clio/clio-android/todo.md

Phone Recorder - Future Ideas / TODO

Questions to Address

Key Rotation

• What happens when server keypair changes?
• Current behavior: Old encrypted files can't be decrypted with new key
• Options:
  1. Keep old private key archived for decrypting old files
  2. Sync all files before rotating keys
  3. Store key version/ID in encrypted file header

Manual Decryption

• Document steps to decrypt files copied via USB from phone
• Need: Python script using server's private key
• Consider: CLI tool in server/ directory

No-Key Behavior

• Current: Deletes recordings if no public key configured (security-first)
• Alternatives to consider:
  1. Store unencrypted with warning (less secure)
  2. Queue recordings until key configured (could fill storage)
  3. Current behavior is probably correct for security model

---

Possible Features

High Priority (if needed)

• QR code for key import (avoid email/typing)
• Server URL configuration in app UI (currently requires code change for non-defaults)
• Manual sync trigger button
• View sync history/status in app
• Display last successful sync time on phone (e.g., "Last sync: 2 hours ago")
• Server status page - HTML dashboard showing:
  • When phone last synced
  • Total recordings count
  • Recent uploads
  • Disk space usage
  • Could be at /status or /dashboard

Medium Priority

• Configurable recording quality (bitrate, sample rate)
• Configurable segment duration (currently 2 hours)
• Delete synced files automatically (with configurable retention)
• Server: Authentication for upload endpoint
• Server: HTTPS with Let's Encrypt

Audio Analysis & Intelligence

• Speech detection bar - Visual indicator showing when speech is detected vs silence
  • Display as additional bar in recording detail view alongside waveform
  • Data comes from batch processing script
• Sleeping detection - Detect periods of sleep/silence for overnight recordings
  • Could help skip to interesting parts
  • Visual indicator in timeline
• Auto voice recognition / Speaker identification
  • Identify who is talking (requires training on voice samples)
  • Label segments by speaker
  • Not text-based (transcripts are unreliable)
• Auto highlight generation
  • Audio-based detection (not text search since transcripts are bad)
  • Example: detect phrase patterns like "Ok, note taken" acoustically
  • Auto-create highlights with tags like "note_taken"
  • Use audio fingerprinting or keyword spotting models

Viewer UX

• Background audio playback - Keep playing current audio while browsing other recordings
  • Mini player that persists across views
  • Allows reviewing recordings list while listening
• Highlight duration - Currently highlights mark only a point; consider adding end time/duration
• Auto-scroll timeline to recordings - If recordings are only at end of day, auto-scroll so user doesn't have to scroll down to see them

Low Priority / Nice to Have

• Multiple server destinations
• Compression before encryption
• Resume interrupted uploads
• Server: Web UI to browse/play recordings
• Server: Transcription integration

Infrastructure

• Systemd service file for server
• Docker container for server
• NixOS module for server
• Backup strategy for server private key

---

Known Limitations

1. No streaming encryption: Entire temp file must complete before encryption
2. No partial sync: If upload fails, whole file retries
3. Single server: Can only sync to one destination at a time
4. No authentication: Anyone who can reach server can upload (relies on encryption for confidentiality)

---

Testing Notes

Unit Tests (server/)

python test_crypto.py

Integration Tests (server/)

# Start server first
python main.py &
python test_integration.py

Manual Testing Checklist

• Record short clip, verify .ogg.enc created
• Verify temp file deleted after encryption
• Check sync triggers after recording stops
• Verify hash verification (corrupt a file, see it retry)
• Test LAN detection with different subnets
• Test sync mode transitions