AI Audio Transcription and Summary

Record voice notes and meetings directly within Obsidian (macOS & iOS), transcribe them using OpenAI Whisper or Google Gemini with automatic size-chunking, and summarize/polish them using your LLM of choice (Anthropic Claude, OpenAI GPT, or Google Gemini).

Support me: https://ko-fi.com/hackerh

🚀 Key Features

• Direct Local Recording: Record high-quality audio directly inside Obsidian. Includes support for selecting specific input devices (microphones) via settings.
• Dual-Provider Transcription:
  • OpenAI Whisper: Automatically chunks larger audio files at 22MB to safely stay under OpenAI's 25MB limit.
  • Google Gemini: Transcribes using the latest stable Gemini API (e.g., gemini-2.5-flash), chunking at 10MB to respect inline payload boundaries.
• Intelligent Post-Processing: Automatically cleans up transcripts using Anthropic (Claude), OpenAI (GPT), or Google (Gemini) based on your custom formatting instructions.
• Flexible Note Output:
  • Create a new Markdown file based on custom date/time templates and title generation.
  • Or paste the transcription/summary directly at your current cursor position.
• Fail-Safe Resilience: If any network error, invalid key, or deprecated model causes transcription or post-processing to fail, the plugin will still save the recorded audio file to your vault and create the note with the media player link so you never lose a recording.
• Premium Visuals: Features dynamic visual wave bars, recording status blinking alerts, and time counters designed to integrate seamlessly with modern Obsidian themes.

⚙️ Configuration & Setup

1. Enable the plugin under Settings > Community Plugins.
2. Go to the AI Audio Transcription & Summary settings tab.
3. Fill in your API Keys:
   • Gemini API Key: Required if you use Google Gemini for transcription or post-processing.
   • Whisper API Key: Required for OpenAI Whisper transcription.
   • OpenAI / Anthropic API Key: Required if you use GPT or Claude for post-processing.
4. Configure your Transcription Provider (OpenAI Whisper or Google Gemini) and select a recommended model from the dropdown.
5. Set your output paths, file templates, and optional post-processing templates.

🛠️ Developer Commands

If you want to modify or compile the plugin yourself:

1. Install dependencies:

   npm install
   

2. Compile the production bundle:

   npm run build
   

   This compiles the TS source files into a single bundle file main.js.

☕ Support

If you find this plugin useful and want to support my work, you can buy me a coffee!

Your support helps me maintain the plugin and keep adding new features. Thank you!

📄 License

This project is licensed under the MIT License.