AI Notes - Obsidian Plugin

An Obsidian plugin to record audio, transcribe it, and enrich your written notes with AI-powered insights.

Features

• 🎙️ Audio recording — Record directly from your microphone while editing a note
• ✍️ Transcription — Convert recordings to text using whisper.cpp or any OpenAI-compatible endpoint
• 🤖 AI enrichment — Combine your notes and transcriptions, then summarize and structure them with an LLM
• 📂 Organized storage — Recordings are saved in per-note folders inside a configurable vault directory
• 🔌 Flexible backends — Works with local models (Ollama, whisper.cpp, llama.cpp) or cloud APIs (OpenAI, etc.)

Commands

Command	Description
Start/Stop recording	Toggle microphone recording — audio is embedded in the active note
Transcribe recordings	Transcribe all embedded audio files in the current note
Enrich note	Generate an AI-powered summary combining your notes and transcriptions

All commands are available via the Obsidian command palette (Ctrl/Cmd + P).

Installation

1. Open Settings → Community Plugins → Browse
2. Search for AI Notes
3. Click Install, then Enable

Manual installation

1. Download the latest release from GitHub Releases
2. Extract main.js, manifest.json, and styles.css into your vault at .obsidian/plugins/ai-notes/
3. Reload Obsidian and enable the plugin in Settings → Community Plugins

Configuration

Open Settings → AI Notes to configure the plugin.

Recording

Setting	Description	Default
Recordings folder	Vault folder where audio files are saved	recordings

Transcription (Whisper)

Setting	Description	Default
Whisper endpoint URL	whisper.cpp server (http://host:port) or OpenAI-compatible (http://host:port/v1)	http://localhost:8080
Whisper model	Model name (only for OpenAI-compatible endpoints)	whisper-1
Whisper API key	Optional API key (only for OpenAI-compatible endpoints)	—

Enrichment (LLM)

Setting	Description	Default
LLM endpoint URL	OpenAI-compatible API base URL	http://localhost:11434/v1
LLM API key	Optional API key for the endpoint	—
LLM model	Model name to use for enrichment	llama3

Usage

1. Record

Open a note and run Start/Stop recording from the command palette. The status bar shows 🔴 while recording. Run the command again to stop — the audio file is automatically embedded in your note.

2. Transcribe

Run Transcribe recordings to convert all embedded audio files in the current note to text. Transcriptions appear in collapsible blocks below each recording.

3. Enrich

Run Enrich note to send your written notes and transcriptions to an LLM. The AI-generated summary is appended to your note.

Note structure

The plugin organizes your note like this:

Your written notes here...

## 🔴 REC
![[recording-file.webm]]
<details>
<summary>Transcription</summary>
Transcribed text...
</details>

## 🤖 AI
AI-generated enrichment...

Privacy First

This plugin does not connect to any external service out of the box. All endpoints are user-configured — nothing is sent anywhere until you explicitly set up and point the plugin to your own services.

Works great with local AI! Use llama.cpp, LM Studio, Ollama, whisper.cpp — or any OpenAI-compatible endpoint — to keep your data entirely on your machine.

• Transcription: Audio recordings are sent to the Whisper-compatible endpoint you configure.
• Enrichment: Note content and transcriptions are sent to the LLM endpoint you configure.

You are in full control of where your data goes.

Credits

AI Notes is made with ♥ by derogab and it's released under the GPL-3.0 license.

Contributors

Tip

If you like this project or directly benefit from it, please consider buying me a coffee:
🔗 bc1qd0qatgz8h62uvnr74utwncc6j5ckfz2v2g4lef
⚡️ derogab@sats.mobi
💶 Sponsor on GitHub

Stargazers over time