Gemini Rolls Out Built-In AI Music Creation

Google is quietly expanding Gemini’s creative toolkit. The app now includes native music generation powered by DeepMind’s Lyria 3, allowing users to produce short songs with vocals and lyrics directly from simple prompts.

The feature is entering global beta for adult users and signals a clear shift: music generation is no longer tucked inside experimental labs or developer consoles. It is moving into the mainstream Gemini experience.

The New “Create music” Option

Inside Gemini’s interface, a new entry point labeled Create music lets users generate a complete 30-second track in one pass.

What the system produces:

• Instrumental backing track
• Auto-written lyrics
• Synthetic vocals
• Basic cover artwork

The interaction stays conversational. Instead of knobs and timelines, users steer the result by describing mood, genre, or theme in plain language.

Visual prompts are also supported. Uploading a photo or short video allows Gemini to compose music that aligns with the scene’s emotional tone.

What Actually Changed Compared to Before

Google has experimented with AI music for years, but earlier efforts largely lived in research demos and developer tools. For most users, there was no direct path inside Gemini to generate songs.

This update changes the access layer more than the core idea.

Previously:

• MusicLM and earlier Lyria models were mostly experimental
• Many features sat inside Vertex AI or limited previews
• Consumer access required workarounds

Now:

• Music generation sits directly inside Gemini
• The workflow is prompt first, friction low
• Outputs include vocals and lyrics by default

The shift is less about capability appearing overnight and more about Google finally productizing it.

Availability Snapshot

The rollout is gradual and still labeled beta.

Current access conditions

• Available to Gemini users aged 18 and older
• Rolling out globally on web and mobile
• Multiple languages supported at launch

Initial language set

English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese.

Google has not yet outlined a timeline for longer compositions or full public release.

Under the Hood: Lyria 3’s Role

Lyria 3 is positioned as DeepMind’s most advanced music model so far. Its job inside Gemini is focused and constrained: generate short, coherent tracks that feel musically structured rather than purely experimental.

Notable capabilities include:

• Text, image, and video to music conversion
• Built-in lyric writing
• Native vocal synthesis
• Natural language style control

Track length is currently capped around 30 seconds, which suggests the feature is optimized for quick creative outputs rather than full song production.

Quiet Expansion Across Google’s Media Stack

The Gemini feature does not exist in isolation. Google is threading Lyria 3 through multiple products.

Developer layer

Lyria 3 is available via the Gemini API and RealTime music endpoints. Developers can stream generated audio and embed dynamic soundtracks into apps or interactive experiences.

YouTube creators

The same model is powering Dream Track, YouTube’s AI music tool, which is expanding beyond its earlier limited release.

YouTube Music

Separately, YouTube Music is testing AI Playlist generation using Gemini models to build playlists from text or voice prompts.

Taken together, the strategy appears coordinated: music generation is becoming infrastructure, not just a novelty feature.

Guardrails and Attribution Controls

AI music tools remain under heavy scrutiny, and Google is clearly trying to get ahead of that conversation.

Key safeguards:

• Artist name prompts are treated as stylistic guidance rather than direct imitation
• Output similarity checks aim to reduce close matches with existing songs
• All generated tracks carry SynthID watermarking

SynthID embeds an inaudible marker that allows later identification of AI-generated audio. Gemini is also expected to gain the ability to detect this watermark in uploaded files.

These measures are meant to address copyright and provenance concerns, though real-world effectiveness will likely be watched closely.

Where This Leaves the AI Music Race

By placing music generation directly inside Gemini, Google is moving into more direct competition with standalone AI music platforms.

The positioning is clear:

• Short-form, fast creation for everyday users
• Built-in tooling for YouTube creators
• API access for developers building audio experiences

What remains less clear is whether the current 30-second format will satisfy serious music creators or primarily serve casual and social use cases.

Key Takeaway

Google’s Lyria 3 rollout inside Gemini marks a structural shift rather than a simple feature drop. Music generation is now embedded alongside text, image, and video creation inside the company’s flagship AI assistant.

The beta remains intentionally lightweight and short-form, but the direction is unmistakable. Gemini is evolving into a multimodal creative hub, and AI music is now part of the core stack rather than an experimental side project.