What Happens During Processing

What Happens During Processing

When you upload a recording, MuniMins runs it through several stages before delivering finished minutes.

Transcription. The audio is transcribed using AI speech recognition. Every word spoken in the meeting is converted to text with timestamps.

Speaker diarization. A separate AI process identifies individual speakers in the recording by analyzing voice characteristics. Each speaker is labeled (Speaker A, Speaker B, etc.) and their segments are mapped across the full meeting.

Speaker pre-matching. If you've set up board members and participants with voice profiles, MuniMins compares the detected voices against your stored profiles. Matches above a confidence threshold are pre-filled with names.

Stage 1 — Extraction. The transcript is sent through MuniMins' AI with instructions to extract structured data: attendance, roll call, every motion with mover and seconder, vote counts, and all discussion items in chronological order.

Stage 2 — Writing. The extracted data is sent through a second AI pass that writes formal, third-person minutes at the detail level you selected. This is where the raw data becomes the formatted minutes you see on the view page.

Save. The finished minutes, transcript, speaker mapping, and extraction data are all stored in your organization. You can review, edit, identify speakers, and move the minutes through the approval workflow.

The entire process is automatic. You upload, MuniMins handles the rest.