Recordings and Media
Understand recordings and media types, and learn how to access audio, video, transcripts, participant events, and meeting metadata.
Recall.ai captures conversation data from sources such as meeting bots and the Desktop SDK. This data is organized into recordings and recording media.
Recordings
A recording is a container for the data produced during a recording session. For example, a meeting recording can contain a video file, an audio file, a transcript, and participant events. Each of these outputs is represented by a separate media object associated with the recording.
Recordings can be created through sources such as Meeting Bots, Desktop SDK, and Meeting Direct Connect. The outputs available for a recording depend on its configuration and the source used to record it.
Recording media must be at least 10 seconds longFor recordings shorter than 10 seconds, Recall appends 10 seconds of blank recording media to the end to ensure the media meets the minimum length.
Recording Media
Recording media is the data produced by a recording. Each media object represents a specific output, such as a transcript or mixed audio file.
Recording media is limited to the data types listed below and it does not include other data (e.g., calendar-related data, logs/telemetry). Availability depends on the recording configuration, recording source, and meeting platform; see the linked guides for support and configuration details.
| Media type | What it contains | Related guides |
|---|---|---|
| Transcript | Conversation text with speaker information and timestamps. | Generating Transcripts |
| Video (Mixed) | Participants’ video feeds combined into one layout for meeting playback. | Real-time · Async |
| Video (Separate) | Separate video files for individual participants and screen shares. These files do not include audio. | Real-time · Async |
| Audio (Mixed) | Participants’ audio combined into a single audio file. | Real-time · Async |
| Audio (Separate) | Separate audio files for individual participants. A participant’s audio can be split into multiple parts, each with its own download URL. | Real-time · Async |
| Participant Events | Structured participant information, timestamped activity such as joins and leaves, chat messages, and a speaker timeline showing who spoke and when. | Meeting Participants & Events |
| Meeting Metadata | Information captured from the meeting platform, such as the meeting title or platform-provided identifiers. Available fields vary by platform. | Meeting Metadata |
Captured meeting metadata is separate from calendar data and custom metadata that your application supplies on API requests.
Access recording media
To access stored outputs:
- Retrieve the Recording to find its associated media objects.
- Check that the media object your application needs has a status of
done, meaning it has finished processing. - Access the output through the media object’s
datafield. Depending on the media type, this includes a download URL.
The recording.done webhook indicates that all media associated with the recording has finished processing and is available. If your application needs a specific output sooner, check that media object’s status and access it once it reaches done.
See Video Playback for displaying retained videos in your application.
Your application can also receive supported data while the recording is in progress through Real-Time Endpoints. Configure these endpoints to receive live events or streams instead of waiting for stored media to finish processing.
Storage and retention
Whether media is stored and how long it remains available depend on the recording’s retention configuration. See Storage and Data Retention for retention, pricing, and deletion. See Zero Data Retention for receiving real-time data without recording-media storage.
Updated 17 days ago
