Google Releases Lyria 3 Pro, an AI Capable of Creating Songs Up to 3 Minutes from Prompts

Google has once again released its latest artificial intelligence (AI) model for creating music, named Lyria 3 Pro. This model is an advancement from the Lyria 3 version introduced about a month ago. With Lyria 3 Pro, users can now create songs or music lasting longer, up to three minutes. Previously, Lyria 3 could only produce songs with a maximum duration of 30 seconds. In addition to enabling longer song creation, Google states that Lyria 3 Pro also offers better creative control and customisation. Through text commands (prompts), users can adjust various elements in a music composition, such as intro, verse, chorus, and bridge. This adjustment capability is provided because the Lyria 3 Pro model is said to have a better understanding of musical structure compared to its predecessor. Google emphasises that this AI music creation model is trained using data from partners as well as data permitted by YouTube and other Google services. The company also states that Lyria 3 Pro does not imitate music or songs created by specific artists. However, if users mention a musician’s name in the prompt, the system will draw general inspiration from that artist’s style to generate the song. Furthermore, every song created using Lyria 3 Pro will be embedded with an invisible watermark named SynthID. This technology serves as an indicator that the music content was created with AI assistance. Lyria 3 Pro is now being rolled out to the relevant applications, but it is only accessible to paid users. In addition, Google is bringing this model to several other services, such as the Google Vids video editing app and ProducerAI. The latter is an AI-based music production tool recently acquired by Google. In the enterprise realm, Lyria 3 Pro’s capabilities are also being integrated into Vertex AI (in public preview stage), Gemini API, and AI Studio, as reported by KompasTekno from TechCrunch.