Resume

I serve as a tech lead manager in innovating speech and conversational AI solutions at Uniphore, mainly with end-to-end speech recognition, language modelling and related areas, in terms of research and engineering. I am interested in most aspects of speech and language processing.

Skills

Tools: PyTorch/Tensorflow, k2/icefall, sherpa-onnx, kaldi
Deployment and automation: Docker, Kubernetes, Kafka, Websockets, FastAPI, Git, Jira
Programming languages: Python, C++, Bash
Leadership strengths: Innovation, teamwork, collaboration

Experience

Uniphore • Staff AI Scientist [2021 - Present]

Google • Research Intern [2020]
Worked on a research project with the audio processing team.

Idiap Research Institute • Research Assistant [2017 - 2021]
Developed a PhD thesis on automatic speech assessment and recognition.

Interactive Intelligence • Senior Speech Engineer [2015 - 2017]
Built production-grade ASR models in multiple languages for small vocabulary ASR systems. Worked on acoustic modelling for improved efficacy and on mitigating systematic ASR errors.

Samsung R&D Institute India • Lead Engineer [2013 - 2015]
Worked on robust feature extraction techniques, implemented data selection techniques for ASR training, built and tested acoustic models using large data for multiple languages.

Education

Docteur ès Sciences (PhD)[2017 - 2021]
École polytechnique fédérale de Lausanne (EPFL), Electrical Engg.

Master of Science (MS) by Research[2010 - 2013]
Indian Institute of Technology Madras, Electrical Engg.

Bachelor of Engineering (BE)[2006 - 2010]
Andhra University, Electronics and Communication Engg.

Professional Services and Outreach

Reviewer at ICASSP, Interspeech, ASRU, SLT, WASPAA and ICMI.
Chaired a session on Speech Signal Analysis at ISCA Interspeech 2023.
Participated in IndiaAI roundtable on Generative AI, April 2023.
Interviewed by IndiaAI, May 2022.
Interviewed by The Interview Portal, Feb. 2023.

Select publications

2024

Shashi Kumar, Srikanth Madikeri, Nigmatulina Iuliia, Esaú Villatoro-Tello, Petr Motlicek, Karthik Pandia D S, S. Pavankumar Dubagunta and Aravind Ganapathiraju, “Multitask Speech Recognition and Speaker Change Detection for Unknown Number of Speakers,” in Proceedings of ICASSP. [DOI]

2023

Lokesh Bansal, S. Pavankumar Dubagunta, Malolan Chetlur, Pushpak Jagtap, and Aravind Ganapathiraju, “On the Efficacy and Noise-Robustness of Jointly Learned Speech Emotion and Automatic Speech Recognition,” in Proceedings of Interspeech. [PDF]

Tilak Purohit, Sarthak Yadav, Bogdan Vlasenko, S. Pavankumar Dubagunta, and Mathew Magimai Doss, “Towards Learning Emotion Information from Short Segments of Speech,” in Proceedings of ICASSP. [PDF]

2022

S. Pavankumar Dubagunta, Edoardo Moneta, Eleni Theocharopoulos, and Mathew Magimai Doss, “Towards Automatic Prediction of Non-Expert Perceived Speech Fluency Ratings,” in Companion proceedings of the International Conference on Multimodal Interaction (ICMI). [DOI]

S. Pavankumar Dubagunta, Rob J. J. H. van Son, and Mathew Magimai Doss, “Adjustable deterministic pseudonymization of speech,” in Computer Speech and Language. [DOI]

2021

S. Pavankumar Dubagunta, “Novel Methods for Incorporating Prior Knowledge for Automatic Speech Assessment,” PhD thesis, École polytechnique fédérale de Lausanne (EPFL). [PDF] [Talk]