AI Decodes Mechanical Keyboards by Ear

https://hackster.imgix.net/uploads/attachments/1993511/_AhrZIwniLV.blob?auto=compress%2Cformat&w=600&h=450&fit=min

Machine learning on microcontrollers has been a hot topic for the past several years, but most of the projects we see are still pretty simple, like keyword spotting or gesture recognition. Psychogenic Technologies shows just how far these tiny devices have come by building two very different edge AI projects: one that can identify which key was pressed on a mechanical keyboard simply by listening to it, and another that transforms a pair of boxing gloves into wireless motion controllers capable of recognizing punches in real time.

Both projects are built around Nordic Semiconductor's Edge AI ecosystem, but they demonstrate two different approaches to running neural networks on embedded hardware. One relies on dedicated AI acceleration, while the other shows that, for the right workload, even a modest processor is capable of getting the job done.

A closer look at the hardware (📷: Psychogenic Technologies)

Most people will likely find...

Copyright of this story solely belongs to hackster.io. To see the full text click HERE

Read more

https://9to5mac.com/wp-content/uploads/sites/6/2026/05/apple-openai.jpg?quality=82&strip=all&resize=1200,628

Apple v. OpenAI: Apple asks the court to let its own experts review forensic images used in the case, and seeks documents tied to parts of OpenAI's hardware R&D

Sponsor Posts Subquadratic: the LLM built for 12M-token reasoning — SubQ can reason across entire codebases and document sets in one pass with no RAG workarounds. Read how SubQ 1.1 Small holds near-perfect retrieval out to 12M tokens. Cloudflare Connect 2026: The Agentic Conference — Stop preparing for agents. Start running