Towards a Professional Web Page
Abstract
I am a postdoc in the HAI-Res lab at MIT, working with Anna Huang and Paris Smaragdis. Previously, I was a member of the Interactive Audio Lab at Northwestern University, advised by Bryan Pardo. I work on building and studying the impacts of AI systems for audio, particularly in music and speech.
1Introduction
I am working on interactive AI systems for audio and music co-creation, including controllable symbolic and audio-domain generation; real-time performance accompaniment; and “decomposing” media artifacts into interpretable, manipulable representations through which we can study and build on the creative processes behind them. Through my research, I aim to (1) develop new tools that broaden the space of possible creative interactions [1, 2, 5, 7, 10]; (2) uncover limitations of existing audio AI methods [3, 4, 9]; and (3) facilitate the safe and beneficial deployment of audio AI systems [6, 8].
I think we can learn a lot about building better AI systems by designing for the tight demands of creative tasks like musical performance – including responsiveness, interpretability, and attribution – and by studying the ways in which musicians and artists navigate these demands.
2News
| Date | Event |
|---|
| 11.2026 | D-Composer won Best Paper at ISMIR 2026 |
| 9.2026 | Started as a Tayebati Postdoctoral Fellow at MIT, advised by Anna Huang and Paris Smaragdis |
| 7.2026 | Defended my PhD at Northwestern |
| 9.2025 | Presented The Rhythm In Anything at ISMIR 2025 in Daejeon |
| 8.2025 | Gave a tutorial on speech watermarking at Interspeech 2025 in Rotterdam (slides and code) |
| 4.2025 | Deep Audio Watermarks are Shallow presented at the first ICLR Workshop on GenAI Watermarking |
| 4.2025 | Presented Code Drift at ICASSP 2025 in Hyderabad |
| 2.2025 | Passed my PhD prospectus |
3Research
Papers I have worked on; click to see demos.
- [1] Patrick O’Reilly, Saumya Pailwan, Annie Chu, Jason Smith, and Bryan Pardo. D-Composer: Language modeling for simultaneous drum transcription and sound event decomposition. In Proceedings of the International Society for Music Information Retrieval Conference (ISMIR), 2026. paper project bibtex
- [2] Patrick O’Reilly, Julia Barnett, Hugo Flores García, Annie Chu, Nathan Pruyne, Prem Seetharaman, and Bryan Pardo. The Rhythm In Anything: Audio-prompted drums generation with masked language modeling. In Proceedings of the International Society for Music Information Retrieval Conference (ISMIR), 2025. paper late-breaking demo project code bibtex
- [3] Patrick O’Reilly, Zeyu Jin, Jiaqi Su, and Bryan Pardo. Deep audio watermarks are shallow: Limitations of post-hoc watermarking techniques for speech. In ICLR Workshop on GenAI Watermarking, 2025. paper project bibtex
- [4] Patrick O’Reilly, Prem Seetharaman, Jiaqi Su, Zeyu Jin, and Bryan Pardo. Code drift: Towards idempotent neural audio codecs. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025. paper project bibtex
- [5] Annie Chu, Patrick O’Reilly, Julia Barnett, and Bryan Pardo. Text2FX: Harnessing CLAP embeddings for text-guided audio effects. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2025. paper project bibtex
- [6] Patrick O’Reilly, Zeyu Jin, Jiaqi Su, and Bryan Pardo. MaskMark: Robust neural watermarking for real and synthetic speech. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024. paper project bibtex
- [7] Hugo Flores García, Christodoulos Benetatos, Patrick O’Reilly, Aldo Aguilar, Zhiyao Duan, and Bryan Pardo. HARP: Bringing deep learning to the DAW with hosted, asynchronous, remote processing. In NeurIPS Workshop on Machine Learning for Creativity and Design, 2023. paper project code bibtex
- [8] Patrick O’Reilly, Andreas Bugler, Keshav Bhandari, Max Morrison, and Bryan Pardo. VoiceBlock: Privacy through real-time adversarial attacks with audio-to-audio models. In Advances in Neural Information Processing Systems (NeurIPS), 2022. paper project video code bibtex
- [9] Patrick O’Reilly, Pranjal Awasthi, Aravindan Vijayaraghavan, and Bryan Pardo. Effective and inconspicuous over-the-air adversarial examples with adaptive filtering. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022. paper project talk code bibtex
- [10] Ethan Manilow, Patrick O’Reilly, Prem Seetharaman, and Bryan Pardo. Unsupervised source separation by steering pretrained music models. In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2022. paper project code bibtex
4Prior Work
Before Northwestern, I received an MS in Computer Science from the University of Illinois Chicago and a BA in Mathematics and Music from Carleton College.
Acknowledgments
The design of this page was inspired by Interactive Audio Lab alumnus Zafar Rafii.
Appendix
AMusic
The Last Will and Testament of Eric Jones: a vocal performance by the Skittles rabbit, separated and denoised with Demucs and harmonized with Coconet. Click the record to play it, or see the music page.
BWriting
[Placeholder: posts will be listed here.]
COther projects
[Placeholder: one-off projects will be listed here.]