Google launches Lyria 3.5 inside Flow Music with improved vocals and control

Google launches Lyria 3.5 in its Flow Music app on July 29, 2026, the company announced, rolling out what it describes as its “newest music generation model” with improvements to musicality, lyrics, vocals and user control. The update arrives inside Flow Music and, according to Google, aims to help creators “craft richer tracks” by making melody, lyrics and vocal performance more expressive and easier to edit.

The rollout emphasises workflow as much as output quality: Google says Lyria 3.5 adds tighter tempo and duration controls, section-level editing and better prompt adherence for lyric structure — features intended to move users from single-shot generation to iterative song construction.

What Google says Lyria 3.5 does

Google’s blog states plainly that “We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control.” The post claims the model can “create richer, more complex melodic structures that sound more natural,” “generate higher quality lyrics with improved prompt adherence and structural awareness,” and “bring more expression and emotion to your songs with more realistic and emotionally nuanced vocals, plus improved pronunciation.” It also highlights user-facing controls: tempo and duration adjustments, and section-level edits that let users expand short motifs into full songs without starting over.

Those features map onto Flow Music’s existing toolbox. The broader Flow Music product had already supported song generation, editing, stem splitting and track extension under Lyria 3 Pro; this release positions Lyria 3.5 as an iterative, creator-focused upgrade rather than a first-of-its-kind service.

How Lyria 3.5 compares to rivals such as Suno

Market watchers frame Google’s move as direct competition with prompt-to-song systems like Suno, where simplicity and speed are selling points. Coverage comparing the two suggests Google’s differentiator is stronger editing and structural controls — tempo, duration and section-level manipulation — rather than raw generation speed. That matters for users who want to craft multi-part songs or refine lyrics over multiple passes rather than accept a single generated clip.

But independent verification is thin. Outside coverage notes Google did not publish benchmark numbers or watermarking details with the launch, making it hard to quantify how much Lyria 3.5 improves over Lyria 3 Pro or competitors. SQ Magazine and others flagged the lack of hard evaluation data, characterising the post as product messaging focused on workflow rather than peer-reviewed comparison.

What’s missing from the announcement

The blog post and tech write-ups do not include side-by-side audio evaluations, objective metrics for lyrical coherence or third-party vocal quality tests — omissions that matter to producers and publishers who will weigh copyright, attribution and commercial usability. Google also did not disclose pricing changes or whether parts of Flow Music powered by Lyria 3.5 will sit behind paid tiers; some secondary pieces speculate on pricing but the company’s post does not confirm fees or free-use limits.

Skeptics caution that without benchmarks and transparent evaluation, claims about “more realistic and emotionally nuanced vocals” remain marketing language until subjected to independent tests. A YouTube reading and several tech outlets summarised the feature set, but none supplied reproducible listening tests tied to Lyria 3.5’s code or datasets.

Flow Music’s expansion has been steady: Google first added generation and editing features under earlier Lyria versions and gradually built a creator workflow. Releasing Lyria 3.5 now appears tactical — consolidating those tools while pitching Flow Music as a one-stop environment for iterative songwriting and production.

Google’s announcement sets the next point to watch: whether independent reviewers publish comparative audio benchmarks and whether Google follows with technical documentation — dataset provenance, evaluation methodology and any watermarking or safety measures. Those disclosures will determine if Lyria 3.5 is a clear technical step forward, or chiefly a more usable interface layered atop existing capabilities.