> ## Documentation Index
> Fetch the complete documentation index at: https://koreai.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Call recording API overview

> Use the call recording API to fetch a session's call recordings with media URLs and to stream recording media for compliance archiving, QA, and analytics.

Use the call recording API to fetch the server-side call recordings that a voice session produced, with a media URL for each recorded file. The API is read-only. It supports unattended jobs such as compliance archiving, QA sampling, and analytics pipelines that need call audio without a person's credentials. It does not return transcripts or other conversation text.

## Prerequisites

* A project-bound platform API key (`abl_…`) with the `recordings.read` scope. Send it in the `x-api-key` header or as `Authorization: Bearer abl_…`. The API refuses Studio user JWTs, JWT App assertions, and public `pk_` keys with `403 PLATFORM_API_KEY_REQUIRED`.
* Call recording enabled for the tenant.
* The project ID and the session ID.

## Workflow and use cases

1. Get all recordings for the session.
2. Check `data.status`. If `data.retryable` is `true`, retry with backoff. Stop retrying when it is `false`.
3. For each file that has a URL, check `urlKind`. For `signed` and `external`, send a `GET` request to the URL. For `proxied`, prefix the relative URL with the API host and send a `GET` request.
4. Stream the media. A `proxied` URL calls the stream endpoint, which supports byte ranges.

Send no API key or `Authorization` header when you request a media URL.

## List of endpoints

| Use case | Help article | Purpose | Credential |
| :- | :- | :- | :- |
| Fetch | [Get all recordings for a session](https://koreai.mintlify.site/agent-platform/api-reference/recording-fetch-calls-per-session) | List a session's recordings and their files, each with a media URL or the reason none is available. | `recordings.read` scope on the key |
| Stream | [Stream a recording](https://koreai.mintlify.site/agent-platform/api-reference/recording-stream) | Stream a recording file from a `proxied` URL, with HTTP range support. | Media token in the URL |

## Behavior to know

* A session can produce several recordings, such as after a transfer. Each recording can have several files, such as separate caller and agent channels.
* An HTTP `200` response does not mean the media is available. The `data.status` field is `success`, `partial`, `call_in_progress`, `media_in_progress`, or `media_unavailable`.
* Media URLs are short-lived. The default lifetime is 900 seconds and the maximum is 3600 seconds. The `expiresIn` field is `null` for external URLs. Treat every media URL as a bearer secret.
* Use the `disposition` parameter to choose `attachment` (default) to download the file or `inline` to play it in a browser.
* A session allows up to 64 recordings and, by default, up to 32 stored files across them. Exceeding a limit returns `413`.
* When the API issues at least one media URL, it records an audit event attributed to the calling key.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.