Sathvik Udupa
Research Assistant at Brno University of Technology
Brno, Southeast, Czechia
Summary
👤
Senior🎓
Top SchoolSathvik Udupa is a research-focused machine learning engineer with 8 years of experience specializing in speech and audio technologies, currently based in Brno and affiliated with Brno University of Technology and BUT Speech. He has a strong track record in open-source speech toolkits—contributing core FastSpeech/FastSpeech2 implementations and HiFi-GAN integration to SpeechBrain and adding end-to-end and Kaldi recipes for ESPnet—demonstrating both model design and production-oriented tooling skills. His background spans academic research at IISc and hands-on industry work in healthcare AI, giving him a practical bent on translating research into applied systems. Notably, he contributed to benchmarking and training Large Audio Language Models at JSALT 2025, reflecting emerging expertise at the intersection of LLMs and audio.
8 years of coding experience
5 years of employment as a software developer
Bachelor of Engineering, Electrical, Electronics and Communications Engineering, Bachelor of Engineering, Electrical, Electronics and Communications Engineering at Ramaiah Institute Of TechnologyPre University Course(11th,12th), Science(PCMCs), Pre University Course(11th,12th), Science(PCMCs) at Poornaprajna College, Udupi - 576102
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (20)
pytorch10
scripting10
model-driven10
audio-processing10
speech-synthesis10
transformer-models10
model-building10
kaldi10
script10
deep-learning10
sh10
speech-recognition10
shell10
speechbrain10
modeling10
Programming languages (5)
Github contributions (5)
espnet/espnet
Jul 2021 - Aug 2021
End-to-End Speech Processing Toolkit
Role in this project:
Backend Developer
Backend DeveloperContributions:7 reviews, 68 commits, 7 PRs in 27 days
Contributions summary:Sathvik added baseline recipes for subtask2, including both end-to-end and Kaldi implementations, to the espnet fork. These changes involved modifying shell scripts related to feature generation, dictionary and JSON data preparation, language model training, and network training. The user also renamed and restructured directories related to the "mucs_subtask2" experiments, indicating a focus on adapting the toolkit to specific speech processing tasks.
speech-recognitionspeech-separationchainerspoken-language-understandingspeech-processing
speechbrain/speechbrain
Mar 2021 - Oct 2023
A PyTorch-based Speech Toolkit
Role in this project:ML Engineer
ML EngineerContributions:3 reviews, 6 PRs, 5 comments in 2 years 7 months
Contributions summary:Sathvik contributed significantly to the FastSpeech and FastSpeech2 model implementations, a core component of the speechbrain toolkit. Their commits show active involvement in model architecture design, particularly in the integration of new features like duration and pitch predictors, and integrating the hifigan vocoder for speech synthesis. They also worked on refactoring and cleaning existing code, ensuring compatibility and correctness.
voice-recognitionasrspeech-recognitionspeech-separationspeaker-verification
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.