About Sytronik

Jeongmin Liu (유정민, 劉政旼)
AI/ML Engineer in Voice model, AI Tech, Naver Cloud Corp.
I research and develop an end-to-end speech synthesis system and a speech enhancement system with my team.

Interests

• Signal Processing (Audio / Speech)
• Deep Learning / Machine Learning
• Python (PyTorch, Fast-API, …)
• Rust Language

Projects

• Thesia: Multi-track Spectrogram / Waveform Viewer
  Nov 2020 - now
  An open-source GUI app that shows multi-track spectrograms and waveform envelopes
  GitHub Repo
• pYIN-rs
  Apr 2022
  A Rust version of pYIN algorithm, one of pitch detection algorithms
  GitHub repo
• Deep Griffin-Lim Iteration
  Sep 2019 - Oct 2019
  An implementation of a paper “Deep Griffin-Lim Iteration”
  GitHub repo

Papers

Conference Papers

• Jeongmin Liu, Byeongho Jo, Jung-Woo Choi, Dereverberation Based on Deep Neural Networks with Directional Feature from Spherical Microphone Array Recordings, in Proc. of the 23rd International Congress on Acoustics (ICA 2019), Aachen, Germany, September 9-13, 2019.

  In the paper, the authors propose a method that uses spatially-averaged acoustic intensity vector as an input feature of the DNN in order to make the DNN perform speech dereverberation by considering spatial information.
  Paper | GitHub Repo

Languages

• 한국어 (모어) / Korean native speaker
• English (good working knowledge)
• 中文 (初學者) / Chinese Mandarin beginner