Google and Apple Add Generative Music AI to Apps
Google has launched a music creation tool within its Gemini AI assistant, powered by the Lyria 3 model from DeepMind. The feature allows users to generate 30-second audio tracks from text prompts, images, or videos. Apple has also introduced music-focused generative AI features, signaling the mainstream adoption of generative audio in consumer applications.
- Google's Lyria 3 model, developed by DeepMind, automatically generates lyrics and allows for control over style and tempo, creating more musically complex tracks than its predecessor. All tracks generated are imperceptibly watermarked with SynthID technology to identify them as AI-created. - Apple’s AI features are integrated into its professional music production software, Logic Pro, and include "Session Players" (virtual AI-powered bass and keyboard players), "Stem Splitter" for separating audio into distinct parts, and "ChromaGlow" for adding sonic warmth. - For commercial use, platforms like Suno, Udio, and Soundful offer royalty-free licenses, allowing freelancers and agencies to create custom audio for branding, video marketing, and podcasts without copyright concerns. - The new tools are part of a larger trend of making audio an API-driven asset; Google's Lyria 3 is also accessible via the Vertex AI API, enabling programmatic music generation for scalable content pipelines and workflow automation. - While Google's Gemini tool is positioned for more casual creative expression, it also integrates with YouTube's Dream Track for Shorts creators, signaling a path for broader content creation applications. Apple's upcoming "Playlist Playground" in Apple Music will use AI to generate 25-song playlists from text prompts. - Both companies are addressing copyright concerns differently; Google states Lyria 3 was trained on music it has the right to use and has filters to avoid mimicking artists, while Apple's Logic Pro tools focus on augmenting a creator's own process. - Competing services like Udio, founded by ex-Google DeepMind researchers, and Suno are also rapidly advancing, offering features like full song generation, multilingual vocals, and the ability to extend or remix tracks. - To prevent misuse, Google limits the number of tracks free users can generate daily (10) and provides higher caps for paid subscribers, while also generating custom cover art using its Nano Banana image model.
