August 8, 2026
Talk to the hand... carefully
Show HN: Sign language translation with smart glasses
Cool demo or total miss? Commenters say it’s spelling, not full sign language
TLDR: A maker demoed smart glasses that read hand-spelled signs, but commenters pushed back hard on calling it full sign language translation. The big debate: this could be a useful tool, yet many say it misses the culture and complexity of how Deaf people actually communicate.
A Hacker News maker showed off smart glasses that can read hand signs and turn them into text through a web app and iPhone setup, pitching it as sign language translation. On paper, it sounds like a classic feel-good future gadget moment. In the comments? Instant chaos. The thread quickly swerved from “neat demo” to a full-on reality check, with Deaf-aware commenters saying the project was being oversold in a big way.
The loudest reaction was basically: this is not the same thing as translating sign language. Multiple people stressed that the demo appears to recognize fingerspelling—spelling words out by hand—while actual sign languages also rely on facial expressions, body movement, grammar, tone, and context. One commenter delivered the thread’s main mic-drop with “fingerspelling != sign language,” then launched into why that difference matters. Another brutally concise hot take? “anything but learn sl,” which landed like a meme and a roast at the same time.
The deeper drama wasn’t just about accuracy—it was about who gets to build disability tech and how. Several commenters warned that disabled and Deaf people often get “helpful” inventions built for them by outsiders who don’t fully understand their lives. So while the demo impressed on a technical level, the community mood was clear: cool idea, but don’t confuse a narrow feature with solving communication. In other words, the gadget got attention, but the comments stole the show.
Key Points
- •Hand Wave is presented as a smart-glasses sign recognition project targeting web and iOS.
- •The system uses a small Python inference service as its backend component.
- •Setup requires pnpm 10.33+, Node 22+, Python 3.11/3.12, uv, and moon.
- •The project requires configuration of VITE_INFERENCE_URL in the environment file before running.
- •The post cites multiple research works related to ASL fingerspelling, handshape recognition, and action recognition.