About Me

My name is Min-Jae Hwang. I am currently a postdoctoral researcher at Meta AI. Prior to Meta AI, I was a research scientist at Naver Corporation. I received my Ph.D. degree in department of Electrical and Electronics at Yonsei University. During my Ph.D. course, I was fortunate to have research experiences as an intern at Microsoft Research Asia and Naver Corporation.

My research interests include Text-to-Speech (TTS) and Speech-to-Speech Translation (S2ST). Specifically, my research history focuses on improving performance of neural vocoder for TTS systems. At Meta, I extended my research field to the expressive S2ST, which preserves source speech’s paralinguistic characteristics during speech translation process.

I’m open to learn new knowledge and enjoy applying them to solve our society’s real-world problems. If you are interested in me, feel free to contact me.

Download my CV

NEWS!
11/2023 : We launched Seamless, a new family of AI translation models that preserve expression and deliver near-real time streaming translations.
11/2023 : SeamlessM4T was recognized by TIME magazine among the best inventions of 2023!
8/2023 : We launched SeamlessM4T, a foundational multilingual and multitask model that seamlessly translates and transcribes across speech and text.
9/2022 : Our paper¹ has been accepted to NeurIPS 2022.
6/2022 : I'll join Meta AI, Seattle, USA as a Postdoctoral Researcher for this October!

Research Interests

• Speech-to-speech translation (S2ST)
  • Expressive S2ST system
• Text-to-speech (TTS) synthesis
  • High-quality and real-time waveform generation method
  • Expressive and emotional TTS system

Recent Publications

• Seamless: Multilingual Expressive and Streaming Speech Translation
  Team Seamless Communication

• SeamlessM4T—Massively Multilingual & Multimodal Machine Translation
  Team Seamless Communication

• HierSpeech: Bridging the Gap between Text and Speech by Hierarchical Variational Inference using Self-supervised Representations for Speech Synthesis
  Sang-Hoon Lee, Seung-Bin Kim, Ji-Hyun Lee, Eunwoo Song, Min-Jae Hwang, Seong-Whan Lee
  Accepted to NeurIPS 2022

• Language Model-Based Emotion Prediction Methods for Emotional Speech Synthesis Systems
  Hyun-Wook Yoon, Ohsung Kwon, Hoyeon Lee, Ryuichi Yamamoto, Eunwoo Song, Jae-Min Kim, Min-Jae Hwang
  Accepted to Interspeech 2022

• TTS-by-TTS 2: Data-selective Augmentation for Neural Speech Synthesis using Ranking Support Vector Machine with Variational Autoencoder
  Eunwoo Song, Ryuichi Yamamoto, Ohsung Kwon, Chan-Ho Song, Min-Jae Hwang, Suhyeon Oh, Hyun-Wook Yoon, Jin-Seob Kim, Jae-Min Kim
  Accepted to Interspeech 2022

  [See more]