A browser-based live note taker that turns spoken words into a running transcript and an organized, editable outline, helping meetings, lectures, and brainstorming sessions become useful notes while they are still happening.
Traditional transcription captures words but still leaves someone with a long block of text to organize later. Taking structured notes manually can also pull attention away from the conversation itself.
The approach
AI Live Scribe pairs browser speech recognition with an outline that is updated as the session develops. The transcript preserves what was said while the outline turns the same material into headings and notes that can be edited immediately.
What it does
Live transcription
Capture speech continuously in a supported Chromium browser and see the transcript grow during the session.
A running outline
Periodically reorganize the conversation into a structure that is easier to scan than raw transcript text.
Human editing remains central
The generated outline is editable, so the user can correct emphasis and shape the final notes.
A focused two-pane workspace
Keep the source transcript and organized notes visible together instead of switching between separate tools.
Design decisions
The choices that shaped how the project works.
Use the browser for speech
The Web Speech API provides an immediate microphone experience without installing a desktop application.
Keep the AI credential off the client
Outline requests go through the site's server-side Gemini proxy rather than shipping a key in the browser bundle.
Organize incrementally
The outline develops during the session so the user receives useful structure before the recording is over.
Built with
React · Vite · Web Speech API · Gemini · Django proxy
Availability
Publicly available in Chrome and Edge. Browser speech-recognition support determines whether live transcription is available.
Explore the project
Open the working app, or continue browsing the rest of the portfolio.
