality of music generated by Lyria 3, hinting at a future where expressive musical creation becomes even more accessible to a global audience.