Chia-Yu Li

Topic: Advanced neural-network based Acoustic Models for Automatic Speech Recognition Systems • Implement the Deep neural network (DNN) , Convolution neural network (CNN) and CNN with Context Expansion and Attention (LACEA) architecture for the acoustic model of Automatic Speech Recognition (ASR) system using Python, theano, lasagne • Investigate the performance of ASR system with DNN/CNN/LACEA based acoustic model training on English Switchboard corpus and Wall Street Journal corpus

• Develop the bash scripts to transform raw data from English/Spanish Speech corpora into KALDI readable data • Develop the bash scripts to generate the Spanish pronunciation of input words • Merge different pronunciation dictionaries with different phone sets into one pronunciation dictionary • Train the language model with Spanish corpus using SRILM toolkit • Interpolate the language model with different Spanish corpora using SRILM toolkit

• Bring up WLAN/Bluetooth function in Android, Windows and Linux system • Develop the automatic factory test tool for Bluetooth product • Debugging WLAN issues reported from customer side (such as ASUS, HTC), Authorized Test Laboratories (SGS Labs, Allion Labs) for Wi-Fi Alliance Certification • Managing customer support and on-site troubleshoot events

• Develop and maintain the core WLAN module in Android system • Bring up WLAN function on different platforms (Broadcom, Qualcomm, Atheros and TI) • Debug the WLAN/Bluetooth co-exist problems and improve the WLAN/Bluetooth performance in Android system

Courses: Information Retrieval and data mining, Statistical Machine Translation, Automatic Speech Recognition using KALDI, Tweet Emotion prediction using machine learning approaches, Deep learning for NLP, Speech Synthesis, Speech Recognition, Sentiment Analysis

Master Thesis: Analytical Study of Two-Tier Peer-to-Peer Traffic Information System. Courses: Algorithm, Network Programming, Peer-to-Peer Computing, Data mining, Operating System, Wireless Security