Gemini gets a free built-in song generator

Google has expanded its AI arsenal with the launch of Lyria 3, a new music generation model integrated directly into the Google Gemini app.

The tool allows users to create 30-second, high-fidelity music tracks from simple prompts — no production skills required. From lo-fi beats and ad jingles to more layered song arrangements, Lyria 3 aims to lower the barrier to entry for AI-powered music creation.

The feature is available now on the desktop version of Gemini, with mobile rollout underway.

How to access Lyria 3

Google confirmed that Lyria 3 is free for all users aged 18 and over. To use it:

  • Open the Gemini app
  • Navigate to the “Tools” menu
  • Select the “Music” option

The system supports multiple languages, including English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese, with additional language support expected.

Paid subscribers to Google AI Plus and AI Ultra will receive higher usage limits, though core functionality remains free.

What Lyria 3 can do

Lyria 3 generates 30-second tracks based on:

  • Text prompts
  • Images
  • Audio inputs
  • Video uploads

Users can specify:

  • Genre and mood
  • Tempo
  • Vocal style
  • Instrumentation

One notable upgrade is built-in lyric generation. Users no longer need to supply their own words — the model can write them automatically based on the prompt.

Google also confirmed that companion creative tools, including Nano Banana, can generate matching album artwork for the AI-produced track.

For content creators, this means a video clip can be uploaded and paired with a custom soundtrack in seconds. A template option with guided suggestions is also available for users seeking inspiration.

Built-in AI watermarking

Every track created inside Gemini includes a SynthID watermark, marking the audio as AI-generated. Google says this measure supports transparency as generative content becomes more widespread.

Copyright scrutiny continues

AI music tools remain under intense legal and ethical debate. Musicians and publishers have accused tech companies of training models on copyrighted works without consent or compensation, leading to multiple ongoing lawsuits across the industry.

Google has not disclosed specific training datasets but stated it has been “mindful” of copyright and partner agreements while developing Lyria 3.

According to the company:

  • The system is designed for original expression
  • It does not aim to replicate specific artists
  • Prompts referencing an artist are treated as stylistic inspiration
  • Output filtering checks content against existing works

Google added that users must comply with its terms of service and generative AI policies, which prohibit intellectual property violations. Rights holders can report suspected infringements.

Another step in Google’s AI expansion

The launch reinforces Google’s steady push to embed creative AI tools across its ecosystem. With Lyria 3 sitting inside Gemini and accessible at no cost, the company is betting that casual users — not just producers — want a personal soundtrack at the click of a button.

Summary

Google has launched Lyria 3, a free AI music generator integrated into Gemini. The tool creates 30-second tracks from text, image, audio, or video prompts, complete with auto-generated lyrics and AI watermarking. While accessible to all adult users, the model enters a market already facing legal challenges over copyright and training data — debates that are unlikely to quiet anytime soon.