Skip to main content

Overview

In-Person Recording lets you capture meetings that happen in the room — no video call needed. Your device’s microphone records the conversation, Deepgram transcribes it with speaker diarization, you label the speakers, and then the full AI analysis pipeline runs. In-person recordings get the same intelligence as video call recordings: executive summary, key decisions, action items, risks, open questions, speaker statistics, and a momentum score.

How to Record

1

Open the recording page

Click the green Record In-Person button from the Meetings page header or the Dashboard. This opens a dedicated recording view at /meetings/record.
2

Enter a title

Give your meeting a descriptive name (e.g., “Weekly Team Standup”, “Client Discovery Call”). This is optional but makes meetings easier to find later.
3

Start recording

Click the red Start Recording button. Your browser will ask for microphone permission the first time.During recording you’ll see:
  • A live audio waveform visualization
  • A running timer (minutes:seconds or hours:minutes:seconds)
  • Pause and Stop buttons
4

Pause and resume (optional)

Click Pause to temporarily halt recording (e.g., during a break). Click Resume to continue. The timer picks up where it left off.
5

Stop recording

Click Stop when the meeting ends. You’ll see:
  • An audio player to preview the recording
  • Upload & Transcribe button to proceed
  • Re-record button to discard and start over
6

Upload and transcribe

Click Upload & Transcribe. The audio is uploaded to secure storage and sent to Deepgram for transcription with speaker diarization. You’re automatically redirected to the meeting detail page.

Speaker Labeling

After transcription completes, the transcript shows generic speaker labels: Speaker 1, Speaker 2, etc. You need to assign real names before analysis can run.
1

Open the meeting detail page

You’re redirected here automatically after upload. If you navigate away, find the meeting in the Recordings tab — it has a green In-Person badge.
2

Rename speakers

An amber Label Speakers panel appears above the transcript. Each generic speaker label has an input field next to it. Type the person’s real name for each speaker.
3

Save and analyze

You have two options:
  • Save Names Only — Updates the transcript with real names but doesn’t run analysis yet
  • Save & Analyze — Updates names AND triggers the full AI analysis immediately
Click Save & Analyze when you’ve labeled all speakers.
The speaker labeling panel only appears when analysis_status is awaiting_speaker_labels. Once you click Save & Analyze, the panel disappears and the analysis runs in the background. The page auto-refreshes when analysis completes.

Audio Quality Tips

For best transcription results:
  • Device placement — Place your laptop or phone near the center of the table, not next to one person
  • Quiet environment — Minimize background noise (close doors, turn off fans if possible)
  • Speak clearly — Encourage participants to speak one at a time for better diarization
  • Microphone check — Test your microphone in browser settings before an important recording
  • Battery — Ensure your device is charged or plugged in for longer meetings

Technical Details

SettingValue
Audio formatWebM with Opus codec
Sample rate48 kHz
ChannelsMono
Noise suppressionEnabled (browser-level)
Echo cancellationEnabled (browser-level)
Transcription modelDeepgram Nova-2
DiarizationEnabled (automatic speaker detection)

Where In-Person Meetings Appear

  • Recordings tab — With a green In-Person badge
  • Meeting detail page — Full transcript with speaker labels, analysis, action items
  • Brain search — Searchable via meeting_intel tool like any other meeting
  • Dashboard — In the “Recent Recordings” widget