Google Brings Its Lyria 3.5 Music Model to Every Gemini User
Google has made its Lyria 3.5 music model available in the Gemini app and API, enabling the generation of full songs at 44.1kHz from text or images with SynthID watermarking. This model builds upon previous versions, including Lyria 3 Clip, which always produces 30-second clips, and Lyria 3.5, capable of creating tracks lasting a couple of minutes.
Key Features:
- Global Availability: Access is available to all Gemini users worldwide on both web and mobile platforms.
- Song Generation: The model generates songs with verses, choruses, and bridges.
- API Functionality: The API produces 44.1kHz stereo audio from text or images, allowing developers to provide up to ten images, write lyrics using section tags, and set timestamps for instrument entry.
- Safety Filters: Safety measures block requests for specific artist voices and the generation of copyrighted lyrics, aligning with a recent European court ruling against Suno.
Legal Context:
A Munich Regional Court ruled on July 31st that Suno was liable for infringing outputs, asserting jurisdiction over training conducted outside the EU. This decision sets a precedent regarding memorized works and training data stored in Germany, even if the training itself took place outside the EU.
Market Saturation:
The market for AI-generated music is already competitive, with Deezer leading in new uploads, despite Spotify’s relatively quiet stance. Approximately 90,000 fully AI tracks are added to Deezer daily, accounting for a small but growing percentage of actual listening time.
Watermarking:
Google and Suno have both implemented watermarking techniques (SynthID) to address copyright concerns, aiming to make it difficult to identify and remove infringing content.
By Ana Maria Constantin, an expert in digital marketing, product management, and branding & identity.