Google Launches Lyria 3.5: A Leap Forward in AI Music Generation
Google has made waves in the tech and music industries with the launch of Lyria 3.5, the latest iteration of its AI music-generation model. This enhanced version, introduced on July 29, promises to deliver tracks that are not only “more natural” but also feature vocals imbued with “more expression and emotion.” Positioned within Google Flow Music, the company’s AI song-creation platform, Lyria 3.5 aims to redefine how musicians and creators approach music production.
Advancements in Musicality and Vocal Quality
One of the standout features of Lyria 3.5 is its ability to produce considerably improved vocals. According to Google’s DeepMind, the AI research division behind Lyria, this version can infuse songs with greater emotional depth, achieving vocals that sound not just realistic but also nuanced. Users are set to benefit from enhanced pronunciation, allowing the AI to deliver lyrics with clarity and intention.
In terms of musicality, Lyria 3.5 enables users to explore “richer, more complex melodic structures” that resonate well with listeners. This evolution in musicality holds the promise of transforming the way musicians conceptualize and create their compositions.
Moreover, Google emphasized that the new model supports “higher quality lyrics” that adhere more closely to user prompts. Creators now have greater control over various aspects of their tracks, such as tempo and duration, directly within the Flow Music platform.
Showcasing New Capabilities
To demonstrate Lyria 3.5’s advanced features, Google released a sample track showcasing vocals reminiscent of a female pop artist. This move serves not just as a marketing strategy, but also as a testament to the model’s prowess in generating contemporary and relatable content.
However, while Google has celebrated the enhancements, details about how Lyria 3.5 was trained remain conspicuously absent from the announcement. The lack of transparency on this front raises questions within the music community about the ethical implications of AI-generated art.
The Ongoing Copyright Debate
In light of previous concerns, the announcement of Lyria 3.5 comes amid a backdrop of legal disputes surrounding Google’s training methodologies. When Google debuted Lyria 3 in February, it claimed to focus on “original expression” and made commitments to handle copyright issues responsibly. Yet, this latest announcement lacks similar reassurances.
Currently, Google is embroiled in legal challenges from a group of musicians and songwriters alleging that the AI was trained on copyrighted material from YouTube without adequate permissions or compensation. Their extensive complaint highlights the staggering amount of material purportedly used—44 million clips and 280,000 hours of music. The plaintiffs argue that Google holds an unfair advantage due to its ownership of YouTube and its rights management tools.
Google’s defense hinges on the assertion that musicians who upload to YouTube license their works, thereby granting Google a wide-ranging usage license. However, the musicians contend that this arrangement should not automatically extend to training AI models like Lyria.
The Evolution of Lyria from Concept to Reality
The Lyria model lineage began its journey in November 2023, alongside initiatives such as Dream Track, a collaboration that used AI voice clones of notable artists. Subsequent iterations, like Lyria 2 and Lyria 3, have expanded the platform’s capabilities, reaching into features like YouTube’s “Speech to Song” tool.
Lyria 3.5’s unveiling has coincided with significant developments in Google’s Flow Music platform, which originated from an open-source project known as Riffusion. This viral sensation of 2022 metamorphosed into ProducerAI in 2025 before being acquired and rebranded by Google in early 2026.
Competitive Landscape and Industry Reactions
As Google pushes forward with Lyria 3.5, other platforms are also evolving. With recent enhancements in Flow Music—including song generation, editing capabilities, and stem splitting—the competition is heating up between Google and rivals such as Suno and Udio.
Partnerships also reflect industry shifts, with music distributor Believe recently collaborating with Google to integrate Flow Music into its services. This collaboration aligns with the broader trend of utilizing generative AI to foster human creativity—a sentiment echoed by Believe’s founder, Denis Ladegaillerie.
Amid this rapid advancement, major music labels are navigating their own legal landscapes, with Universal Music Group and Warner Music Group settling disputes with Udio. Conversely, Sony Music continues to litigate, adding further complexity to the conversation around copyright and AI in music.
As Lyria 3.5 takes its place in an ever-evolving music landscape, it stands as a pivotal moment for both Google and the future of AI-driven music creation. The ongoing dialogue regarding copyright, transparency, and artistic integrity will likely shape the model’s trajectory and reception in the months to come.