The archive · Product Ideas · Technical decision · 2020s
See Sound translates household noises into a visual readout for Deaf users
See Sound used machine learning and two million YouTube sound clips to identify 75 household noises and show them visually.
See Sound
What it had to solve
Millions of deaf and hard-of-hearing people had no home product that could distinguish a microwave from a baby crying. The challenge was to turn ambiguous sound into a simple visual cue.
How it works
Deaf and hard-of-hearing people can miss crucial household information because ordinary products do not distinguish one sound from another. A microwave, a baby crying and other events can all remain invisible without an alternative signal.
See Sound answered with a smart-home hearing system. It verified noises against a data library and used a machine-learning model powered by two million YouTube sound clips to translate them into a simple visual readout.
The system reported 75 sounds with high accuracy, turning audio recognition into a visual accessibility layer. Its data-driven design also left room for sensitivity to improve as the library was mined more deeply.
Why it lands
- It solves a concrete accessibility gap at the moment a household event happens.
- The visual output is simpler to act on than asking users to interpret raw technical data.
- A large sound library makes the product more useful as it encounters more real-world variation.
- Machine learning is used for translation and independence, not as a novelty layered onto an existing product.
What it did
A smart-home hearing system made household audio legible without relying on hearing, creating a visual layer for 75 sounds and a model that could improve through deeper data mining.
What you can take
Accessibility improves when the system translates information into the user’s strongest channel instead of asking the user to compensate for a missing one.
Since then
The D&AD jury highlighted the deep mining of data as the concept’s standout feature: sensitivity could keep improving while the product made sound detection useful to people who could not hear it.
Sources
spotted an error? The archive wants to know.
Your turn
You just read one. Describe the brief you are staring at, and see who has been given the same problem.
Free account · 3 free questions · no card