Introduction - Kair Voice

What is Kair Voice?

Kair Voice is a platform for recording, transcribing, and analysing spoken sessions — meetings, workshops, citizen assemblies, or any group conversation where you need a reliable record of what was said and by whom. It combines:

Hardware devices (ESP32-based recorders) or browser-based recording via the web app
Automatic transcription — either via a local Whisper model or a hosted inference cluster (Foresight)
Speaker diarisation — segments labelled by who was speaking
AI-powered analysis — summaries, concern detection, and a knowledge graph (HiRAG) that lets you query across sessions
Multi-workspace support — separate groups or projects can each have their own sessions, members, and settings

Who uses it?

Role	What they do
Moderator / Facilitator	Starts/stops recordings, watches live transcription, manages sessions
Admin	Configures transcription models, manages users, views all sessions across workspaces
Participant	Joins a session as a guest; their speech is captured and attributed
Your colleague	Reads this documentation to understand how the system works

How a session flows

Device / Browser  →  Audio upload (chunks)  →  Transcription worker
                                                      ↓
                                             Summarisation worker
                                                      ↓
                                              HiRAG indexing (knowledge graph)

See Platform Overview for a fuller picture, or jump straight to Recording a Session.