Lyria 3
Explore new genres, play with soundscapes, and find musical inspiration.
Slide 1 of 4
Lyria 3
Our most advanced music generation model yet
Compose with images
Upload an image and ask Lyria to transform it into a custom high-fidelity track.
Direct the details
Compose with freedom and technical control. Define realistic vocal styles and acoustic preferences to find the sound you’re looking for.
Export professional-grade audio
Background ambience or mainstage anthem – produce crisp, clear tracks ready for your projects.
Explore the latest
Lyria 3 helps you express, explore, and experiment with high-fidelity music, using prompts to create tracks with natural flow from note to note.
Model family
Fine-tune the details and generate professional-grade audio with Lyria. Our family of music generation models can create clips and tracks, or deliver a constant stream of music.
Lyria RealTime
Designed for real-time, interactive music generation.
Magenta RealTime
Our Lyria open model explores the role of machine learning as a tool in the creative process.
Your musical collaborator
Lyria 3 is our most advanced AI music generation tool yet. We’ve developed it with input from producers and musicians so it understands musicality – from rhythm to arrangement.
Describe your track, then dial in the details you want. Lyria 3 handles the complexity, putting new musical possibilities at your fingertips.
Slide 1 of 5
Create cohesive tracks
Turn quick ideas into high-fidelity music that flows naturally from start to finish.
Explore global languages and genres
Generate vocals in different languages, and create music spanning genres from pop to funk to Motown.
Compose with images
Upload an image and ask Lyria to transform it into a custom high-fidelity track.
Dial in the details
Compose with freedom and technical control. Define realistic vocal styles and direct the details to find the sound you’re looking for.
Export professional-grade audio
Background ambience or mainstage anthem – produce crisp, clear tracks ready for your projects.
Creating your prompts
Keep it simple with a rough idea of what you want – like ‘an upbeat birthday tune’. Or describe the details for more control, like tempo and dynamics. Don’t forget to experiment as you go.
Powerful composition, endless exploration
Create pieces in countless styles using Lyria 3.
See it, hear it
Lyria 3 takes images and turns them into music.
Slide 1 of 8
Lyrics and vocals
Write your own lyrics, or generate them around specific subjects or themes. Then express them with realistic, natural, and nuanced vocals.
Slide 1 of 4
Safety
We believe AI should enhance human creativity, not replace it. We’re working directly with artists to gather feedback and shape guardrails that address their concerns. By doing this, we ensure this tool empowers creators and helps them realize the possibilities of AI in their work.
We use extensive filtering and data labeling to minimize harmful content in datasets and reduce the likelihood of harmful lyrics. Audio generation in Lyria has our latest privacy and safety features. All of our tracks are imperceptibly watermarked with SynthID technology, allowing you to detect whether music has been created or edited using AI.
Limitations
While Lyria 3 can create a wide range of music, we’re still working on improving key capabilities. You should always carefully check that the tracks you create fit your vision.
Try Lyria
Gemini
Supercharge your creativity and productivity
Dream Track
Create an AI-generated soundtrack in Shorts
Acknowledgements
Lyria 3 was developed by Ágoston Weisz, Andrea Agostinelli, Andrea Debellis, Calin Cruceru, Cătălina Cangea, Cristina Nader Vasconcelos, Eyal Rozenberg, Gladys Tyen, Hongliang Fei, Jason Baldridge, Jeff Chang, Jingwei Zhou, Jon Weinman, Louis Owen, Matej Kastelic, Mauricio Zuluaga, Mauro Verzetti, Michael Dooley, Moon Park, Myriam Hamed Torres, Nick Zammuto, Norman Casagrande, Ondrej Skopek, Oscar Akerlund, Ozgun Bursalioglu, Pen Li, Vikas Verma. Additional contributions provided by Adam Roberts, Antoine Caillon, Ari Troper, Asahi Ushio, Beat Gfeller, Catherine Zhang, Carlos Hernandez, Carrie Zhang, Cassie Tarakajian, Chris Deaner, Colin McArdell, Di Li, Eleni Shaw, Ethan Manilow, Ian Simon, Ilaria Manco, Jesse Engel, Jordi Pont-Tuset, Kehang Han, Krishmin Rai, Li Li, Lluis Castrejon, Mikhail Dektiarev, Noah Constant, Rey Coaguila, Tayniat Khan, Tobenna Peter Igwe, Vivek Kumar, and Yotam Mann.
Special thanks to Aäron van den Oord, Mahyar Bordbar, Douglas Eck, Tulsee Doshi, Tom Hume, Koray Kavukcuoglu and Demis Hassabis for their insightful guidance and support throughout the development process.
We also acknowledge the many other individuals who contributed across Google DeepMind and Alphabet, including our colleagues at YouTube (a particular shout out to the YouTube Artist Partnerships team led by Vivien Lewit for their support partnering with the music industry).
Lyria 2 was developed by Andrea Agostinelli, Asahi Ushio, Beat Gfeller, Brian McWilliams, Ethan Manilow, Kazuya Kawakami, Keyang Xu, Matej Kastelic, Mauricio Zuluaga, Mauro Verzetti, Myriam Hamed Torres, Ondrej Skopek, Pavel Khrushkov, Pen Li, Tobenna Peter Igwe, and Zalan Borsos. Additional contributions provided by Adam Roberts, Benigno Uria, Carrie Zhang, Chris Deaner, Colin McArdell, DY Kim, Eleni Shaw, Hongliang Fei, Jason Baldridge, Jesse Engel, Li Li, Luyu Wang, Nemanja Spasojevic, Noah Constant, Ruba Haroun, Tayniat Khan, Volodymyr Mnih, Yan Wu and Zoe Ashwood.