Google Lyria 3.5 Brings Music Controls and Templates to Gemini Creators
Lyria 3.5 expands Google Gemini's music-generation tools with configurable song structure, templates, vocal options, and API access for creators and developers.
Google has released Lyria 3.5 across the Gemini app and Gemini API, adding more direct control over how AI-generated music is shaped. The update gives creators options for duration, tempo, genre, vocals, lyrics, and song structure, while a template gallery offers starting points for faster experimentation. For businesses producing video, social, product, or presentation content, the practical change is greater control over generating music that better fits a defined format and creative brief.
Google describes Lyria 3.5 as an upgrade in musicality and vocal fidelity. In its official Gemini announcement, the company says users can create vocal or instrumental tracks, choose genres and vocal styles, adjust track length, and begin with templates. The release is also available through the Gemini API, opening a route for developers to build music generation into their own tools and workflows.
What Lyria 3.5 changes for music generation
The central improvement is not simply that Gemini can make music. It is that users have more ways to specify the result. Rather than relying solely on a broad text prompt, Lyria 3.5 supports instructions that can guide a track's progression through sections such as verses, choruses, and bridges. The Gemini API documentation also supports text and image inputs through the Interactions API, allowing an image to act as creative context for music generation.
More control over the finished track
Google's published materials identify several controls and output options that are relevant to both hands-on creators and product teams:
- Tempo and duration controls help users align music with the pace and runtime of a project.
- Vocal and instrumental outputs allow a choice between a lyric-led track and background music.
- Lyrics and multilingual support broaden the types of vocal music that can be generated.
- Structured prompting can guide musical sections, including verses, choruses, and bridges.
- Templates provide a quicker way to start a project when a user needs direction rather than a blank prompt.
- 44.1 kHz stereo audio is the stated output format for Lyria 3.5.
In the Lyria 3.5 user interface, Google says tracks can run up to about three minutes. The API documentation describes the Pro model's output as lasting a couple of minutes, so developers should treat the interface and API guidance as related but not identical implementation contexts. That distinction matters when designing a workflow around a fixed video length or a repeatable content format.
| Area | Gemini app and Lyria UI | Gemini API |
|---|---|---|
| Starting point | Text prompts and a Lyria template gallery | Text prompts and image inputs through the Interactions API |
| Track length guidance | Up to about 3 minutes | A couple of minutes for the Pro model |
| Creative controls | Genre, vocals, duration, and templates | Duration guidance and prompt-driven song structure |
| Typical use | Direct music creation in Gemini | [Building music generation into software or workflows](https://scalevise.com/services/api-system-integrations) |
Availability, pricing, and provenance
Lyria 3.5 is available across the Gemini app and Gemini API. Google also identifies Flow Music for artists and AI creatives, AI Studio integrations, and Google Vids for developers among the places connected to the release. These are distinct routes into the model rather than a single uniform product experience, so available controls may depend on the interface being used.
For Gemini API usage, Google's pricing documentation lists $0.08 per full song for Lyria 3.5 and $0.04 for a 30-second clip. Those listed per-song prices give teams a clearer basis for testing than an open-ended production estimate. They do not, however, remove the need to assess generated results for fit with the intended campaign, audience, and creative direction.
Google also says Lyria 3.5-generated content is watermarked with SynthID, its system for identifying AI-generated media. That provenance signal is relevant for teams that need to distinguish generated assets within their own creative process. It should not be confused with a statement about whether any particular generated track is suitable for every commercial, editorial, or brand use case.
Why templates and structure matter to business content
The most useful application of Lyria 3.5 for many teams is likely to be iterative production. A marketer creating a short social video may need upbeat instrumental audio with a specific runtime. A training team may need a more restrained background track for a product walkthrough. A creative agency may want to prototype several directions before deciding whether a concept deserves further production work.
Templates can reduce the time needed to form an initial prompt, while duration and tempo controls can make a generated track easier to test against an existing edit. Structured prompts add another layer of direction when a project requires an opening, recurring hook, or defined transition. These controls do not guarantee a production-ready result, but they can make experimentation more deliberate and repeatable.
The API route has a different implication. Developers can evaluate whether music generation belongs inside an existing content system, campaign tool, or internal production workflow. The supplied documentation supports multimodal prompting and full-song structure, but it does not establish how every third-party editor, design platform, or marketing system will integrate with Lyria. Any implementation will depend on the team's own application requirements and the Gemini API capabilities available to it.
For businesses that want to turn new AI capabilities into dependable processes, the important question is not only whether a model can generate an appealing track. It is whether staff can reliably move from a brief to a usable asset with appropriate review points. Scalevise's AI workflow automation service can help map that process, connect AI outputs to the tools your team already uses, and reduce repetitive handoffs without losing human creative oversight. Discuss an AI automation project with Scalevise.
Frequently Asked Questions
What is Google Lyria 3.5?
Google Lyria 3.5 is a music-generation model available across the Gemini app and Gemini API. Google describes it as providing improved musicality and vocal fidelity, along with more controls for creating vocal or instrumental tracks.
How long can Lyria 3.5 tracks be?
Google says the Lyria 3.5 interface can generate tracks up to about three minutes. Gemini API documentation describes the Pro model's output as lasting a couple of minutes.
Can Lyria 3.5 generate songs with vocals and lyrics?
Yes. Google says Lyria 3.5 can create vocal or instrumental outputs, generate lyrics, and support multiple languages. Users can also choose vocal-related options in the Gemini experience.
How much does Lyria 3.5 cost through the Gemini API?
Google's Gemini API pricing documentation lists Lyria 3.5 at $0.08 per full song and $0.04 per 30-second clip.
Does Lyria 3.5 label AI-generated music?
Google says Lyria 3.5-generated content includes a SynthID watermark, which is intended to provide provenance for AI-generated media.
Conclusion
Lyria 3.5 makes Gemini's music generation more controllable through templates, timing settings, vocal options, structured prompts, and API access. For creators and teams, its value lies in making music experiments easier to direct toward a specific format or creative purpose. The API pricing and cross-product availability also give developers a defined starting point for evaluating where generated music could fit into existing content workflows.