Profile 03 00@2x
" Keep moving forward "

Yun-wen Lin (Vincent)

- Responsible, active, eager to work hard, and willing to explore new things.
- Good communication and teamwork.

- Rich practical experience and leadership skills.
- Research in speech enhancement with machine learning.


Web Development

  • HTML, CSS, Javascript
  • React.js, Redux, JQuery
  • Node.js, Express, MongoDB 
  • Browser Extension
  • Webpack

Machine Learning

  • Keras, Tensorflow
  • Pytorch
  • Speech Enhancement


  • C/C++, Android Java/ Kotlin
  • Git
  • Photoshop
  • GoldWave, Corel VideoStudio


Master in Computer Science & Information Engineering

           National Cheng Kung University, 2018.09 ~ 2020.07

Bachelor in Computer Science & Information Engineering

           National Cheng Kung University, 2014.09 ~ 2018.06


SideeX, Lead Committer, 2016.09 ~ 2020.12

Participated in development of SideeX which extends from Selenium IDE.

  • For rich text function in web pages, added "editContent" command in SideeX to resolve the problems on at least 14 kinds of special sites.
  • Migrated SideeX from XUL-based to WebExtension-based one, which is the base of the new Selenium IDE.
  • Developed SideeX 3, giving more friendly UI, more powerful record-playback mechanism, and increased execution speed at least 50% at the same time.
  • Designed and developed SideeX Runner for operation in CLI which can be bridged on CI/CD.
  • Develop SideeX Launcher which is electron-based application, and it can run on different OSs and call system commands


KKBOX, Software Engineer (intern), 2017.07 ~ 2017.12

Implemented two projects with KKBOX Open API for demonstrations, promotions, and being examples.

  • One is a chatbot for music searching. After typing specific commands and keywords, it returns the related searching results (Deploy on Innovation Chatbot FB fans page now).
  • The other is an android APP with Android Speech Recognizer to control, search and play the musics.



Paragraph image 03 00@2x

AI Speech Enhancement

Pytorch CNN DNN ML 

  • Use machine learning to automatically extract noise, speaker and phoneme features. Then, generate the mask to remove noise in the speech.
  • Consider both the voice quality of STOI and PESQ and the acoustic probability obtained from the Kaldi ASR to improve the word error rate
  • Compared with the result of the basic model, the ASR word error rate is reduced by about 10%

Paragraph image 03 00@2x


Framework7 JS Express MongoDB

  • It is a PWA APP with front-end/back-end techniques.
  • Designed a new version of the NCKU APP for students and faculties, which has been used by more than 10,000 people now.
  • Implemented the timetable, schedule page, and arranged the information of offices and students.
  • Integrated the event registration page with the original web for automatic login.

Paragraph image 03 00@2x

Music Search APP

Flask Kotlin Android

  • Implemented with KKBOX Open API.
  • (1st, 2nd from the left) It is a chatbot in FB Messenger. After entering specific commands and keywords, it searches and returns related musics, songs.
  • (1st from the right) It is an APP with Android Speech Recognizer for searching musics by speaking and playing the songs.

Paragraph image 03 00@2x


JS React Browser Extension

  • It is an automated web UI testing software, which creates test cases and runs in simple, fast way with recording and playback mechanism.
  • Added "editContent" command in SideeX 2 (left) for rich text function on some web pages that cannot be recorded and played.
  • Designed the architecture of SideeX 3 (right) and participated in its function and interface development.