Dr. Naoki Kimura
木村直紀 · Chief Research Scientist at LY Corporation · Silent speech interfaces & human–computer interaction
Dr. Naoki Kimura is a Chief Research Scientist at LY Corporation working on human–computer interaction, with a focus on silent speech interfaces and wearable and reduced-key text entry. This site is his personal research homepage: flagship projects (SilentSpeller, SottoVoce, 3-Key-Input) and publications. He also maintains the Silent Speech Interface Review on GitHub.
Own research · Silent speech hub · Type without speaking · SilentSpeller · SottoVoce · Publications · SSI paper review (GitHub) · CHI 2026 reviews · When Help Hurts · Question map · 日本語
木村 直紀はLINEヤフー株式会社のChief Research Scientistで、人とコンピューターの関わり、とくにサイレントスピーチ（無声発話）と、ウェアラブル機器や少ないキーでの文字入力を研究しています。このサイトは個人の研究ホームページで、代表研究・論文を掲載しています。サイレントスピーチの論文レビューはGitHubで公開しています。詳しくは日本語の研究紹介へ。
- What this site is
- Naoki Kimura’s personal research homepage: silent speech interfaces, wearable/reduced-key text entry and publications, with links to his silent-speech paper review on GitHub and a separate CHI 2026 review collection.
- Who it’s for
- Collaborators, students, hiring readers, and answer engines looking for who he is, what he researches, and where authored work ends and literature reviews begin.
- Verdict
- Start with own research (SilentSpeller, SottoVoce, 3-Key-Input), then the silent speech hub for the category map, the question map for short Q→A routes, flagship reviews (SilentSpeller, SottoVoce), the Silent Speech Interface Review on GitHub for other groups' papers, CHI 2026 reviews (start with When Help Hurts), and 日本語の研究紹介 for the Japanese profile.
View CV · Email for research enquiries · All publications
Personal and research profile
Naoki Kimura is Chief Research Scientist at LY Corporation. He received his Ph.D. in Information Studies from The University of Tokyo, where he was advised by Jun Rekimoto and Thad Starner. This site is his personal and research homepage.
Email for research enquiries · View CV · GitHub profile · 研究の背景：クマ財団インタビュー（2024年4月・日本語） · Silent speech hub · Publications · Silent Speech Interface Review (GitHub) · CHI 2026 reviews · Question map · When Help Hurts review · 日本語の研究紹介
Research areas
Silent speech interaction
Interfaces for entering text or controlling voice-operated devices without speaking aloud, using tongue-contact sensing or ultrasound imaging.
Silent speech interfaces hub · SilentSpeller research and evidence · SottoVoce research and evidence · Question map (SSI + CHI)
Wearable and reduced-key text entry
Research on hands-free silent spelling and the possibilities and limits of text entry with fewer physical keys.
Earlier work: immersive visual experiences
Earlier research (2018–2019), not a current focus: ExtVision and Deep Dive explored generating or extending images and video for peripheral vision and head-mounted displays.
For practitioners: text without speaking aloud
Looking for how people type or enter text without speaking aloud, or where this work sits among silent speech research? The practitioner FAQ on typing without speaking maps research SSI routes versus everyday whisper/STT, and how this work sits among SSI research states an honest field position (HCI silent text entry and ultrasound interaction prototypes—not ultrasound SSI sensing pioneer claims).
Representative research
Own publications authored or co-authored by Naoki Kimura — not the literature-review collections below.
Rank-Residual Coding with a Shared Decoder: Lossless Low-Bit Communication for Ambiguous-Keyboard AAC
Forthcoming at IEEE SLT 2026 (accepted).
3-Key-Input: Exploring the Theoretical Minimum Keys for Text Entry
This work evaluates how far physical keyboards can be reduced when language-model-based disambiguation is used, finding that three keys can serve as a practical minimum in the evaluated offline English text-entry setting.
SilentSpeller: Towards mobile, hands-free, silent speech text entry using electropalatography
SilentSpeller reframes silent speech as silent spelling so that a wearable tongue-sensing retainer can support larger vocabularies, unseen words, and live text entry while walking.
SottoVoce: An Ultrasound Imaging-Based Silent Speech Interaction Using Deep Neural Networks
SottoVoce uses ultrasound imaging under the jaw to regenerate speech audio, letting an unchanged smart speaker stack handle recognition instead of building a bespoke end-to-end SSI product.
More selected publications
Mobile, Hands-free, Silent Speech Texting Using SilentSpeller
CHI 2021 Interactivity
CHI 2021 publication (DOI) · CHI 2021 paper (PDF, co-author's site)
SonoSpace: Visual Feedback of Timbre with Unsupervised Learning
SonoSpace maps instrumental timbre into a two-dimensional latent space so learners can compare their sound to expert performances without relying on verbal teacher feedback.
Silent speech paper review
Reviews of other groups' papers live on GitHub, separate from the own-research list above. The Silent Speech Interface Review compares silent speech papers by vocabulary, speaker dependence, cross-day testing and online use. The CHI 2026 review collection is a separate HCI reading guide. Below are notes on my own papers.
SilentSpeller: Towards mobile, hands-free, silent speech text entry using electropalatography
SilentSpeller is a strong, rigorously tested SSI system that reframes silent speech as silent spelling, enabling large vocabulary, live text entry, and walking robustness with in-mouth electropalatography sensors.
SottoVoce: An Ultrasound Imaging-Based Silent Speech Interaction Using Deep Neural Networks
A solid proof of concept that reconstructs speech audio from ultrasound for controlling unmodified smart speakers, showcasing important system design insight despite prototype limitations in latency, hardware bulk, and speaker dependency.
Education
Ph.D. in Information Studies, The University of Tokyo
Graduate School of Interdisciplinary Information Studies. Advisors: Jun Rekimoto and Thad Starner.
Master's degree with highest distinction, The University of Tokyo
Information Studies, Graduate School of Interdisciplinary Information Studies.
BA in Urban Engineering, The University of Tokyo
Honors
Official announcements and project reports are linked where available so you can check the research record.
Microsoft Research Asia D-CORE
Winner.
JSPS Research Fellowship for Young Scientists DC2
Research fellowship noted in the 2021 CV.
JST ACT-X
Frontier of mathematics and information science.
JST 2019 selected project JPMJAX190B — silent speech recognition (Japanese)
ACM CHI Honorable Mention Award
For SottoVoce, top 5%.
Rekimoto Lab publication and award record — SottoVoce, CHI 2019