EchoEtcher

An automated system that converts audio notes into formatted Markdown notes with allowed tags using local AI processing.

Project Status ⚠️

Experimental / Work in Progress

This project is a rapid prototype developed through collaborative AI assistance and personal iteration. As such, it comes with some important caveats:

• 🧪 Experimental: The codebase is in active development
• 🛠 Unstable: Expect potential breaking changes
• 🐛 Limited Testing: Minimal comprehensive testing has been performed
• 🚧 Use at Your Own Risk: Not recommended for production environments without significant review and modification

Contributions, feedback, and improvements are welcome! If you encounter issues or have suggestions, please open an issue on the repository.

Disclaimer

By using EchoEtcher, you acknowledge that you understand and accept the risks associated with using an experimental project. You agree to hold harmless the developers and contributors of EchoEtcher for any damages or losses resulting from its use.

Features

• Monitors a selected folder for new audio files
• Transcribes audio locally using Whisper
• Processes transcriptions locally with Ollama
• Auto-generates formatted Markdown notes with allowed tags
• Links original audio files in Obsidian vault
• Completely local processing - no cloud services required

Prerequisites

• Python 3.8+
• Ollama installed locally
• FFmpeg installed for audio processing
• Obsidian vault set up

File Processing and Storage

Notes Folder Structure

The Daily Notes folder is used for processed files with the following conventions:

• A sub-folder audio is created to store processed audio files
• Markdown files are generated for each processed audio file
• A markdown file is saved in the Daily Notes folder for each processed audio file and follows the format: yyyy-MM-dd-Succinct-Generated-File-Name.md

Tag Processing

When processing tags from the transcribed audio, EchoEtcher uses a unique approach to tag management:

• The system allows you to specify a set of allowed_tags
• When writing the final tags to the Markdown file, these tags are prepended with '#echo-etcher/'
• Prepending makes it easier to filter and search for specific tags that were created by EchoEtcher in your notes
• This approach ensures consistency and improves note organization and discoverability

Setup

1. Clone this repository
2. (Optional) Create and activate a virtual environment:

   # Using venv (recommended for most users)
   python -m venv venv
   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`
   
   # Alternative: Use conda or your preferred environment manager

3. Install dependencies:

   pip install -r requirements.txt

4. Copy .env.example to .env and configure your paths
5. With Ollama.ai installed locally, open a new terminal and run ollama pull mistral (or whatever model you want to use)
6. Ensure the Ollama model you want to use is set in the .env file (By default the example .env is set to mistral)
7. Ensure Ollama is running locally by running ollama serve
8. Run the watcher:

   python main.py

Configuration

Edit the .env file to configure:

• Audio input folder path
• Obsidian vault path
• Ollama model name and settings
• Note template settings
• Allowed tags file location

Allowed Tags File

The allowed_tags.md file defines the list of tags that can be used in your notes. This helps maintain consistency and organization in your note-taking system.

Location

By default, the allowed tags file should be placed in your Obsidian vault's root directory. The location is specified by the ALLOWED_TAGS_FILE setting in the .env file.

File Format

Create a Markdown file with a list of allowed tags, one per line. Each tag should:

• Start with a #
• Use lowercase letters
• Use hyphens for multi-word tags
• Optionally use hierarchical tags with /

Example allowed_tags.md:

# Allowed Tags

#meeting
#idea
#project
#journal
#family
#biking
#bills
#books/fiction
#books/non-fiction
#health
#finance
#quote
#career
#gaming
#rant

Purpose

• Restricts tags to a predefined list to prevent hallucinations
• Ensures consistent tag usage across notes
• Helps with organization and searchability
• Can be easily modified as your note-taking needs evolve