Google has introduced Lyria 3, its most sophisticated generative music model to date, directly into the Gemini ecosystem. This update allows users to transform simple text descriptions or uploaded images into 30-second musical tracks, complete with custom lyrics and AI-generated cover art. Unlike previous iterations, Lyria 3 handles the entire creative process—from composing melodies to drafting verses that match the user’s specific mood or memory—significantly lowering the barrier for high-fidelity personal expression.

The model offers granular control over musical elements such as vocal style, genre, and tempo, enabling the creation of complex tracks ranging from R&B slow jams to energetic Afrobeat rhythms. To address persistent concerns within the creative industry, Google has implemented robust filters designed to prevent the direct mimicking of existing artists. Instead, the model uses artist names only as broad stylistic inspiration, ensuring that the outputs remain original while respecting intellectual property rights.

Safety and transparency are central to this rollout through the expansion of SynthID. This technology embeds an imperceptible watermark within the audio data, allowing the Gemini app to verify if a file was originally generated by Google AI. By broadening these verification capabilities across audio, image, and video, the platform aims to provide a more robust defense against misinformation and uncredited synthetic media in an increasingly multimodal digital landscape.

Google has introduced Lyria 3, its most sophisticated generative music model to date, directly into the Gemini ecosystem. This update allows users to transform simple text descriptions or uploaded images into 30-second musical tracks, complete with custom lyrics and AI-generated cover art. Unlike previous iterations, Lyria 3 handles the entire creative process—from composing melodies to drafting verses that match the user’s specific mood or memory—significantly lowering the barrier for high-fidelity personal expression.

The model offers granular control over musical elements such as vocal style, genre, and tempo, enabling the creation of complex tracks ranging from R&B slow jams to energetic Afrobeat rhythms. To address persistent concerns within the creative industry, Google has implemented robust filters designed to prevent the direct mimicking of existing artists. Instead, the model uses artist names only as broad stylistic inspiration, ensuring that the outputs remain original while respecting intellectual property rights.

Safety and transparency are central to this rollout through the expansion of SynthID. This technology embeds an imperceptible watermark within the audio data, allowing the Gemini app to verify if a file was originally generated by Google AI. By broadening these verification capabilities across audio, image, and video, the platform aims to provide a more robust defense against misinformation and uncredited synthetic media in an increasingly multimodal digital landscape.