MeetStream Guide: Per-Participant Video Streams

This guide explains how to capture individual video recordings per participant in a meeting. Instead of a single composite screen recording, you receive a separate WebM video file for each participant’s webcam and any screen shares.

Supported platforms: Google Meet, Zoom, Microsoft Teams.

1) What you need

1. A MeetStream API key (see the Quickstart Guide).
2. A meeting link (Google Meet, Zoom, or Microsoft Teams).
3. Set video_separate_streams: true in your Create Bot request.

2) Create a bot with per-participant video

Use the Create Agent endpoint: https://docs.meetstream.ai/api-reference/api-reference/bot-endpoints/create-agent

Minimal example

$
curl -X POST "https://api.meetstream.ai/api/v1/bots/create_agent" \
>
  -H "Authorization: Token <YOUR_API_KEY>" \
>
  -H "Content-Type: application/json" \
>
  -d '{
>
    "meeting_link": "<YOUR_MEETING_LINK>",
>
    "bot_name": "RecorderBot",
>
    "video_separate_streams": true
>
  }'

Full example with recording config

1
{
2
  "meeting_link": "https://meet.google.com/abc-defg-hij",
3
  "bot_name": "RecorderBot",
4
  "video_required": true,
5
  "video_separate_streams": true,
6
  "recording_config": {
7
    "transcript": {
8
      "provider": {
9
        "deepgram": {
10
          "model": "nova-3",
11
          "language": "en"
12
        }
13
      }
14
    }
15
  },
16
  "automatic_leave": {
17
    "everyone_left_timeout": 5
18
  }
19
}

Parameter reference

Tip: You can also pass video_separate_streams nested inside recording_config:

1
"recording_config": {
2
  "video_separate_streams": true
3
}

The top-level parameter takes precedence if both are set.

3) What happens during the meeting

Once the bot joins, it automatically:

• Captures each participant’s webcam feed as a separate video stream.
• Captures screen shares as separate streams.
• Records up to 6 concurrent webcam streams. If more than 6 participants have cameras on, additional cameras are queued and start recording when a slot frees up.
• Screen shares are always captured (1 concurrent on Teams/Zoom, multiple possible on Google Meet).

No additional configuration is needed per participant — the bot handles everything.

4) Retrieve per-participant video streams

After the bot leaves and video processing completes, call:

$
curl -X GET "https://api.meetstream.ai/api/v1/bots/<BOT_ID>/get_recording_streams" \
>
  -H "Authorization: Token <YOUR_API_KEY>"

Response when processing is complete

1
{
2
  "bot_id": "f923cd21-da86-4d77-b37b-0707d37751c8",
3
  "video_status": "Success",
4
  "recording_streams_available": true,
5
  "participants": [
6
    {
7
      "participant_name": "John Doe",
8
      "streams": [
9
        {
10
          "stream_id": "user_456",
11
          "segments": [
12
            {
13
              "segment_index": 0,
14
              "url": "https://s3.amazonaws.com/...?X-Amz-Signature=...",
15
              "filename": "John_Doe_456_0.webm",
16
              "duration_seconds": 125.5,
17
              "width": 640,
18
              "height": 360,
19
              "codec": "vp8"
20
            }
21
          ]
22
        }
23
      ]
24
    },
25
    {
26
      "participant_name": "Jane Smith",
27
      "streams": [
28
        {
29
          "stream_id": "user_789",
30
          "segments": [
31
            {
32
              "segment_index": 0,
33
              "url": "https://s3.amazonaws.com/...?X-Amz-Signature=...",
34
              "filename": "Jane_Smith_789_0.webm",
35
              "duration_seconds": 130.2,
36
              "width": 854,
37
              "height": 480,
38
              "codec": "vp8"
39
            }
40
          ]
41
        }
42
      ]
43
    }
44
  ],
45
  "screenshares": [
46
    {
47
      "participant_name": "John Doe",
48
      "streams": [
49
        {
50
          "stream_id": "source_101",
51
          "segments": [
52
            {
53
              "segment_index": 0,
54
              "url": "https://s3.amazonaws.com/...?X-Amz-Signature=...",
55
              "filename": "John_Doe_101_0.webm",
56
              "duration_seconds": 60.0,
57
              "width": 1280,
58
              "height": 720,
59
              "codec": "vp8"
60
            }
61
          ]
62
        }
63
      ]
64
    }
65
  ],
66
  "summary": {
67
    "total_participants": 2,
68
    "total_screenshares": 1,
69
    "total_segments": 3
70
  }
71
}

Response when processing is still in progress

1
{
2
  "video_status": "processing",
3
  "stage": "Core Processing",
4
  "message": "Video is being processed. Try again later."
5
}

HTTP status 202 is returned when processing is not yet complete. The stage field indicates the current processing step. Poll again after 10-30 seconds.

5) Download the video files

Each segment in the response contains a url field — a presigned S3 URL that allows direct download without additional authentication.

$
# Download a participant's video
>
curl -o "John_Doe.webm" "<PRESIGNED_URL_FROM_RESPONSE>"

Important: Presigned URLs expire after 10 minutes. If a URL has expired, call get_recording_streams again to get fresh URLs.

File format

Audio is available separately via the standard Get Bot Audio endpoint.

6) Understanding segments

A participant may have multiple segments. A new segment is created when:

• A screen share resolution changes (e.g., switching from fullscreen 1080p to a smaller window) on Teams or Zoom.
• A participant shares their screen multiple times (each share session is a separate segment).
• A webcam video dimension change occurs on Teams.

Camera off/on does not create a new segment — the bot maintains a continuous recording file per participant across camera toggles.

Segments are ordered by segment_index and are chronologically sequential. Use the duration_seconds field to determine segment timing.

7) Video resolution by platform

Webcam and screen share resolutions differ by platform:

8) Using both composite and per-participant video

You can enable both video_required and video_separate_streams simultaneously:

1
{
2
  "meeting_link": "<MEETING_LINK>",
3
  "video_required": true,
4
  "video_separate_streams": true
5
}

This produces:

• A composite MP4 (single screen recording of the meeting view) — retrieve via the Get Bot Video endpoint.
• Per-participant WebM files — retrieve via the Get Recording Streams endpoint.

9) Webhook notifications

If you have webhooks configured, you will receive notifications when video processing starts and completes. Use the video_status field in webhook payloads or poll get_recording_streams until video_status is "Success".

10) Troubleshooting

• recording_streams_available: false with video_status: "Success"?

  • The bot may have been in the meeting but no participant had their camera on. Per-participant streams are only produced for participants whose cameras were active.

• Missing a participant’s video?

  • The bot captures up to 6 concurrent webcam streams. If more than 6 cameras were active, some participants may not have been captured.
  • Participants who never turned on their camera will not appear in the response.

• Getting video_status: "processing"?

  • Video processing takes a short time after the bot exits. Poll again after 10-30 seconds.

• Presigned URL returns 403 Forbidden?

  • The URL has expired (10-minute lifetime). Call get_recording_streams again for fresh URLs.

• Video file has no audio?

  • Per-participant video files are video-only. Use the Get Bot Audio endpoint to retrieve the meeting audio separately.