CV
Education, experience, selected publications, awards, and domestic presentations.
General information
| Name | Hien Ohnaka |
| Title | Engineer |
| ohnaka.hien@gmail.com | |
| Location | Japan |
| Summary | Full-time engineer at LY Corporation, Japan. |
| GitHub | i17oonaka-h |
| X | nehi_h |
Experience
- 2026-04 – present
Full-time Engineer
LY Corporation
Japan
- 2024-12 – 2026-03
Part-time Researcher
LY Corporation R&D
Japan
- Improved phonemic and prosodic annotation models for text-to-speech; the work was accepted to Interspeech 2025.
- Researched neural vocoders for high-quality speech generation from self-supervised speech features; the work was accepted to ICASSP 2026.
- 2025-02 – 2025-03
Research Intern
NTT Communication Science Laboratories
Japan
- Researched post-refinement of ASR hypotheses for joint multi-channel distant speech recognition.
- 2024-08 – 2024-10
Research Intern
LY Corporation R&D
Japan
- Improved phonemic and prosodic annotation models for text-to-speech and related subtasks.
- 2022-06 – 2022-07
Research Intern
The University of Tokyo
Japan
- Researched environmental sound synthesis from visual onomatopoeia; the work was accepted to ICASSP 2023.
Education
- 2024-04 – 2026-03
Nara Institute of Science and Technology
M.S., Information Science
Japan
- Graduate School of Science and Technology, Master's Course.
- Intelligent Robot Dialogue Laboratory (RIKEN), supervised by Prof. Koichiro Yoshino.
- Received the NAIST Best Student Award.
- 2022-04 – 2024-03
National Institute of Technology, Tokuyama College
B.E., Computer Science and Electronic Engineering
Japan
- Advanced Course.
- Miyazaki Laboratory, supervised by Prof. Ryoichi Miyazaki.
- 2017-04 – 2022-03
National Institute of Technology, Tokuyama College
Computer Science and Electronic Engineering
Japan
- Regular Course.
- Miyazaki Laboratory, supervised by Prof. Ryoichi Miyazaki.
Selected Publications
- 2026-05
Wave-Trainer-Fit: Neural Vocoder with Trainable Prior and Fixed-Point Iteration towards High-Quality Speech Generation from SSL Features
Hien Ohnaka, Yuma Shirahata, Masaya Kawamura
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
- 2025-08
Grapheme-Coherent Phonemic and Prosodic Annotation of Speech by Implicit and Explicit Grapheme Conditioning
Hien Ohnaka, Yuma Shirahata, Byeongseon Park, Ryuichi Yamamoto
Interspeech
- 2025-08
Modeling Turn-Taking Speed and Speaker Characteristics
Kazuyo Onishi, Hien Ohnaka, Koichiro Yoshino
Annual Meeting of the Special Interest Group on Discourse and Dialogue (SIGDIAL)
- 2025-02
Unsupervised Speech Enhancement with Spectral Kurtosis and Double Deep Priors
Hien Ohnaka, Ryoichi Miyazaki
Acoustical Science and Technology
- 2023-06
Visual Onoma-to-Wave: Environmental Sound Synthesis from Visual Onomatopoeias and Sound-Source Images
Hien Ohnaka, Shinnosuke Takamichi, Keisuke Imoto, Yuki Okamoto, Kazuki Fujii, Hiroshi Saruwatari
IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
Awards
- 2026-03
Best Student Award
Awarded by Nara Institute of Science and Technology. Announcement
- 2025-03
Committee Special Award
Awarded by the Association for Natural Language Processing for work on jointly predicting response timing and short responses for real-time spoken dialogue systems. Award details
- 2024-12
IEEE SPS Tokyo Joint Chapter Student Award
Awarded for Visual Onoma-to-Wave. Award details
Domestic Presentations (Non-peer-reviewed)
- 2026-03
音声の意味的類似度に基づいた対照学習によるリアルタイム応答選択
- 2026-03
Voice Activity Projection とガンマ分布ハザード関数を用いて対話の間を考慮したターンシフトモデル
- 2025-03
- 2024-11
- 2024-03
- 2024-03
事前学習不要なDNN音声強調の反復処理に基づく性能改善
- 2024-03
クリーン不要なDNN音声雑音除去に用いる目的信号の分析
- 2023-12
過学習の抑制に基づく事前学習不要なDNN音声強調
- 2023-03
Double-DIPに基づく教師なし雑音除去の分析と性能改善
- 2023-02
Visual onoma-to-wave:音源画像と画像オノマトペを利用した環境音合成の提案
- 2022-11
クリーン音声を必要としない時間領域でのDNN音声強調
- 2022-05
Double-DIP に基づく単一雑音混入音声のみを用いた雑音除去の検討
- 2021-12
深層ニューラルネットワークにおける学習過程の信号分布に着目した雑音抑圧の検討