DataVonix

sensa ai memory wearable

An AI personal-memory wearable we are building from scratch.

Industry
Consumer hardware
Category
AI Products
Core stack
ESP32-S3 · Hardware
  • What it is

    An AI personal-memory wearable (prototype)

  • Who it's for

    People who want to remember their conversations

  • What it does

    Captures conversations to build a searchable memory; audio capture works today

the problem we were asked to solve

Personal-memory wearables are a new category. We are building one from first principles, starting with the part that is hardest to get right: clean audio from a very small device.

  • Audio captureSpeech is recorded to the device's flash memory as WAV files.
  • Wi-Fi control pageThe device runs its own access point and web page.
  • Start and stop from the pageRecording is controlled from any phone or laptop.
  • Download over Wi-FiRecordings come off the device without a cable.
  • Clear speechThe microphone captures clear, usable speech.

how it works, step by step

Chip

ESP32-S3

Microphone

INMP441, over I2S

Version 0.1

Records clear speech

  1. 01

    Capture

    The device's microphone records clear speech.

  2. 02

    Store

    Recordings are saved as WAV files on its flash memory.

  3. 03

    Control

    Recording starts and stops from the device's own Wi-Fi page.

  4. 04

    Download

    Files come off the device over Wi-Fi.

see it in action

Sensa AI Memory Wearable: the main screen
The main screen
Sensa AI Memory Wearable: on a phone
On a phone
Sensa AI Memory Wearable: in the browser
In the browser

how it's built

An ESP32-S3 board with 16 MB of flash and 8 MB of PSRAM and an INMP441 I2S microphone, powered over USB-C, with firmware written in C++.

This is version 0.1. The phone app, backend, transcription and the AI memory itself come next.

Want a Similar Project?

Tell us about your idea and we'll build something you're proud of.

let us make the web talk about your project

Send over the problem, the deadline, or the half-finished system. We come back with an honest scope, an architecture direction, and a number.