Google DeepMind Launches Lyria 3: New AI Music Generation Arrives in Gemini App

Google DeepMind on Wednesday announced Lyria 3, its latest artificial intelligence (AI) music generation model. As per the Mountain View-based tech giant, the new model is integrated into the Gemini app for Android and iOS, and is designed to help users generate high-quality music tracks using simple text prompts, leveraging multimodal generative AI. Lyria 3 is positioned as a creator-focused tool aimed at simplifying music production for social media, video, and digital storytelling.

Lyria 3 Features

Lyria 3 powers new music creation capabilities inside the Gemini app, Google DeepMind explained in a blog post. Users can generate up to 30-second original tracks by describing the desired genre, mood, tempo, or instruments. The music generation model translates text prompts into fully produced audio clips, complete with layered instrumentation and polished sound output.

Compared to the previous iterations, Lyria 3 is claimed to offer several enhanced capabilities. To begin with, it can automatically generate lyrics based on the prompt, while also providing more control over elements like style, vocals, and tempo. Apart from this, it offers improved audio quality, richer instrumentation, and better coherence across longer compositions.

Beyond text-to-track generation, Lyria 3 in Gemini allows users to create music inspired by photos and videos. For example, a sunset image or short travel clip can be used as creative input, with Lyria 3 generating a soundtrack that matches the visual tone and atmosphere. Each generated track can also include custom AI-generated cover art, making it easier for creators to package and share their work.

Google is also extending Lyria's capabilities to YouTube Dream Track, where users can experiment with AI-generated music for Shorts and other content formats. Dream Track enables creators to explore different musical styles and themes using similar generative AI technology.

To address concerns around AI-generated media authenticity, Google has embedded its SynthID watermarking technology into Lyria 3 outputs. It invisibly marks generated audio with metadata that helps identify it as AI-created content for transparency and responsible use.

The company has also indicated that the model is part of its broader push toward multimodal AI, where text, images, audio, and video generation coexist within a unified platform.

As per the tech giant, Lyria 3 is available in the Gemini app for all users over 18 years of age in English, German, Spanish, French, Hindi, Japanese, Korean and Portuguese, on desktop today. Its rollout on the Gemini app for Android and iOS will take place over the next several days. Google AI Plus, Pro and Ultra subscribers can avail of higher limits.