Conversation

1. ~~input capture
   • ~~text based input
     • ~~shared doc
     • ~~use webhook service to capture data and send to LLM
   • voice based input (speech to text tools)
     • whisper-stt
     • vosk-stt
     • deepspeech-stt (no longer actively supported, but could fine-tune)
   • storing inputs
     • Use a local database (e.g., SQLite or PostgreSQL) to log conversations.
     • Implement a timestamped message queue to track dialogue progression
2. processing
3. user interface

Real time interjections and interactions
voice-activity-detection (VAD)

[speaker-diarisation]