Email Audio Analyzer

Check email accounts for audio attachments, transcribe them (Lithuanian support), and extract event details (type, date, time, people count, contact info). Outputs structured summaries via Telegram or to file.

Email Audio Analyzer Skill ⚡

Automatically checks email accounts for audio attachments, transcribes them using Whisper, and extracts structured event information using AI analysis.

Quick Start

Manual trigger:

Check work email for audio messages
Analyze audio emails from RUBIX

On-demand with options:

Check email audio --account work --output telegram
Check email audio --account gmail --since 24h

Features

• ✅ Check Gmail (personal) or RUBIX work email
• ✅ Download audio attachments automatically
• ✅ Transcribe with Whisper (Lithuanian language support)
• ✅ Extract event details:
  • Event type (concert, party, private, normal day)
  • Date and time
  • Number of people/guests
  • Contact information (name, phone, email)
  • Special requirements/notes
• ✅ Output to Telegram, file, or both
• ✅ Supports multiple audio formats (mp3, wav, m4a, ogg, etc.)

Configuration

Edit config.json to customize:

{
  "defaultAccount": "work",
  "checkInterval": "30m",
  "output": {
    "telegram": true,
    "file": true,
    "memory": true
  },
  "languages": ["lt", "en"],
  "whisper": {
    "model": "base",
    "timeout": 120
  }
}

Commands

Check Email Now

cd ~/.openclaw/workspace/skills/email-audio-analyzer
node scripts/check-email.js --account work

Transcribe Audio

node scripts/transcribe.js --file /path/to/audio.mp3 --language lt

Analyze Transcription

node scripts/analyze.js --text "transcription text here"

Automation

Heartbeat (periodic check): Add to HEARTBEAT.md:

- Check RUBIX email for new audio messages

Cron job:

# Check every 30 minutes
*/30 * * * * cd ~/.openclaw/workspace/skills/email-audio-analyzer && node scripts/check-email.js --account work

Output Format

Summaries are saved to output/YYYY-MM-DD-HH-mm-summary.md:

## 📋 Event Inquiry

**Contact:**
- Name: Jonas Petraitis
- Phone: [phone]
- Email: [e-mail]

**Event Details:**
- Type: Birthday party (private)
- Date: May 15, 2026
- Time: 20:00
- Guests: ~30 people
- Notes: Needs sound system, private room

**Source:** Email from [e-mail], received 2026-03-05 11:30
**Audio:** inquiry_20260305_113045.mp3
**Transcription:** [Full Lithuanian text...]

Dependencies

• himalaya CLI (work email)
• gog CLI (Gmail)
• whisper CLI (transcription)
• Node.js 18+

Install dependencies:

cd ~/.openclaw/workspace/skills/email-audio-analyzer/scripts
npm install

Troubleshooting

Audio not downloading:

• Check email CLI is configured (himalaya configure or gog auth)
• Verify attachment permissions

Transcription failing:

• Ensure Whisper is installed: which whisper
• Try smaller model: --model tiny
• Check audio format is supported

Lithuanian accuracy poor:

• Use --model medium or --model large for better accuracy
• Ensure --language lt is specified