Lyria 3.5 makes AI music more intentional
Google’s latest Lyria music model lifts melody arrangement, lyric generation, vocal realism, and song control—so your prompts turn into richer, more deliberate tracks.
Announced for Google Flow Music · July 2026
Style, tempo, duration, and vocals travel with your prompt into the studio.
What’s new
Four upgrades that change how songs come together
Compared with the previous generation, Lyria 3.5 focuses on musicality, lyric quality, vocal performance, and how tightly you can steer a track.
Richer melody arrangement
Build more complex melodic structures that still feel coherent—hooks, progressions, and phrasing that follow the song’s intent.
Stronger lyric generation
Lyrics align more closely with your brief and emotional arc, so verses and choruses read like deliberate writing, not filler.
Finer vocal expression
Voices sound closer to real performances: clearer pronunciation, more natural emotion, and better phrasing across languages and styles.
Flexible song control
Set rhythm and duration with more confidence. Shape pacing without fighting the model for basic structure.
Musical intent
Not just more audio—music that follows what you meant.
Lyria 3.5 is positioned to better understand song intent from prompts, then deliver richer arrangements and more human-sounding vocals.
Capabilities
What Lyria 3.5 improves in practice
Use it when you need prompt-faithful songs, expressive vocals, and controllable length and groove for production, content, or creative drafting.
Melody
Complex structure, clear intent
Melody and arrangement with more depth
Lyria 3.5 supports richer, more complex melodic structures while staying closer to the story you describe in the prompt.
- Stronger harmonic and melodic continuity across sections
- Better adherence to genre and mood cues
- Useful for both full songs and production beds
Lyrics
Words that match the song
Lyrics that respect the brief
Lyric generation is a focus area for 3.5—so themes, imagery, and emotional tone track your instructions more reliably.
- Closer match between prompt themes and lines
- More natural verse-to-chorus development
- Better starting point for rewrite and polish
Vocals
Closer to a real take
Vocals with finer emotional detail
Google highlights more realistic singing, subtler emotion, and more accurate pronunciation—key for vocal-first tracks.
- More natural phrasing and breath-like delivery
- Improved clarity for multilingual or accented lines
- Stronger fit for demos, social songs, and narrative vocals
Control
Tempo and length on purpose
Rhythm and duration under tighter control
Beyond generation quality, Lyria 3.5 strengthens creative controls so you can set pace and track length more flexibly.
- More flexible tempo and pacing choices
- Duration settings that better match the use case
- Faster iteration when structure must fit a brief
Generation jump
Lyria previous gen vs Lyria 3.5
A practical view of where 3.5 concentrates improvements. Exact product surfaces and limits depend on the platform you use.
| Dimension | Previous generation | Lyria 3.5 |
|---|---|---|
| Melody arrangement | Solid generation, simpler structures more often | Richer, more complex melodic structures |
| Lyric generation | Usable drafts with uneven theme fit | Stronger alignment with prompt and song intent |
| Vocal performance | Serviceable vocals, coarser emotion | More realistic tone, finer emotion, clearer diction |
| Song control | Basic steering of form and feel | More flexible rhythm and duration control |
| Prompt following | Good baseline adherence | Better understanding of song intent from prompts |
Official demo
Hasta Que Salga El Sol
Google showcased a Lyria 3.5 demo track titled “Hasta Que Salga El Sol” (Until the Sun Rises)—an example of the model’s melodic, vocal, and arrangement direction.
Demo naming referenced from public coverage of Google’s Flow Music announcement.
Demo track
Until the Sun Rises
Hasta Que Salga El Sol · Lyria 3.5 showcase
Who it’s for
Where Lyria 3.5 shines
Video and short-form creators
Need a mood-true song or bed that follows a script beat, tempo, and runtime? 3.5’s control and arrangement upgrades help.
Podcast and brand audio
Build intros, transitions, and themed cues with clearer prompt fit and more consistent musical structure.
Songwriters and demo makers
Draft vocal songs with stronger lyric and performance quality, then refine the idea in your own production stack.
FAQ
Lyria 3.5 questions
What is Lyria 3.5?
Lyria 3.5 is Google’s updated music generation model announced for Google Flow Music. It improves musicality, lyrics, vocals, and creative control versus the prior generation.
What improved versus the previous Lyria model?
Public materials highlight four focus areas: melody arrangement, lyric generation, more realistic and expressive vocals, and stronger control over tempo and duration.
Does Lyria 3.5 follow prompts better?
Yes—Google positions 3.5 as better at following user prompts and understanding song intent, which should improve consistency between brief and output.
Is vocal quality better?
Google states that generated vocals are closer to real singing, with finer emotional layers and more accurate pronunciation.
Can I control tempo and length?
Creative control is a headline upgrade: users can set song rhythm and duration more flexibly for different formats and briefs.
Where can I try Lyria-style music on this site?
Open the Lyria3 AI music workspace from this page, write a prompt, choose style and vocal direction, and generate in the browser.
Start a track with Lyria-level musical intent
Write the mood, structure, and vocal notes you care about—then generate and iterate in the studio.