Voice Interface for My Second Brain
A voice-driven interface that turns natural speech into structured tasks, projects, events, and notes inside a customized Obsidian knowledge system.
The Problem
My Obsidian system already organized projects, tasks, events, notes, and daily information—but adding something still required opening the application, navigating to the right place, and entering the correct metadata.
I wanted the system to understand intent without requiring the interface.
The Solution
I built a voice layer that accepts natural speech, transcribes it, identifies the user's intent, and converts it into structured Markdown.
The system determines whether the input is a task, event, project, note, or other supported object and applies the appropriate metadata, template, folder, and date information.
The resulting item appears inside the existing Obsidian system without requiring manual organization.
How It Works
Example: Spoken input — "Add Toyota interview Tuesday at 9 AM." The system identifies the event, extracts the date and time, generates the structured entry, and places it into the appropriate workflow.
Highlights
- Speech-to-text input
- Local LLM intent parsing
- Structured Markdown generation
- Automatic metadata
- Template and folder routing
- Tasks, events, projects, and notes
- Integration with existing Obsidian workflows