SIP Server currently supports video calls. In order to support recording of video streams, SIP Server must connect both audio and video streams to Media Server. Media Server will bridge both audio and video streams between the participants and replicate both audio and video streams towards the media recorder. This diagram is effectively the same as the basic call recording model.
Since Media Server is modeling call recording internally as a conference, Media Server must treat a video call as a simple video-switching type conference. This means the parties are seeing each other and no video rendering is done by Media Server.
While the basic video recording model is straightforward, there are a few additional considerations and design requirements:
- A client of SIP Server such as the media recorder or agent desktop may make requests for specific media streams to be recorded. The client may choose either audio only or audio+video streams. Media Server acknowledges the request and will only offer the requested media streams to the media recorder. For example, if the client only asks for audio only, then Media Server will only send audio stream in the SDP offer in the recording session to the media recorder.
- The media recorder may have its own local policy to determine which media streams to record if Media Server decides to offer both audio and video in the recording session. The media recorder may only want to store audio streams; to do so, the media recorder sends the SDP answer to accept the audio media stream by setting a non-zero port and reject the video media stream by setting a zero port.
- The client of SIP Server needs to get separate recording indication events for audio recording and video recording.
- For delivering inband indications for video recording, it can either be an explicit announcement before entering the recording, or provide visual indication continuously for the duration of the recording. The mechanism for handling the latter part is undefined for now, but it is expected that Media Server will need to provide rendering of the visual indication for the duration of the recording.
- When delivering the recorded media streams over a single SIP recording session, it will end up having four SDP m= lines in the SDP offer. The mapping of media streams to the devices is explained in the Interfaces section.