Vocalog

English | 简体中文

Obsidian plugin for automatic audio transcription and AI-powered note generation.

Features

• 🎤 Automatic transcription using STT API
• 🤖 AI-powered summaries using LLM (OpenAI-compatible)
• 📝 Automatic writing to daily notes
• 📅 Batch processing by date range
• 🗓️ Calendar interface for date selection

Installation

From Obsidian Community Plugins (Recommended)

[To be available after community review]

Manual Installation

1. Download the latest release from GitHub
2. Extract to <vault>/.obsidian/plugins/vocalog/
3. Reload Obsidian and enable the plugin

Configuration

Open Settings → Vocalog to configure:

Required Settings

1. Audio Folder: Where your audio recordings are stored (e.g., Inbox/Voice)
2. STT API Configuration:
   • API URL (Whisper-compatible endpoint)
   • API Key
   • Model name (default: whisper-1)
3. LLM API Configuration:
   • API URL (OpenAI Chat-compatible endpoint)
   • API Key
   • Model name (default: deepseek-chat)
   • System Prompt (customize how notes are structured)

Optional Settings

• Output Folder: Where to save daily notes (defaults to vault root)
• Daily Note Format: Date format pattern (default: YYYY-MM-DD)

Usage

Generate Notes from Today's Audio

Click the microphone ribbon icon or use command: Vocalog: Generate Audio Notes

Process Specific Dates

Use command: Vocalog: Generate Audio Notes by Date Range

Process Selected Files

Select audio files in file explorer, then use command: Vocalog: Generate from Selected Audio Files

Context Menu

Right-click any audio file → 🎤 Generate Audio Note

API Compatibility

This plugin works with OpenAI-compatible APIs:

STT (Speech-to-Text):

• Compatible with OpenAI Whisper API format
• Tested with: OpenAI Whisper, Groq Whisper

LLM (Summarization):

• Compatible with OpenAI Chat Completion API format
• Tested with: OpenAI GPT models, DeepSeek, other OpenAI-compatible services

Network and Data Use

Vocalog only sends data when you run a transcription or summarization command.

• Audio files you choose to process are sent to the configured speech-to-text API endpoint.
• Generated transcripts are sent to the configured AI summarization API endpoint.
• API keys are stored locally in this plugin's Obsidian settings data.
• Vocalog does not collect analytics, telemetry, or tracking data.

License

MIT

Contributing

Contributions welcome! This is a "vibe coding" project developed with Claude AI assistance.