Yashish M. Siriwardena

Senior Research Scientist, Hippocratic AI · hpersym@gmail.com · LinkedIn · Google Scholar

Download as PDF

Experience

Industry Academic

1. Sep 2025 – Present
   Hippocratic AI, Menlo Park, CA
   • Senior Research ScientistJun 2026 – Present
   • Senior Research EngineerSep 2025 – Jun 2026
   • Research and development of speech and voice AI, including text-to-speech and speech-to-speech systems, for healthcare conversational agents.
2. Jan 2024 – Sep 2025
   OmniSpeech, College Park, MD
   • Senior Research Engineer
   • Developed a deepfake detection system with the best single-system performance (EER, minDCF) in the open condition of the ASVspoof5 (2024) challenge, and researched interpretable acoustic features for deepfake detection.
   • Delivered a speech enhancement system that suppresses background and machine noise of an MRI scanner to improve communication between operator and subject.
   • Developed a full-band (48 kHz) noise suppression system for PC and cloud video conferencing.
   • Developed an API that extracts a key set of vocal biomarkers from speech.
3. Jun 2023 – Sep 2023
   Dolby Laboratories, San Francisco, CA
   • Research Intern, Multimodal Science
   • Developed a foreign accent conversion system with voice conversion transformers (PyTorch, HiFi-GAN, ESPnet, HuBERT, AWS SageMaker) that outperformed the state-of-the-art system.
4. May 2022 – Aug 2022
   Skylyte, New York, NY
   • Machine Learning Engineer Intern
   • Analyzed vocal biomarkers to find the best predictors of burnout and resilience, and built models to classify burnout, resilience, valence and arousal from workplace speech.
   • Designed and started a speech data collection for burnout symptoms at the workplace.
5. May 2020 – Dec 2023
   University of Maryland, College Park, MD
   • Graduate Research Assistant, Speech Communication Lab
   • Built a HuBERT-based, speaker-independent speech inversion system that estimates nasal and glottal activity.
   • Developed the MirrorNet, an unsupervised learning algorithm inspired by sensorimotor interaction that learns the control parameters of an articulatory or audio synthesizer.
   • Improved acoustic-to-articulatory speech inversion with data augmentation and multi-task learning.
   • Built mispronunciation detection for children's speech, a speaker-independent articulatory synthesizer, and multimodal pipelines for detecting schizophrenia symptoms and emotion.
6. Aug 2018 – May 2020
   University of Maryland, College Park, MD
   • Graduate Teaching Assistant
   • ENEE 645 Compilers and Optimization; ENEE 322 Signals and Systems; ENEE 222 Elements of Discrete Signal Analysis.
7. Oct 2015 – Jul 2018
   Synergen Technology Labs, Nugegoda, Sri Lanka
   • Research EngineerMay 2017 – Jul 2018
   • Research InternOct 2015 – Apr 2016
   • Developed and delivered the communication interface of a wearable ring that captures PPG and motion signals for a cardiac arrest monitoring system.
   • Built an Android app that talks to a wearable baby monitor over BLE and computes heart rate, respiratory rate and SpO2 from the PPG signal.

Education

• 2023 PhD, Electrical and Computer Engineering, University of Maryland, College Park Dissertation: Towards Extending Acoustic-to-Articulatory Speech Inversion and Learning Articulatory Representations. Advisers: Carol Espy-Wilson, Shihab Shamma.
• 2022 MS, Electrical Engineering, University of Maryland, College Park
• 2017 BSc (Honours), Biomedical Engineering, University of Moratuwa, Sri Lanka

Interests

Text-to-speech · Speech-to-speech · Speech and audio enhancement · Audio deepfake detection · Voice and accent conversion · Speech and music synthesis · Articulatory speech production · Multimodal systems · Vocal biomarkers for mental health

Selected publications

• HARP: agentic hybrid retrieval and analysis for long-form audio Chin-Jou Li, Masao Someki, Woojeong Jin, Yashish M. Siriwardena, Tanmay Laud, Shanil Puri, Shinji Watanabe. arXiv preprint arXiv:2609.14116. [Preprint]
• Speaker-independent speech inversion for recovery of velopharyngeal port constriction degree Yashish M. Siriwardena, Suzanne E. Boyce, Mark K. Tiede, Liran Oren, Brittany Fletcher, Michael Stern, Carol Y. Espy-Wilson. Journal of the Acoustical Society of America 156(2): 1380-1390.
• Accent conversion with articulatory representations Yashish M. Siriwardena, Nathan Swedlow, Audrey Howard, Evan Gitterman, Dan Darcy, Carol Espy-Wilson, Andrea Fanelli. Interspeech 2024.
• Improving speech inversion through self-supervised embeddings and enhanced tract variables Yashish M. Siriwardena*, Ahmed Adel Attia*, Carol Espy-Wilson. European Signal Processing Conference (EUSIPCO) 2024.
• A multimodal framework for assessment of the schizophrenia spectrum Gowtham Premananth, Yashish M. Siriwardena, Philip Resnik, Sonia Bansal, Deanna L. Kelly, Carol Espy-Wilson. Interspeech 2024.
• A multi-modal approach for identifying schizophrenia using cross-modal attention Gowtham Premananth, Yashish M. Siriwardena, Philip Resnik, Carol Espy-Wilson. IEEE Engineering in Medicine and Biology Society (EMBC) 2024.
• Learning to compute the articulatory representations of speech with the MirrorNet Yashish M. Siriwardena, Carol Espy-Wilson, Shihab Shamma. Interspeech 2023. [Preprint]
• Speaker-independent speech inversion for estimation of nasalance Yashish M. Siriwardena, Carol Espy-Wilson, Suzanne Boyce, Mark K. Tiede, Liran Oren. Interspeech 2023. [Preprint]
• Acoustic-to-articulatory speech inversion features for mispronunciation detection of /ɹ/ in child speech sound disorders Yashish M. Siriwardena*, Nina R. Benway*, Jonathan L. Preston, Elaine Hitchcock, Tara McAllister, Carol Espy-Wilson. Interspeech 2023. [Preprint]
• The Secret Source: incorporating source features to improve acoustic-to-articulatory speech inversion Yashish M. Siriwardena, Carol Espy-Wilson. IEEE ICASSP 2023. [Paper][Poster][Video][Code]
• Audio data augmentation for acoustic-to-articulatory speech inversion Yashish M. Siriwardena, Ahmed Adel Attia, Ganesh Sivaraman, Carol Espy-Wilson. European Signal Processing Conference (EUSIPCO) 2023. [Preprint]
• Acoustic-to-articulatory speech inversion with multi-task learning Yashish M. Siriwardena, Ganesh Sivaraman, Carol Espy-Wilson. Interspeech 2022. [Paper][Slides]
• The MirrorNet: learning audio synthesizer controls inspired by sensorimotor interactions Yashish M. Siriwardena, Guilhem Marion, Shihab Shamma. IEEE ICASSP 2022. [Paper][Poster][Video][Code]
• Multimodal approach for assessing neuromotor coordination in schizophrenia using convolutional neural networks Yashish M. Siriwardena, Carol Espy-Wilson, Deanna L. Kelly, Chris Kitchen. ACM International Conference on Multimodal Interaction (ICMI) 2021. [Paper][Poster][Video]
• Emotion recognition with articulatory coordination features Yashish M. Siriwardena. 181st Meeting of the Acoustical Society of America 2021. [Poster]
• Inverted vocal tract variables and facial action units to quantify neuromotor coordination in schizophrenia Yashish M. Siriwardena. 12th International Seminar on Speech Production (ISSP) 2020. [Preprint]

Skills

• Languages: Python, C, C++, Java, Matlab
• ML frameworks: PyTorch, TensorFlow/Keras, ONNX Runtime, ExecuTorch
• Speech and audio: Librosa, Praat, ESPnet, HuBERT, HiFi-GAN, openSMILE, SciPy, scikit-learn
• Infrastructure: AWS SageMaker, Ray

Honors and awards

• IEEE ICASSP Rising Star in Signal Processing, 2023, for outstanding PhD thesis work
• Jacob K. Goldhaber Travel Award, Interspeech 2022
• ASA travel grant and Jacob K. Goldhaber Travel Award, 181st Meeting of the Acoustical Society of America, 2021
• Mahapola Higher Education Merit Scholarship (top 100 of about 40,000 candidates, G.C.E. Advanced Level 2011)
• Sisu Udana Scholarship and Dialog Merit Scholarship (7th of about 0.4 million candidates, G.C.E. Ordinary Level 2008)

Professional activities

• Reviewer: Speech Communication (journal), Interspeech 2024, IEEE ICASSP, 13th International Seminar on Speech Production (ISSP) 2024
• Past chair, IEEE Robotics and Automation Society student chapter, University of Moratuwa
• University category organizer, Sri Lanka Robotics Challenge 2014