I built this after getting frustrated with losing great ideas while driving, cooking, or exercising. You know that moment when inspiration strikes but your hands are occupied?
The Problem: Doubao AI earphones (popular in China, similar to AirPods but with built-in AI) are great for Q&A, but all conversations disappear after listening. It's a closed ecosystem with no way to export valuable content.
My Solution: A Python tool that monitors the Doubao web interface and automatically syncs voice notes to Obsidian in real-time.
What makes it interesting:
30+ speech variation recognition - Works even if you say "note" instead of "take note" or use filler words like "um, note this down" Hands-free operation - Just say "Doubao, take a note, [your content]" and it appears in Obsidian instantly Smart deduplication - Won't create duplicate entries for similar content Cross-platform - Works on Windows, macOS, Linux Technical approach:
Uses Playwright to monitor DOM changes and network requests Regex engine handles speech variations and colloquialisms SQLite for deduplication logic Async I/O for real-time file operations Real use cases I've tested:
Capturing meeting insights while driving between offices Recording workout thoughts during runs Noting recipe improvements while cooking Quick idea capture during walks The tool essentially breaks the walled garden of AI earphones and turns them into a proper knowledge management device. It's like having a voice assistant that actually remembers and organizes everything you tell it.
Demo: Say "Doubao, note this: remember to research async patterns" → Instantly appears in Inbox/Voice Notes/2026-01-21.md
Built with Python + Playwright + SQLite. MIT licensed.
What do you think? Have you faced similar issues with voice assistants that don't persist information? Would love to hear about your workflows for capturing ideas on-the-go!
GitHub: https://github.com/GptsApp/doubao-earphone-to-obsidian
I totally get the concern—continuous recording is definitely a gray area in everyday life. My tool is more focused on active recording (like saying "Doubao, take a note..."), not passive monitoring. The key differences are:
Intent - You clearly know what you're recording. It feels more like "writing something down" than "recording audio"
Visibility - Earbuds are already so commonplace that they don't draw attention like a dedicated voice recorder would
Social boundaries - I typically only record personal thoughts, not conversations with others (which would require clear consent)
The 2 line explanation is sort of vague, but from the code I surmise the Python "app" watches a webpage (configured as https://www.doubao.com/chat/624642496948226) and every time the DOM there is modified, it sees that new prompt, looks for the word "note", and if so, creates an Obsidian note with the transcription of the prompt.
Alexa has "build your own app", this seems less convoluted.Google Gemini also records my prompts (under My Activity), I guess with an always-listening Gemini Assistant and a similar Python script that monitors https://myactivity.google.com/product/gemini (I'm guessing this page needs a hard reload to update), it's possible to build something similar.
I don't have my phone to respond to "Hey Google", but I have an alarm clock that has that (not Gemini, but Google Assistant), and I often tell it to "Remind me about [...] in x hours". I just tested the phrase "Add a note about...", and it added a note in Google Keep. But with an analog Python script one could trigger many more things.