Google Gemini Launches Lyria 3 AI Music Model

Google has launched Lyria 3, its latest AI music generation model, as a beta feature within the Gemini app, enabling users to create original 30-second tracks through text prompts or image uploads. The model, described in a Billboard report as designed "not to create a musical masterpiece, but rather to give you a fun, unique way to express yourself," represents the third iteration of Google’s Lyria series with significant technical refinements.

Lyria 3 eliminates the need for users to provide lyrics—a limitation in earlier versions—while granting greater control over musical elements like style, vocals, and tempo. According to Google’s official Lyria 3 documentation, the model generates "high-fidelity music" with "natural flow from note to note," supporting genres ranging from pop to Motown. Users can prompt for specific moods ("a comical R&B slow jam about a sock finding their match") or upload photos to inspire tracks, with AI-generated cover art produced via Google’s Nano Banana image model.

Google emphasizes Lyria 3’s copyright-conscious design, stating it trains only on music "that YouTube and Google have a right to use under our terms of service, partner agreements, and applicable law." The model includes filters to check outputs against existing content, though Google acknowledges these may not be "foolproof," allowing users to report potential copyright violations. All generated tracks are imperceptibly watermarked with SynthID technology to identify AI-generated audio, a feature highlighted in Google’s Gemini app announcement.

Lyria 3’s technical improvements center on structural coherence: earlier versions produced simpler tracks with limited musical complexity, while Lyria 3 handles "full instrumental arrangements" and timed lyrics. The model’s API integration, detailed in Google’s music generation documentation, supports two variants—Lyria 3 Clip (30-second clips) and Lyria 3 Pro (up to 3-minute tracks with customizable sections like verses and choruses). Lyria 3 Pro is now available in the Gemini app for paid subscribers and integrated into tools like Google Vids and Vertex AI for enterprise use.

Industry collaboration underpins Lyria 3’s development. Google partnered with musicians through its Music AI Sandbox initiative, including Grammy-winning producer Yung Spielburg, who used Lyria to score the Google DeepMind short film Dear Upstairs Neighbors. Spielburg noted Lyria 3’s "fidelity and musicality" and its role as a "versatile part of my arsenal" for refining creative ideas. Similarly, DJ Françoise K collaborated with Google to iterate on a song using Lyria 3, underscoring the model’s utility for professional workflows.

Google’s approach contrasts with competitors by prioritizing "original expression, not mimicking existing artists." When prompted to generate a song "in the style of Kate Bush," Lyria 3 responded with a track description emphasizing "gothic romanticism" and "atmospheric synthesizers" while explicitly avoiding direct imitation. This aligns with Google’s broader commitment to "enhancing human creativity, not replacing it," as stated in its Lyria 3 documentation.

Availability is currently limited to English, German, Spanish, French, Hindi, Japanese, Korean, and Portuguese, with plans to expand language support. The model is accessible via the Gemini app (for users 18+), YouTube’s Dream Track feature for Shorts soundtracks, and enterprise tools like Vertex AI. For developers, Lyria 3’s API integration enables custom music generation in applications, with sample code provided in Google’s documentation for Python, JavaScript, and other languages.

While Lyria 3’s focus on brevity (30-second tracks) may limit its use for full-length compositions, its emphasis on creative experimentation and copyright safeguards positions it as a responsible entry into AI music generation. As Google continues to refine Lyria 3 Pro for longer tracks and broader integrations, the model’s success will likely hinge on balancing user creativity with industry trust—a challenge Google aims to address through ongoing partnerships with artists and developers.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn