User built an automated pipeline to transcribe Plaud recordings and create structured Obsidian notes, stating that getting recordings into a usable, searchable format was tedious even with Plaud's built-in transcription.
Hey everyone, I've been using Plaud for a while and kept running into the same problem — getting recordings into a usable, searchable format was tedious even with the built-in transcription. I wanted something that would automatically land a clean, structured note in Obsidian the moment a recording synced to iCloud. So I built it. **What it does:** The pipeline runs entirely on your Mac in the background. When a new recording syncs from the Plaud app to iCloud, it gets automatically transcribed using OpenAI's Whisper model (running locally in Docker), then Claude structures it into a formatted Obsidian note using whatever template you define. The note appears in your vault within a few minutes with no manual steps. **The full flow:** Plaud → iCloud → Mac watcher script → Docker/Whisper transcription → Claude API → Obsidian note → iCloud sync to mobile **Tech stack:** * Whisper (local, runs in Docker) for transcription * Claude API for summarization and note structuring * Obsidian Local REST API plugin for writing notes directly into the vault * Python for orchestration and file watching * Docker for containerization **Cost:** About $0.01-0.02 per recording in Anthropic API costs. For 50 recordings a month that's roughly $0.50-$1.00. **What you need:** * Mac with Docker Desktop * Obsidian with the Local REST API plugin * Anthropic API key (separate from Claude Pro subscription) * Node.js 18+ The repo is fully documented with step by step setup instructions, a startup script for after reboots, and example config files with no API keys committed. GitHub repo: [https://github.com/holzerchristopher-tech/Plaud-Claude-Obsidian](https://github.com/holzerchristopher-tech/Plaud-Claude-Obsidian) Would love feedback from anyone who tries it — particularly if you're using a different vault structure.