Voice interface
VoiceKey
An M5StickS3 voice-input experiment: hold a button, speak, transcribe, and pass the result to a Mac as USB keyboard input.
ESP32-S3 · Firmware · Applied AI
Small devices.
Real utility.
WangLab is developing practical smart hardware where embedded systems, voice interfaces, and AI-assisted firmware development meet.
Prototype 01Voice interface
Physical input,
intelligent output.ESP32-S3
USB · Wi-Fi · Audio
intelligent output.ESP32-S3
USB · Wi-Fi · Audio
01 / ACTIVE EXPERIMENTS
Hardware experiments designed around a clear job.
AI companion firmware
Connected gadgets
Firmware experiments for compact AI companions, exploring resilient network paths, CJK display support, and device-level interaction on constrained hardware.
02 / CAPABILITIES
From breadboard uncertainty to firmware that behaves.
WangLab takes a build-first approach. The focus is not speculative hardware—it is compact tools that can be flashed, tested, carried, and used.
01
Embedded firmware
ESP-IDF components, FreeRTOS task design, peripheral control, and careful debugging under real memory and timing limits.
02
Voice and interaction
Button-first voice capture, transcription pipelines, compact displays, and interfaces built for devices without room for complexity.
03
Connected device systems
Wi-Fi provisioning, proxied connectivity, API integration, and failure handling that starts at the device—not in a slide deck.
AI in the development loop.
WangLab plans to use Claude as a firmware co-developer: drafting ESP-IDF components, reviewing embedded C/C++, tracing concurrency failures, and shortening the path between an idea and a tested build.
$ wanglab build voicekey
→ inspect board constraints
→ review task ownership
→ compile ESP-IDF target
✓ firmware image ready
# Human decisions stay in the loop.
$ flash → test → observe → refine
03 / CONTACT
Building something small, useful, and technically stubborn?
Share the device, constraint, or prototype you are working through.