National Institute of Informatics / SOKENDAI

S. Koyama's Lab

skoyamalab-secnii.ac.jp

NII S. Koyama's Lab

Welcome to Audio Processing Research Group at National Institute of Informatics (NII), Tokyo, Japan! This group led by Prof. Shoichi Koyama was created in 2023.

Our primary research interests lie in signal processing and machine learning in audio and acoustics. Specifically, we study the analysis and control of acoustic fields, which bridges wave physics and machine learning. Our work is validated through real-world systems, with applications in VR/AR audio, noise control, acoustic imaging, and more.

We are always looking for motivated Ph.D. students, interns, PostDocs, and collaborators. Check out the Opportunities page if you are interested!

Keywords

Acoustic signal processing, machine learning for audio, microphone and loudspeaker array processing, spatial audio

News

Mar 16, 2026 Ms. Yazhou Li (Queen Mary University of London, UK) joined our lab as an intern 🚀
Feb 12, 2026 A new paper has been published in IEEE Transactions on Audio, Speech and Language Processing 🎉
Jan 13, 2026 A new article has been published in IEICE Journal 🎉
Dec 2, 2025 Prof. Koyama gave an invited talk at SIG-Challenge Workshop 🧑‍🏫
Nov 27, 2025 Presentation at ASA/ASJ Joint Meeting 🇺🇸

Selected works

2025

  1. Spatial Upsampling of Head-Related Transfer Function Using Neural Network Conditioned on Source Position and Frequency
    IEEE Open Journal of Signal Processing, 2025
  2. fnt_sfest.jpg
    Sound Field Estimation: Theories and Applications (Foundations and Trends® in Signal Processing)
    2025
  3. piml_sfest.png
    Physics-Informed Machine Learning For Sound Field Estimation: Fundamentals, state of the art, and challenges
    Shoichi Koyama, Juliano G. C. Ribeiro, Tomohiko NakamuraNatsuki Ueno, and Mirco Pezzoli
    IEEE Signal Processing Magazine, 2025
  4. Physics-Informed Machine Learning For Audio Processing
    In European Signal Processing Conference (EUSIPCO), Tutorial, Sep 2025

2024

  1. Sound Field Estimation Based on Physics-Constrained Kernel Interpolation Adapted to Environment
    Juliano G. C. Ribeiro, Shoichi Koyama, Ryosuke Horiuchi, and Hiroshi Saruwatari
    IEEE/ACM Transactions on Audio, Speech, and Language Processing, Sep 2024