Never Forget a Million-Dollar Idea Again

You never know when creativity might strike, but one thing is certain: it rarely seems to happen when a notepad and pen are within reach. The best ideas usually pop into our heads when we are otherwise occupied — perhaps when we are driving or out for a jog. That can make it hard to capture our thoughts before they slip away.

A mechanical engineer who goes by the handle “The Raccoon Lab” has run into this situation too many times to count. He is fed up with his ideas vanishing, so he came up with a solution. It is a dedicated voice recorder that can capture an idea anywhere — you just press a button and speak to use the device. It transcribes the recordings and generates summaries to make it easier to digest those voice notes later on.

The recorder was designed around an inexpensive ESP32 development board. An INMP441 digital microphone captures audio over an I2S connection, while recordings are stored locally on an SD card. A rechargeable LiPo battery powers the system, and a small display provides menu navigation and playback information.

To give the device some personality, The Raccoon Lab included a spinning external control disc driven by a small DC motor. A magnetic encoder beneath the motor tracks its position in real-time. Users can physically stop the spinning wheel with their fingers to pause playback or recording, then release it to continue. Spinning the wheel rapidly in reverse or forward produces exaggerated “chipmunk-style” audio effects.

The hardware was assembled on a DIY PCB that was made with the help of a 3D printer. After printing the substrate, copper foil was carefully applied to both sides and trimmed by hand using a hobby knife to isolate the electrical pathways. Dozens of holes were drilled manually, and metal pins were soldered through the board to create electrical vias connecting the two layers.

The enclosure follows a retro cassette-inspired aesthetic with minimalist Bauhaus influences. Flexible “live hinge” buttons were printed directly into the shell, eliminating the need for separate moving parts. After extensive sanding, filler application, and paint work, the finished device looks closer to a polished consumer product than a homemade prototype.

Once recordings are captured, the ESP32 hosts its own Wi-Fi access point so nearby devices can retrieve audio files wirelessly. Those recordings are then processed through OpenAI’s Whisper transcription model before being summarized using Anthropic’s Claude API, turning spontaneous spoken ideas into concise, readable notes.