Jonathan Le Roux
Cambridge, Massachusetts, United States
Summary
🤩
RockstarJonathan Le Roux is a distinguished research scientist at MERL with nine years of experience leading speech and audio research, currently focused on machine learning for speech enhancement and source separation. He blends deep academic roots in audio signal processing—spanning Ph.D. and postdoc work on F0 estimation, spectrogram-aware reconstruction, and model-based source separation—with practical leadership of the Speech & Audio team. Jonathan’s expertise bridges model-based and deep-learning approaches, covering the full speech pipeline from acoustic and language modeling to dialog systems and user prediction. Known for inventive extensions to NMF and harmonically-constrained GMMs, he brings a rare combination of statistical inference rigor and hands-on machine-listening practice. Based in Cambridge, MA, he pairs a strong mathematical background with a track record of shipping advanced algorithms from research prototypes toward real-world systems.
9 years of coding experience
Github Skills (70)
spectrogram10
python10
windows10
powerpoint10
pretrain10
matlab10
latex10
ai10
artificial-intelligence10
vba10
dvi10
pytorch10
machine-learning10
xetex9
music9
Programming languages (7)
Github contributions (5)
Jonathan-LeRoux/lws
May 2017 - Feb 2021
Fast spectrogram phase recovery using Local Weighted Sums (C/Python/Matlab)
Contributions:3 releases, 65 commits, 1 PR in 3 years 10 months
matlabpythonspectrogram
Jonathan-LeRoux/IguanaTex
Dec 2016 - Sep 2022
A PowerPoint add-in to insert LaTeX equations into PowerPoint presentations on Windows and Mac
Contributions:7 releases, 36 commits, 2 PRs in 5 years 10 months
latexpowerpointwindowsinsert-latex-equationspowerpoint-presentations
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.