Biography
Hello! I am Haohan Guo (郭浩瀚), a Research Scientist at ByteDance Seed. I received my Ph.D. in Speech AI from The Chinese University of Hong Kong, advised by Prof. Helen Meng, and my M.S. and B.S. degrees from Northwestern Polytechnical University, advised by Prof. Lei Xie. My research focuses on speech and audio generation. Please feel free to contact me if you are interested in my work.
Interests
- Speech & Audio Processing
- Speech Synthesis
- Voice Conversion
- Audio Generation
Education
Ph.D. in Speech AI, 2021 - 2026
The Chinese University of Hong Kong
MSc in Computer Science, 2017-2020
Northwestern Polytechnical University
BSc in Computer Science, 2013-2017
Northwestern Polytechnical University
Publications
Work Experience
lab, internship, full-time employee
Applied Scientist Intern
Work as an applied scientist intern to develop large-scale TTS system based on large language models (LLM).
Research Intern
Work as a researcher intern to investigate the application of speech representations in TTS.
Researcher
Work as a researcher on singing voice conversion. We aim to develop a commercial singing conversion system which can convert arbitrary singing voice to the target timbre. High sound quality and accurate melody expression are both required.
Research Intern
Research topic is multi-singer singing voice conversion. We propose a MelGAN based end-to-end PPG-SVC model. It significantly improves the sound quality and singer similarity over the conventional PPG-SVC framework. The work is summarized to the paper, Phonetic Posteriorgrams based Many-to-Many Singing Voice Conversion via Adversarial Training.
Research Intern
Supervised by Frank K. Soong and Lei He. We aim to improve the robustness and naturalness of end-to-end TTS. Two main works are published to INTERSPEECH 2019, A New GAN-based End-to-End TTS Training Algorithm and Exploiting Syntactic Features in a Parsed Tree to Improve End-to-End TTS. We also investigate the conversational TTS using the end-to-end approach. The work is published to SLT 2021, Conversational End-to-End TTS for Voice Agents.
Research Intern
Be responsible for the optimization of the front-end modules of TTS system, including G2P and Prosody model.
Education
Ph.D. in Speech AI
Supervised by Prof. Helen Meng.
M.S. Student
Supervised by Prof. Lei Xie.