Selected Works · 2021 — 2026
Portfolio Issue 01

Tao Yun
Huang 黄 韜 云

Creative technologist. Real-time systems for spatial audio, machine learning, and live performance — on stage, in installation, at scale.

Taipei · Available worldwide
bernie5@dataraw.tech
02 — About Index · I
Practice

Between engineering and live experience, I build systems that have to work once —
in front of an audience,
without a retry.

My practice covers spatial audio, real-time machine learning, and mixed reality. From hardware control and sensor fusion, through neural audio synthesis and multichannel rendering, to cross-engine AV pipelines — I work across the full stack of systems that turn technological possibility into physical presence.

Audio & ML

  • Spatial Audio · Ambisonics
  • Real-time Neural Synthesis
  • RAVE · Custom models
  • Max/MSP · Pure Data
  • Dante · multichannel routing

XR & Systems

  • Unity · ARKit · Swift
  • TouchDesigner
  • UWB sensor fusion
  • YOLO · Computer Vision
  • Adversarial ML

Hardware & Code

  • Python · PyTorch · C++
  • Motor & actuator control
  • Embedded systems
  • UDP / OSC protocols
  • Hardware prototyping
2010 — 2013
Feng Chia University
逢甲大學

B.Eng. Electrical Engineering. Signal processing, embedded systems, control theory.

2014 — 2019
Nat'l Taiwan University of Arts
國立臺灣藝術大學

B.A. Drama & Theatre Arts. Performance, dramaturgy, staging.

2020 — 2024
NYCU — Institute of Music
國立陽明交通大學

MFA Multimedia Music (New Music Theatre). Advisor: Prof. Chao-Ming Tung.

Part · One

Commercial Work

商 業 製 作
03 — Commercial · 01 if.plus / work / 208
Commercial · 01

AI Living Lab

MODA · 2025 資訊月 · 台北
CLIENT   Ministry of Digital Affairs (MODA)
VENUE   Taipei Computer Festival 2025
STUDIO   IF Plus Art
AUDIENCE   1,000+ participants
DURATION   4 days
YEAR   2025
Interactive cycling station, MODA booth
Fig. 01 — Interactive cycling station, MODA booth

Pedal, choose a route, answer a prompt. The system generates a one-of-a-kind audio souvenir — cassette postcard — from your ride.

Visitors mounted a custom-built stationary bike and turned the handlebars to choose a virtual route, answering short prompts about how AI shows up in daily life. Pedaling speed, route choices, and prompt responses streamed in real time to an AI music-generation model I developed and trained specifically for this project.

At the end of each ride, the system printed a one-of-a-kind cassette postcard — an audio souvenir generated uniquely by the rider's physical and semantic input. Over four days, more than 1,000 visitors rode the bike, producing everything from aggressive electronic tracks to ambient loops.

My role

AI music model development & training · pedaling sensor data acquisition & processing · real-time program integration · custom bike hardware design.

Python · Generative Audio ML · Real-time synthesis · Hardware sensors · Cassette print system
04 — Commercial · 02 if.plus / work / 209
Commercial · 02

Diesel-Electric Workshop

國家鐵道博物館 柴電工場
CLIENT   National Railway Museum, Taipei
EXHIBITION DESIGN   archicake
DIGITAL   IF Plus Art
YEAR   2025
TYPE   Permanent exhibition

A permanent exhibition staging the invisible: how energy flows through a 100-year-old locomotive repair hall.


For the S212 Diesel-Electric locomotive, a large-format transparent screen sits in front of the actual machinery, animating the flow of electric power while keeping the physical mechanisms visible behind — an X-ray overlay aligning virtual animation to real hardware.

For the armature maintenance area, a full-scale 60 kg armature model is choreographed with the overhead crane, replaying the actual motion sequence of factory workflow.

My role

Hardware engineering — armature & overhead crane motion control · transparent display control · multi-system timing synchronization between animation playback and physical actuation.

Motor control · Transparent LED · Crane kinematics · Animation sync · Hardware timing
S212 transparent screen
Armature and crane
Workshop floor
05 — Commercial · 03 if.plus / work / 199
Commercial · 03

Moss Floating Layer

苔域浮層 · Shin Kong Mitsukoshi A9 · 台北
CLIENT   SCOPE TECH
VENUE   Shin Kong Mitsukoshi A9, Taipei
STUDIO   IF Plus Art
YEAR   2025
Moss terrarium installation, Xinyi district
Fig. 01 — Moss terrarium installation, Xinyi district
Concept

Against the acceleration of the city,
a small terrarium breathing
at its own pace.


In the middle of one of Taipei's busiest commercial districts — an immersive moss ecosystem terrarium. A quiet counterpoint to retail velocity. Visitors stop, look up, and listen.

My role — sound design

Multichannel spatial music composition and system design for the installation's audio environment.

Multichannel audio · Spatial composition · Retail AV integration
06 — Commercial · 04 if.plus / work / 206
Commercial · 04

Culture-Tech Exchange

文化科技交易所 · TTXC INNOVATIONS 2025 · 高雄
CLIENT   TAICCA (Taiwan Creative Content Agency)
VENUE   Pier-2 Art Center, Kaohsiung
STUDIO   IF Plus Art
TOTAL VIRTUAL INVESTMENT   NT$4.7 billion
DURATION   4 days

An exhibition turned into a live investment game. Every visitor gets NT$10M virtual seed capital.


Via a LINE OA app, visitors invest in culture-tech works on display; the aggregated behavior reveals real market signal back to the cultural agency. Phone-based investments sync instantly to venue screens — a live investment ticker, and a top-10 leaderboard refreshing every 3 minutes.

Over four days, total virtual investment hit NT$4.7 billion — a behavioral dataset reflecting the public's appetite for emerging creative-tech works.

LINE OA venue leaderboard
My role

Unity-based real-time software development · cross-platform interaction integration (mobile ↔ venue display) · live-data visualization sync.

Unity · LINE OA · Real-time networking · Live data viz
Part · Two

Research & Practice

研 究 與 創 作
07 — Research · 01 Series · 3 Versions · 2022 — 2024
Research · 01

Threshold

闁 · Electromechanical Neural Audio Series

A three-version series examining recursive feedback between real-time neural audio synthesis and physical resonance systems.

Threshold v1, Taichung 2022

v1 · Taichung 2022

Site-specific

Collected and algorithmically transformed physical space vibration and environmental audio in real time.

Threshold v2, C-LAB Taipei 2023

v2 · C-LAB Taipei 2023

Latent Space

Modified RAVE model for real-time latent-space audio reconstruction across a 49.4-channel Ambisonics system — first deployment of real-time neural audio at this scale in Taiwan's institutional sound-art context.

Threshold v3, electromechanical apparatus 2024

v3 · NYCU 2024

Electromechanical

Gate-shaped apparatus: brass plate driven by stepper motors, solenoids, and pneumatics forms a continuous feedback loop with RAVE synthesis; k-means redistributes output into a five-voice generative score.

My role — solo

Composition · modified RAVE pipeline · electromechanical apparatus design · Ambisonics mapping · feedback-loop engineering across all three versions.

DOCUMENTATION   youtu.be/YVT2q0_gJyA
08 — Research · 02 Digital Art Center Taipei
Research · 02

Circle of Confusion — Reaction

彌散圈反應 · VR Installation · 2024
Viewer wearing custom 3D-printed haptic VR rig

A VR installation exploring the photographic notion of the "circle of confusion" — mediation, diffusion, blur, process.

Music triggers cascading algorithmic transformations inside a non-representational VR environment. A custom 3D-printed sculpture attached to the VR headset translates virtual image into tactile sensation.

The spatial audio uses a deliberately incomplete Ambisonics array — speaker absences become bidirectional "channels," thresholds through which the viewer becomes self-aware of their body against the technical field. A CO₂ / O₂ gas system gives the installation a respiratory rhythm.

Medium

VR · generative algorithm · Ambisonics · haptic · gas

Role

Solo — concept · system · composition · 3D print

Venue

Digital Art Center, Taipei · 2024

Documentation

youtube · YpsZ0EtSozE

09 — Research · 03 NYCU · iOS AR
Research · 03

River of Shadows

川影 · Site & Time-Specific iOS AR · 2024
AR overlay, botanical garden fragment on corridor

A site-specific, time-specific iOS AR performance. At a designated moment, audiences open the app inside the ground floor of the university's administrative building.


Area Target spatial recognition activates the performance from a first-person perspective. The work superimposes two absent spaces onto the physical corridor — the artist's private room and the desert pavilion of Taichung Botanical Garden. The room functions as a negative space — a chroma-key cut-out subtracted from the visible field, producing a ruptured scene. A moving white cube becomes a navigation score, directing each viewer's physical movement through the corridor in sync with spatial electroacoustic composition.

My role — solo

iOS AR app development · Area Target spatial recognition · spatial electroacoustic composition · audience navigation choreography.

Swift · ARKit · Area Target · Spatial audio · Doc
10 — Current · In Development Submitted · ZKM Karlsruhe 2026/27
Current · In Development

LATENT BREACH

Adversarial Subspaces as Performative Resistance
Concept visualization, adversarial AR entity in shared airspace

A 20-minute live performance in which every audience member's smartphone becomes a sensing instrument.

An iOS / Android AR app runs real-time YOLO26 object detection while six UWB beacons track each device to centimeter-level precision. For the first ten minutes, the system classifies the room accurately — building trust.

At minute ten, PGD adversarial perturbation collapses the detection confidence, and an AR entity materializes in the shared airspace. Object labels are replaced by site-specific narrative text authored from archival research. The machine does not malfunction. It narrates.

Status

Residency proposal · ZKM Karlsruhe 2026/27 · Rauschenberg Residencies

Stack

YOLO26 · UWB · ARKit · PGD · PyTorch · 8-ch audio

Role

Solo — system · narrative · adversarial pipeline

Full document

View PDF

Get in touch

Open to full relocation worldwide.
Available immediately.

+886 983 717 444 · Taipei
LinkedIn GitHub Works
EU Blue Card eligible   ·   Canada Global Talent Stream eligible