Google has launched Lyria 3.5, the latest version of its AI music-generation model, promising tracks that “sound more natural” and vocals with “more expression and emotion.”
The model rolled out on Wednesday (July 29) inside Google Flow Music, the company’s AI song-creation platform.
Google DeepMind, the tech giant’s AI research division, said Lyria 3.5 delivers “significant advancements across musicality, lyrics and vocal quality.”
On vocals, Google said the model can “bring more expression and emotion to your songs with more realistic and emotionally nuanced vocals, plus improved pronunciation.”
On musicality, Google said users can create “richer, more complex melodic structures that sound more natural.”
The company also cited “higher quality lyrics” with better prompt adherence, and easier control over the tempo and duration of a track in Flow Music.
To showcase the capabilities of Lyria 3.5, Google released the following track, which features the vocal style of a female pop artist.
What Google‘s announcement did not say is how Lyria 3.5 was trained.
The blog post also made no mention of SynthID, the watermarking technology Google embeds in its AI-generated tracks, nor of the copyright framing that accompanied earlier Lyria launches.
When Google launched Lyria 3 in February, it described that model as “designed for original expression, not for mimicking existing artists.”
Google said at the time that it had been “very mindful of copyright and partner agreements,” and that it had sought to “develop this technology responsibly in collaboration with the music community.”
The Lyria 3.5 announcement carried no equivalent language.
Unlike the Lyria 3 launch, which was bylined by two Google product managers, it also carried no named author.
Google has never specified what music it used to train the Lyria model line.
The company has said its training uses material that it and YouTube “have the right to use” under their “terms of service, partner agreements, and applicable law.”
Google‘s position is now being tested in court.
In March, a group of independent musicians and songwriters sued Google, accusing it of training Lyria 3 on copyrighted recordings pulled from YouTube without permission or payment.
Their 118-page complaint claimed Google had copied “at least 44 million clips and 280,000 hours of music.”
The plaintiffs argued that Google holds “structural leverage” because it owns YouTube and runs Content ID, the platform’s rights-management system.
Google moved to dismiss the case in June.
Its lead argument is not a fair use defense, but that the musicians licensed their recordings to YouTube – and to Google – when they uploaded them.
Each plaintiff, Google said, granted YouTube and its “Affiliates” a “worldwide, non-exclusive, royalty-free, sublicensable and transferable license” to use their uploaded content.
The lawsuit, Google added, rests on the “unsupported hypothesis that Google trained on their specific works.”
Google DeepMind first unveiled Lyria in November 2023, in partnership with YouTube.
It arrived alongside Dream Track, a YouTube Shorts experiment that generated tracks using AI voice clones of artists including Charlie Puth, T-Pain and Alec Benjamin.
A later model, Lyria 2, powered YouTube‘s “Speech to Song” tool for turning dialogue into music.
Lyria 3 followed in February 2026 inside Google’s Gemini app, generating 30-second tracks from text prompts or images.
Flow Music, where Lyria 3.5 now sits, began as Riffusion, an open-source project that went viral in 2022, and launched as ProducerAI in 2025.
Google acquired the platform in February and rebranded it Flow Music in April.
In May, Believe partnered with Google to offer Flow Music to artists across the distributor and its self-release arm, TuneCore.
“The adoption of Gen-AI is going to enhance human creativity,” Believe founder and CEO Denis Ladegaillerie told MBW at the time.
Last week, Google added song generation, editing and stem splitting to Spaces, a “vibe coding” tool inside Flow Music.
The additions move Flow Music closer to rivals Suno and Udio.
Universal Music Group and Warner Music Group both settled their copyright lawsuits against Udio, agreeing to build licensed AI music platforms. Sony Music, however, has not settled with Udio and continues to litigate, filing a fresh suit on July 20 that names more than 30,000 recordings. Warner has also settled with Suno, while Universal and Sony‘s cases against Suno remain ongoing.Music Business Worldwide
