Tomoki Hayashi
Postdoctoral Researcher at 株式会社 Human Dataware Lab.
Nagoya, Japan
Summary
🤩
Rockstar🎓
Top SchoolTomoki Hayashi is a postdoctoral researcher at Nagoya University, a researcher at TARVO Inc., and COO of Human Dataware Lab with eight years of experience in statistical speech and audio processing. He holds a Ph.D. in information science from Nagoya University and is a principal developer of the widely used ESPnet end-to-end speech toolkit. His open-source work spans pragmatic engineering (installation and compatibility fixes) to core model building—implementing residual and upsampling modules and generator code for ParallelWaveGAN—bridging research prototypes to production-ready toolchains. A recipient of the IEEE SPS Japan 2020 Young Author Best Paper Award and the ASJ 2020 Itakura Award, he blends rigorous academic research with hands-on software leadership from Nagoya, Japan.
8 years of coding experience
五条高等学校博士, 情報科学研究科 メディア科学専攻, 博士, 情報科学研究科 メディア科学専攻 at 名古屋大学
Github Skills (15)
neural-network10
convolutional-neural-network10
pytorch10
neuralnetwork10
audio-processing10
convolutional-neural-networks10
python10
scripting9
bash9
model-building9
versioncontrol9
version-control9
script9
git9
software-packaging7
Programming languages (11)
Github contributions (5)
espnet/espnet
Jun 2018 - Jan 2023
End-to-End Speech Processing Toolkit
Role in this project:
Back-end Developer
Back-end DeveloperContributions:50 releases, 252 reviews, 3190 commits in 4 years 7 months
Contributions summary:Tomoki's commits focus on updating the version number in the project's setup file, adding compatibility features, and fixing bugs related to the installation process. The contributions include adding fill_missing_args for compatibility, fixing issues related to pretrained models, and making adjustments to the scripting. The work primarily involves modifying existing code files, specifically relating to setup and utility scripts.
speech-recognitionspeech-separationchainerspoken-language-understandingspeech-processing
kan-bayashi/ParallelWaveGAN
Oct 2019 - Jan 2023
Unofficial Parallel WaveGAN (+ MelGAN & Multi-band MelGAN & HiFi-GAN & StyleMelGAN) with Pytorch
Role in this project:Back-end Developer
Back-end DeveloperContributions:22 releases, 28 reviews, 1150 commits in 3 years 3 months
Contributions summary:Tomoki's contributions focus on implementing and integrating fundamental audio processing building blocks for a speech generation model. Specifically, the user implemented a residual block module and an upsampling module, both crucial components for the generation process within the WaveGAN framework. The user also built the generator class, which combines and configures different audio layers to build a usable model.
realtimebandneural-vocodermelganparallel-wavenet
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.