Your Pocket Research Whisperer – Built with ESP32-S3 + n8n AI
A Pocket-Sized AI That Reads Papers to You 📚🔊
Just a fun little weekend project—but it turned out surprisingly powerful.
Imagine carrying a personal research assistant in your pocket. One that searches papers, summarizes them, syncs notes to Notion, and even reads them out loud—all without needing a phone or computer.
With an ESP32-S3 microcontroller and an n8n-powered AI agent, I built a low-cost voice assistant that does exactly that.
🎒 Imagine this:
You’re walking around with a tiny device in your pocket that:
- 📡 Connects to Wi-Fi autonomously
- 🔍 Searches for academic research
- ✍️ Generates summaries using AI
- 📤 Sends notes directly to your Notion workspace
- 🔊 Speaks the summary aloud using on-device audio playback
All you need is Wi-Fi—once it’s connected, the agent runs autonomously. No phone, no laptop, just plug in and go.
⚙️ Under the Hood: What Makes It Tick
Here’s a breakdown of the tech stack and magic behind it:
-
Voice Recording & Audio Handling
Using theRecorder
module, the device captures raw audio, converts it into WAV format with a custom MicroPythoncreate_wav_from_raw()
function, and sends it to the OpenAI Whisper API for speech-to-text transcription. -
Natural Language Understanding
Transcribed text is sent to a custom GPT agent hosted via n8n webhook, which intelligently responds in the same language as input — even handling Unicode text like Japanese or Chinese. -
Screen Display
The device displays both the user’s voice input and the AI response on the onboard LCD, with colored, paginated output for clarity and flair. -
Touch Control
Start and stop recording with a simple screen tap. No buttons, no UI overhead — just responsive design. -
Audio Playback
The system can also broadcast summaries as audio, making it the perfect assistant for hands-free reading while walking or commuting.
All of this runs without any external device. No phone. No laptop. Just Wi-Fi.
10 minutes to set up, a lifetime of geeky joy.
📸 Here’s a peek at the actual build:
Youtube video
This was built in a single day, but I’m just getting started. I’ll be sharing:
- 🧠 The full MicroPython codebase
- 🔗 My n8n automation workflows
- 🔊 Audio playback enhancements (featuring the ES8311 DAC)
Whether you’re into automation, embedded AI, or just love building cool stuff—stay tuned.
Hashtags for the curious: #Microcontroller #ESP32S3 #AIVoiceAssistant #Notion #MicroPython