Chris Donahue
Chris Donahue is an AI researcher who develops models and interactive tools for making music. He grew up playing piano and drums, studied computer science at the University of Texas at Austin, and completed a PhD at UC San Diego before starting a Stanford postdoc with Percy Liang in 2019. His work includes Dance Dance Convolution for music-game choreography, WaveGAN for audio synthesis, and Piano Genie for improvising with eight buttons. He helped launch Beat Sage in 2020, co-authored SingSong in 2023, and contributed to Magenta RealTime in 2025. He is the Dannenberg Assistant Professor in Carnegie Mellon University's Computer Science Department, leads its Generative Creativity Lab, and works part-time on Google DeepMind's Magenta team.
- Now at
- Carnegie Mellon University and Google DeepMind
- Works
- 84
- Citations
- 4,363
- h-index
- 17
- Badges
Publication numbers from OpenAlex, as of 27 October 2026.
Timeline
-
By 2013
Music and computer science
He grows up playing piano and drums and studies computer science at the University of Texas at Austin. In December 2013, he completes the undergraduate honors thesis “Applications of Genetic Programming to Digital Audio Synthesis,” which uses genetic programming to make sound synthesis algorithms mimic acoustic and electronic instruments.
-
2017
Dance Dance Convolution
In 2017, he coauthors Dance Dance Convolution with Zachary C. Lipton and Julian McAuley. The system learns to turn an audio track into a Dance Dance Revolution step chart, using neural networks to decide when to place steps and which steps to select.
-
2019
WaveGAN and Piano Genie
His 2019 ICLR paper with Julian McAuley and Miller Puckette introduces WaveGAN, which uses generative adversarial networks to synthesize audio waveforms. Piano Genie, published with Ian Simon and Sander Dieleman in 2019, uses recurrent neural networks to let people improvise piano music through eight buttons.
-
2019
A doctorate and Stanford
His UC San Diego PhD, jointly advised by Miller Puckette in music and Julian McAuley in computer science, includes a 2019 dissertation on music, AI and interaction. In 2019, he starts a postdoc in Stanford University's Computer Science Department, advised by Percy Liang.
-
April 2020
Launching Beat Sage
In April 2020, he and Abhay Agarwal launch Beat Sage. The service uses AI and neural networks to generate custom Beat Saber maps from songs.
-
July 2020
Language models fill blanks
At ACL in July 2020, he coauthors “Enabling Language Models to Fill in the Blanks” with Mina Lee and Percy Liang. Their method trains language models to fill missing passages within documents and tests sentence completion in short stories, scientific abstracts and lyrics.
-
2022–2023
Google Magenta and SingSong
In 2022, he becomes a research scientist at Google Magenta. In January 2023, he coauthors SingSong, a system that generates instrumental music to accompany a person's singing by adapting AudioLM and training on separated vocal and instrumental tracks.
-
2023–2024
A professorship at Carnegie Mellon
He joins Carnegie Mellon's School of Computer Science faculty in 2023 as an assistant professor in the Computer Science Department. In 2024, he becomes the first recipient of the Dannenberg Career Development Professorship, established by Roger Dannenberg to support research and teaching in computation and music.
-
2025
Live music with Magenta
In 2025, he coauthors Live Music Models, which introduces models that produce a continuous stream of music with live user control. The team releases Magenta RealTime, an open-weights model that users can steer with text or audio prompts.
-
As of October 2026
Music AI at CMU
As of October 2026, he is the Dannenberg Assistant Professor in Carnegie Mellon's Computer Science Department and leads the Generative Creativity Lab (G-CLef). He also works part-time as a research scientist on Google DeepMind's Magenta team, developing generative AI for music and creative interaction.
Career
- NowCarnegie Mellon UniversityDannenberg Assistant Professor, Computer Science, Lead, Generative Creativity Lab (G-CLef)
- NowGoogle DeepMindPart-Time Research Scientist, Magenta Team
- 2022–2023GoogleResearch Scientist, Magenta Team
- 2019–2023Stanford UniversityPostdoctoral Researcher, Computer Science
Education
- 2019University of California, San DiegoPhD
- University of Texas at AustinUndergraduate studies in computer science
Sources and credits
Photo: We have no published photo of Chris Donahue yet. Send us a free-licensed photo.
- Singing a New Tune: Computational Music — The Link - The Magazine of the Kenneth C. Griffin School of Computer Science
- Peter Stone :: Professor at the University of Texas at Austin, Department of Computer Science :: Students
- Dance Dance Convolution
- Adversarial Audio Synthesis | Chris Donahue
- Piano Genie | Chris Donahue
- Chris Donahue | Carnegie Mellon University Computer Science Department
- Chris Donahue
- Machine Learning Group
- Get Rhythm: How Beat Sage Uses AI To Create Beat Saber Maps
- Enabling Language Models to Fill in the Blanks - ACL Anthology
- SingSong: Generating Musical Accompaniments from Singing | Chris Donahue
- SCS Faculty Receive Endowed Professorships | Carnegie Mellon University Computer Science Department
- G-CLef @ CMU
- Chris Donahue

