What Google Lyria 3.5 changes in the music generation model
Google launched Lyria 3.5, its newest music generation model, in Google Flow Music on July 29, 2026. The launch post says the model brings significant advancements in musicality, lyrics and vocal quality, and that it was rolling out that day in Flow Music. Google lists four improvements:
| Area | What Google says |
|---|---|
| Musicality | Melodic structures become richer and more complex, and sound more natural. |
| Lyrics | Higher quality, with better prompt adherence and structural awareness. |
| Vocals | More expression and emotion, more realistic and nuanced vocals, and improved pronunciation. |
| Creative control | Tempo and duration of the outputs are easier to control. |
The Google DeepMind Lyria page describes Lyria as Google DeepMind's most advanced music generation model, now up to 3 minutes long. The page's model list describes Lyria 3.5 as the latest model, with improved vocals, musicality and variable song lengths.
We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control. ／ Improved musicality: Create richer, more complex melodic structures that sound more natural. Enhanced lyrics: Generate higher quality lyrics with improved prompt adherence and structural awareness. Improved vocals: Bring more expression and emotion to your songs with more realistic and emotionally nuanced vocals, plus improved pronunciation. Creative control: More easily control the tempo and duration of your outputs. — From the launch announcement and the list of what is newView official source →
Our most advanced music generation model, now up to 3 minutes long ／ Our latest model with improved vocals, musicality, and variable song lengths — From the page header and the model family listView official source →
How long can a Lyria track be?
Google DeepMind's Lyria page says tracks run up to 3 minutes, and you can prompt for the length you need. The page's examples are a quick 60-second clip, a half-length track and a full 3-minute song, and it says a track can match a project's exact duration.
The page presents this as a feature of Lyria, its current model. The Lyria 3.5 launch post itself talks about controlling the tempo and duration of outputs.
Generate cohesive tracks matching your project's exact duration — whether you need a quick 60-second clip, a half-length track, or a full 3-minute song. — From the section on prompting for lengthView official source →
Can Lyria turn an image into a track?
Yes. The Google DeepMind page says you upload an image and ask Lyria to transform it into a custom high-fidelity track.
Upload an image and ask Lyria to transform it into a custom high-fidelity track. — From the section on composing with imagesView official source →
Lyria 3.5 in Google Flow Music: access, tiers and regions
Google's July 29, 2026 post says Lyria 3.5 is rolling out in Google Flow Music, where the stated aim is to help people create songs they love. In Google's May 19, 2026 wording, Flow Music has always let people create, iterate on and share studio-quality songs.
The rollout sentence in the Lyria 3.5 post names Flow Music. The DeepMind Lyria page, which covers Lyria in general, links to both Google Flow Music and Gemini under its "Try Lyria" section.
The access conditions are in Google's May 19 post, which ends with a note that features may vary by Google AI subscription tier, platform (web versus mobile) and region, plus an 18+ marker. The same post says Omni Flash is available to all Google AI subscribers. It also says that as of May 19, 2026, the Flow Music app was available on iOS, with Android coming soon.
We’re rolling it out today in Google Flow Music, where we want to help you create songs you love, with creative control. — From the opening paragraphView official source →
Google Flow Music has always allowed you to create, iterate and share studio-quality songs. ／ For those 18+, the Flow app is available on Android in beta (iOS coming soon), and the Flow Music app is available now on iOS (Android coming soon). ／ Features may vary by Google AI subscription tier, platform (web vs. mobile), and region. 18+. — From the Flow Music section, the app availability paragraph and the footnoteView official source →
Try in Google Flow Music ／ Try in Gemini — From the Try Lyria sectionView official source →
AI music editor tools in Flow Music: edits and covers
The tools below come from Google's May 19, 2026 Flow Music update, which came out before Lyria 3.5. Google introduced them in the same post as new agents and mobile apps for Flow and Flow Music.
| Tool | What Google says |
|---|---|
| Section editing | Highlight any part of a song and change it without altering the rest of the track. |
| Lyrics and beat drop | Instantly rewrite or translate lyrics, or restyle the beat drop. |
| Sampling | Sample a specific section of a song and extend it in a completely different direction. |
| Covers | Transform the style of a full song made in Flow Music while keeping the original melody and structure. |
| Music videos with Omni Flash | Work with the agent to direct shareable music videos, guiding styles, subjects and scenes to match the narrative and pacing of the track. |
If a song is finished except for one verse, you can highlight that verse and rewrite or translate its lyrics without touching the rest of the track. Covers is Flow Music's AI music cover feature, and Google's example is a "lo-fi study" version of a favorite playlist, where the style of the full songs changes but their melody and structure stay. Omni Flash is the model behind the music video option, and Google says it is available to all Google AI subscribers.
Now you can highlight any part of a song and make changes. Instantly rewrite or translate lyrics, restyle the beat drop and more — without altering the rest of your track. Or you can sample a specific section of one of your songs, and extend it into a completely different direction. ／ Transform the style of full songs you make in Google Flow Music, while keeping the original melody and structure. For instance, try creating a “lo-fi study” version of your favorite playlist. ／ Our newest model is also coming to Google Flow Music, allowing you to work conversationally with the agent to direct shareable music videos. With Omni Flash you can guide the styles, subjects and scenes to match the narrative and pacing of your track. This is available to all Google AI subscribers. — From the section-by-section editing, covers and music video itemsView official source →
SynthID watermark on AI generated music, and what to check
Google DeepMind says on its Lyria page that all of its tracks are imperceptibly watermarked with SynthID technology. SynthID, in the page's words, is what allows you to detect whether music has been created or edited using AI. For the same watermark on another vendor's audio, SynthID Watermark on ChatGPT Audio: What It Proves covers how it applies to ChatGPT.
The page also states a limit. Lyria can create a wide range of music, but Google is still working on improving key capabilities. Google's advice is to always carefully check that the tracks you create fit your vision.
All of our tracks are imperceptibly watermarked with SynthID technology, allowing you to detect whether music has been created or edited using AI. ／ While Lyria can create a wide range of music, we’re still working on improving key capabilities. You should always carefully check that the tracks you create fit your vision. — From the notes on SynthID watermarking and on current limitationsView official source →
FAQ
- What improved in Google Lyria 3.5?
- Google names four areas: musicality, lyrics, vocals and creative control. Melodic structures are richer, lyrics follow prompts more closely, vocals carry more expression and better pronunciation, and tempo and duration are easier to control. The launch post describes these in words and gives no figures.
- Does access to Flow Music features depend on tier, platform or region?
- Yes, according to a note in Google's May 19, 2026 Flow update. Features may vary by Google AI subscription tier, platform (web or mobile) and region. The same post says Omni Flash is available to all Google AI subscribers.
- How can you detect whether a Lyria track was made with AI?
- Google DeepMind says all of its Lyria tracks are imperceptibly watermarked with SynthID. That watermark is what lets you detect whether music has been created or edited using AI.
Related articles
Reddit to require logins as search referrals fallReddit will test a logged-in-only site outside the US in Q4. Search referrals fell over 1M a week in October; its CEO says Google's AI Mode drives no traffic.
Gemini create files: edit PDFs and docs in chatGemini create files: docs, PDFs, slides and sheets now open in the chat as previews you can edit. Phones edit only PDFs; the web handles the rest.
What is Genie 4? Google's world model with audioGenie 4 is Google DeepMind's world model that adds ambient sound and live dialogue to explorable worlds. Only testers have it; Ultra subscribers are next.
Google Deep Review: what the AI paper checker doesGoogle Deep Review checks papers for logic gaps, math errors and weak evidence. Two reviews a week are free, and its predecessor caught 89.7% of known errors.