Research

Neural acoustic modelling and sound field control for spatial audio.

My work explores how machine learning and acoustic knowledge can be combined to model sound fields and control their reproduction through loudspeaker arrays. I am particularly interested in the relationship between acoustic reconstruction accuracy and downstream reproduction performance.

Neural acoustic modelling

I study continuous neural representations of binaural room impulse responses (BRIRs), including sinusoidal representation networks (SIREN), physics-informed architectures, and frequency-dependent modelling. I evaluate both reconstruction accuracy and the effect of model errors on spatial audio reproduction.

  1. Sub-band Neural Interpolation of Binaural Room Impulse Responses for Personal Sound Zone Control

    Yazhou Li, Lin Wang, Joshua Reiss

    6th AES International Conference on Audio for Virtual and Augmented Reality and Immersive Games, Paris · 2026

    Publisher Code Slides
    BibTeX
    @inproceedings{li2026subband,
  title = {Sub-band Neural Interpolation of Binaural Room Impulse Responses for Personal Sound Zone Control},
  author = {Li, Yazhou and Wang, Lin and Reiss, Joshua},
  booktitle = {6th AES International Conference on Audio for Virtual and Augmented Reality and Immersive Games},
  address = {Paris},
  year = {2026},
  organization = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=23307},
}

  2. Binaural Room Impulse Responses Interpolation using Physics-Informed Neural Networks in Three Dimensions

    Yazhou Li, Lin Wang, Joshua Reiss

    51st German Annual Conference on Acoustics (DAS|DAGA), Copenhagen · 2025

    Publisher PDF Poster
    BibTeX
    @inproceedings{li2025pinn,
  title = {Binaural Room Impulse Responses Interpolation using Physics-Informed Neural Networks in Three Dimensions},
  author = {Li, Yazhou and Wang, Lin and Reiss, Joshua},
  booktitle = {51st German Annual Conference on Acoustics (DAS|DAGA)},
  address = {Copenhagen},
  year = {2025},
  url = {https://pub.dega-akustik.de/DAS-DAGA_2025/konferenz-1672.html?article=108},
}

Personal sound zones

I investigate neural-network-based filter design for personal sound zones and transaural reproduction. My work connects sound field estimation to control, benchmarking neural methods against conventional optimisation and examining their robustness when room acoustic models are imperfect.

  1. Neural Regularization for Personal Sound Zones

    Yazhou Li, Lin Wang, Joshua Reiss

    6th AES International Conference on Audio for Virtual and Augmented Reality and Immersive Games, Paris · 2026

    Publisher Code Slides
    BibTeX
    @inproceedings{li2026regularization,
  title = {Neural Regularization for Personal Sound Zones},
  author = {Li, Yazhou and Wang, Lin and Reiss, Joshua},
  booktitle = {6th AES International Conference on Audio for Virtual and Augmented Reality and Immersive Games},
  address = {Paris},
  year = {2026},
  organization = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=23308},
}

  2. Impact of Mismatched Room Acoustic Modeling on Transaural Reproduction with Loudspeaker Arrays

    Yazhou Li, Lin Wang, Joshua Reiss

    155th Audio Engineering Society Convention, New York · 2023

    Publisher
    BibTeX
    @inproceedings{li2023impact,
  title = {Impact of Mismatched Room Acoustic Modeling on Transaural Reproduction with Loudspeaker Arrays},
  author = {Li, Yazhou and Wang, Lin and Reiss, Joshua},
  booktitle = {155th Audio Engineering Society Convention},
  address = {New York},
  year = {2023},
  organization = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=22259},
}

Personalised spatial audio

During my research internship at the National Institute of Informatics, I developed an HRTF-conditioned neural framework for personalising BRIRs to unseen listeners. Earlier work at Peking University mapped features from 3D head meshes to HRTF representations, using data from more than 400 subjects.

  1. Binaural Room Impulse Responses Personalization From Head-related Transfer Function Conditioning

    Yazhou Li, Shoichi Koyama, Lin Wang, Joshua Reiss

    Manuscript under submission · 2026

Audio software and earlier projects

Alongside my research, I develop audio software. At PASACO Studio (2024), I worked on a real-time iOS audio engine using AVAudioEngine and C++ buffers, together with a modular SwiftUI interface. My audio software projects received the Most Fun Award at Muse Hub Hackathon 2024 and the Best Creative Hack Award at Timbre Tools Hackathon 2024.

I also co-developed Daily Life Vocoder, a mobile vocoder sequencer for exploring everyday sounds and musical timbres.

Browse earlier audio projects →