Per-Participant Audio Streams: Separate Audio for Every Speaker

Get one audio file per speaker, aligned to the meeting timeline, alongside the usual mixed recording.

Supported platforms: Zoom, Google Meet and Microsoft Teams.

Platform support

See the capability matrix for the rest of what each platform supports.

1) What you need

1. A MeetStream API key (see Dashboard Setup).
2. A meeting link (Zoom, Google Meet or Microsoft Teams).
3. audio_separate_streams: true in your Create Bot request.

2) Create a bot with per-participant audio

Use the Create Bot endpoint.

Minimal example

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

To try it without code, open API Playground in the dashboard and go to the Streams step:

Full example with transcription and MP3 copies

{
  "meeting_link": "https://meet.google.com/abc-defg-hij",
  "bot_name": "RecorderBot",
  "video_required": false,
  "audio_separate_streams": true,
  "audio_separate_mp3": true,
  "recording_config": {
    "transcript": {
      "provider": {
        "deepgram": {
          "model": "nova-3",
          "language": "en"
        }
      }
    }
  },
  "automatic_leave": {
    "everyone_left_timeout": 60
  }
}

Parameter reference

Tip: You can also pass audio_separate_streams inside recording_config:

"recording_config": {
  "audio_separate_streams": true
}

The top-level parameter takes precedence if both are set. Set audio_separate_streams: true explicitly when you need per-participant files.

3) What happens during the meeting

Once the bot is admitted, it automatically:

• Captures each speaker’s audio as a separate file.
• Records up to 16 speakers. A 17th distinct speaker is not captured.
• Keeps the standard mixed audio file (audio.wav) as well.

Files are on the meeting timeline: gaps between a person’s turns are filled with silence, and each file’s start_timestamp tells you where it begins, so files can be laid side by side. If you pause recording, every file is silent for the paused stretch.

4) Retrieve per-participant audio streams

After the bot leaves and audio processing finishes (the audio.processed webhook), call:

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

Response when processing is complete (200)

The response is an array with one item per speaker file:

[
  {
    "id": "f923cd21-da86-4d77-b37b-0707d37751c8-spaces_AbCdEf_devices_185-0",
    "created_at": "2026-09-20T17:01:58.913Z",
    "is_bot": false,
    "participant": {
      "id": 48213377,
      "name": "Jane Smith",
      "is_host": true,
      "is_bot": false,
      "platform": "desktop",
      "extra_data": {
        "google_meet": { "static_participant_id": "spaces/AbCdEf/devices/185" }
      }
    },
    "start_timestamp": { "absolute": "2026-09-20T17:01:58.913Z", "relative": 0.0 },
    "duration": 1843.2,
    "download_url": "https://...",
    "download_url_mp3": "https://..."
  }
]

Other responses

5) Download the audio files

curl -o "jane.webm" "<download_url from the response>"

Links expire after 1 hour. Call get_audio_streams again for fresh links. Check the file extension in the link: Google Meet files are .wav, Zoom and Teams files are .webm.

6) Per-participant audio by platform

If exact isolation during crosstalk matters on Teams, rely on the post-call transcript’s speaker labels for overlapping segments.

7) Using both mixed and per-participant audio

Per-participant audio and the mixed audio are always captured together:

• Mixed audio WAV (everyone combined): Get Bot Audio (/api/v1/bots/<BOT_ID>/get_audio). See retrieve recordings.
• Per-speaker files: Get Audio Streams (/api/v1/bots/<BOT_ID>/get_audio_streams).

8) Using per-participant audio with per-participant video

{
  "meeting_link": "<MEETING_LINK>",
  "video_required": true,
  "audio_separate_streams": true,
  "video_separate_streams": true
}

Audio and video come as separate files; they are not muxed together. Match them with participant.extra_data (static_participant_id) across the two responses, and line them up with start_timestamp. On Zoom, per-participant video also needs video_required: true.

9) Webhook notifications

Subscribe to audio.processed in webhooks and events, then call get_audio_streams.

10) Troubleshooting

• Empty array [] after the meeting?

  • Confirm audio_separate_streams: true was set when the bot was created.
  • Wait for audio.processed; files are uploaded after the bot leaves.
  • Participants who never spoke do not get a file.

• A speaker is missing?

  • The bot captures up to 16 speakers.
  • The bot only records after it is admitted. If it never got in, see bot status and error codes.

• Getting in_progress?

  • The bot is still in the meeting. Files are produced after it leaves.

• The download link returns 403 Forbidden?

  • The link has expired (1-hour lifetime). Call get_audio_streams again.

• On Teams, one file contains two voices?

  • Expected during crosstalk. See section 6.