YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

KR 한국어	US English	JP 日本語	CN 简体中文	ES Español
FR Français	DE Deutsch	RU Русский	SA العربية	IN हिन्दी

Johyo Jeun (`johyojeun-netizen`)

       __      __               __                     
      / /___  / /_  __  ______ / /__  __  ______       
 __  / / __ \/ __ \/ / / / __ \/ / _ \/ / / / __ \      
/ /_/ / /_/ / / / / /_/ / /_/ / /  __/ /_/ / / / /      
\____/\____/_/ /_/\__, /\____/_/\___/\__,_/_/ /_/       
                 /____/                                 
 Senior Systems & Real-Time Audio DSP Engineer · Seoul / Berlin

"Strict real-time safety, mathematical acoustic fidelity, and tactile physical sound modeling."

I am an audio digital signal processing (DSP) engineer and systems developer specializing in hard real-time spatial acoustics, binaural psychoacoustic rendering, and physical modeling sound synthesis. My primary work centers around developing low-latency audio plugins (CLAP / VST3), SIMD-accelerated filtering pipelines, embeddable audio runtimes for bare-metal/Linux platforms, and open mathematical tools for immersive multi-channel soundscapes.

Outside commercial plugin work, I build DIY Eurorack modules, conduct impulse response recordings in resonant architectural spaces, and optimize lock-free audio buffer architectures for sub-millisecond audio callbacks.

🧭 Engineering Philosophy

Hard Real-Time Audio Thread Safety: The audio callback is sacred. Zero dynamic memory allocations (malloc/new), zero mutual exclusion locks (mutex), zero file or network I/O, and zero thread synchronization that can trigger priority inversion or buffer under-runs (xruns).
Physical Modeling Over Static Sampling: Rather than storing multi-gigabyte static PCM sample tables, simulate the continuous differential wave equations of physical matter—resonant strings, acoustic air columns, vibrating membranes, and spatial room reflections.
Data-Oriented SIMD Processing: Structure audio buffer layouts as Struct-of-Arrays (SoA) to leverage 8-wide AVX2 / AVX-512 and ARM NEON vector instructions, processing blocks of floating-point audio data in continuous cache lines without branch mispredictions.
Open Standards & Modular Sound Graphs: Build against open standards such as the CLAP plugin format, non-interleaved IEEE 32/64-bit float streams, and modular node-graph engines that can run seamlessly from microcontrollers to high-density digital audio workstations.

🛠️ Core Technical Arsenal

Languages & Toolchains

Technology	Proficiency Level	Primary Use Cases
C++ (C++20 / C++23)	Production Daily	Real-time audio engines, DSP algorithms, CLAP/VST3 plugins, low-overhead templates
Rust	Production Daily	Memory-safe DSP toolchains, lock-free audio ring buffers, WebAssembly audio modules
C (C99 / C11)	Advanced	Embedded microcontrollers, bare-metal audio codecs, platform abstraction layers
Zig	Intermediate	Cross-compilation toolchains, standalone audio CLI utilities, lightweight runtimes
Python	Proficient	Filter coefficient calculation, impulse response deconvolution, acoustic modeling
POSIX Shell / Make / CMake	Proficient	Multi-platform build systems, real-time Linux kernel configuration, reproducible benchmark suites

DSP, Audio Architecture & Acoustics

Domain	Focus Areas & Technologies
Spatial Acoustics & HRTF	High-Order Ambisonics (HOA), SOFA format HRIR convolution, spherical head shadow models
Synthesis & Physical Modeling	Digital Waveguide Mesh (DWM), Karplus-Strong string synthesis, nonlinear state-variable filters
Filter Design & Spectral Math	Polyphase FIR/IIR resamplers, FFTW3 / pffft, Hilbert transform, zero-delay feedback (ZDF)
Audio Plugin Frameworks	CLAP (Clever Audio Plugin) API, JUCE 8, VST3 SDK, WebAudio AudioWorklet
Real-Time Platforms	Linux RT-PREEMPT kernel tuning, ALSA / JACK / PipeWire, Bela Platform, Teensy 4.1

🚀 Key Open-Source Projects

1. `reson-dsp` — Lock-Free Physical Modeling Waveguide Synthesizer

Language: C++23 / CLAP API
Description: A polyphonic physical modeling sound engine simulating acoustic resonance in struck metallic bars, bells, and nonlinear taut strings using 2D digital waveguide networks.
Key Characteristics:
- Deterministic $O(1)$ voice allocation using compile-time bounded lock-free circular queues.
- Zero heap allocation throughout the entire runtime processing lifecycle.
- Native AVX-512 and ARM NEON vectorized biquad filter cascades running at 192 kHz with less than 2.8% CPU overhead on a single core.

2. `spatial-orb` — Real-Time Binaural Ambisonics 3D Spatializer

Language: Rust / WebAssembly
Description: A 3D spatializer engine converting multi-channel High-Order Ambisonics (up to 5th-order HOA, 36 spherical harmonics) into binaural headphone audio with custom HRTF profiles.
Key Characteristics:
- Partitioned overlap-save fast convolution supporting arbitrary length impulse responses under 1.2ms latency.
- Smooth parametric Doppler shift and distance attenuation without audible zipper noise.
- Dual-target compilation: native static library (.a / .so) and WebAssembly for WebAudio AudioWorklet integration.

3. `wav-crunch` — SIMD 64-bit Audio Analysis & Polyphase Resampler CLI

Language: Rust
Description: A zero-dependency command-line utility for bit-accurate audio analysis, inter-sample peak detection (True Peak via 4x oversampling), and high-fidelity sample rate conversion.
Key Characteristics:
- Polyphase sinc interpolation with dynamic window attenuation exceeding 144 dB signal-to-noise ratio (SNR).
- Multi-threaded batch processing leveraging cache-aligned work-stealing thread pools.
- Comprehensive export capabilities: LUFS loudness analysis (ITU-R BS.1770-4), dynamic range meters, and phase correlation matrices.

4. `patch-wire` — Micro-Modular Audio Signal Graph Runtime

Language: C++20 / C
Description: An embeddable directed acyclic graph (DAG) audio computation graph tailored for embedded Linux systems (Raspberry Pi CM4, Bela) and headless Eurorack digital modules.
Key Characteristics:
- Topological sorting performed once at graph configuration; zero branch evaluation inside the inner audio buffer block loop.
- Strict bounded memory footprint: entire node graph memory mapped to a single contiguous cache-aligned arena.
- Hardware CV/Gate control voltage I/O synchronization via real-time SPI and I2C ADC bus drivers.

📊 Real-Time Audio Processing Architecture

[ Hardware Audio Driver (ALSA / ASIO / CoreAudio) ]
                        |
                        v (Interrupt / Callback)
+-------------------------------------------------------------------+
|                     Audio Thread Callback                         |
|                                                                   |
|   +-----------------------------------------------------------+   |
|   | 1. Lock-Free SPSC FIFO Input Ingestion (Zero Lock / Wait) |   |
|   +-----------------------------------------------------------+   |
|                               |                                   |
|   +-----------------------------------------------------------+   |
|   | 2. SIMD Block Processing (Vectorized DSP Filters / DWM)   |   |
|   |    - 64-sample buffer chunks                              |   |
|   |    - SoA floating-point registers (AVX-512 / NEON)        |   |
|   +-----------------------------------------------------------+   |
|                               |                                   |
|   +-----------------------------------------------------------+   |
|   | 3. Binaural HRTF Partitioned Fast Convolution             |   |
|   |    - Time-domain overlap-save buffers                     |   |
|   +-----------------------------------------------------------+   |
|                               |                                   |
|   +-----------------------------------------------------------+   |
|   | 4. Saturation Limiter & Dither to DAC Output Ring Buffer  |   |
|   +-----------------------------------------------------------+   |
+-------------------------------------------------------------------+
                        |
                        v
          [ Physical DAC / Headphone Out ]

⏱️ Target Latency & Benchmarks

All benchmarks measured on Linux kernel 6.11 with PREEMPT_RT patch, Intel Core i7-13700K (fixed 4.8 GHz clock) and AMD Ryzen 9 7950X:

Audio Buffer Size	Sample Rate	Round-Trip Latency (Hardware I/O)	Worst-Case DSP Block Time
32 samples	96 kHz	0.67 ms	18.4 µs
64 samples	96 kHz	1.33 ms	34.2 µs
128 samples	192 kHz	1.33 ms	56.8 µs
256 samples	48 kHz	5.33 ms	41.1 µs

🔬 Studio, Acoustic Rig & Hardware Lab

Category	Equipment & Instruments
Monitoring & Headphones	Neumann KH 120 II DSP Monitors, Sennheiser HD 800 S (calibrated diffuse-field diffuse EQ), Beyerdynamic DT 1990 Pro
Acoustic Measurement	Earthworks M30 measurement microphone, Ambisonics 1st/2nd-order tetrahedral arrays, REW (Room EQ Wizard)
Hardware Synthesizers & Modular	Custom DIY 9U Eurorack system (analogue VCOs, Serge wave multipliers, custom Teensy DSP modules)
Measurement & Debugging Tools	Rigol DS1054Z 4-channel Digital Oscilloscope, Saleae Logic Pro 8 logic analyzer, Audio Precision APx515
Embedded Computing Testbeds	Bela Mini Starter Kit, Teensy 4.1 Audio Shield rev D, Raspberry Pi Compute Module 4 with custom HiFiBerry DAC+ ADC Pro

📬 Connect & Collaborate

GitHub Issues & Discussions: Preferred channel for open-source bug reports, algorithm RFCs, and DSP mathematical discussions.
Code Signing & PGP Fingerprint: 4A8F B120 79E3 5DC6 D982 314F 88B1 C990 2A3E 761F
Location: Seoul, Republic of Korea / Berlin, Germany (UTC+9 / UTC+1)
License Preference: MIT License (for algorithm utilities) / ISC or BSD-2-Clause (for embedded core headers)

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Johyo Jeun (johyojeun-netizen)