Product updates

Developer Changelog

Keep track of changes and upgrades to the Zoom Developer platform. You can also subscribe to our Changelog RSS feed below.


RTMS

RTMS November release

New featureAPI release

Summary:Adds V2 protocol features, transcript rewind, Zoom Phone support, new participant and audio data, and a TLS 1.2 or TLS 1.3 requirement.

The RTMS November release adds V2 protocol features, transcript rewind, Zoom Phone support, new participant and audio data, and a TLS 1.2 or TLS 1.3 requirement.

RTMS V2 protocol

The V2 protocol adds new features and some restrictions compared to V1. For details, see Protocol versions.

  • Protobuf wire format: A binary alternative to JSON with smaller payloads, faster serialization, and native binary data handling without Base64 encoding. JSON messages are unchanged and work with both V1 and V2. For details, see Protobuf definitions.
  • Application-layer media payload encryption: V2 supports AEAD encryption of outbound media and text content. It adds a layer of security on top of the existing TLS and WebSocket transport encryption. The Zoom team must approve and enable it for each app. This feature is available by request only. For details, see Signaling handshake request.
  • SRT transport for data connections: A low-latency transport with selective packet loss recovery that performs better on unstable networks. Existing WSS data connections remain supported. The signaling handshake response includes srt_passphrase and srt_pbkey_length. For details, see Connecting over SRT.
  • Event codes: V2 uses event_code, a globally unique identifier, in place of the legacy service-specific event_type. Zoom Meetings and Zoom Contact Center voice keep V1 compatibility. Zoom Phone uses event_code only. For details, see RTMS_EVENT_CODE.
  • Multiple individual video streams: Lets you open a separate video data connection for each participant you want to receive video from. You subscribe each connection to one participant over the signal connection, and each connection receives only that participant's video. Capacity depends on the requested resolution. By default, the service supports up to four concurrent 720p subscriptions per stream. The server-side multiple-video-subscription feature must be enabled and is available by request only. For details, see Multiple individual video streams.
  • Video format: Standard definition video is 640x360, and PNG is no longer available for video or screen share. For details, see Video.

Error responses

RTMS adds a REQUEST_ERROR_RESP message type. It returns request_msg_type, status_code, reason, and timestamp for failed requests, and it adds more validation status codes. For details, see Request error response.

Zoom Phone

Zoom Phone developers can use RTMS to retrieve audio or transcript data, consistent with the existing Zoom Meetings and Zoom Contact Center voice capabilities. The initial Zoom Phone events use event codes 3000 to 3003 and cover call information, hold, resume, and transfer completion. For details, see Zoom Phone events.

New Zoom Phone webhook events:

  • phone.call_rtms_started
  • phone.call_rtms_stopped
  • phone.call_rtms_interrupted

Transcript rewind

Transcript rewind lets RTMS clients receive corrections to transcript sentences that were already delivered, as the ASR refines its results. Rewind is on by default when enable_lid is enabled, and it's off when enable_lid is set to false.

Initial sentences arrive as NEW events (attribute: 1). Revisions arrive as UPDATE events (attribute: 2) with the same sentence_id and the complete replacement text. On an UPDATE, replace the existing sentence instead of adding a new one. With speaker diarization enabled, speaker-specific segments keep the parent sentence_id and update attribute. For details, see Handling transcript updates.

Participant and audio data

  • MEDIA_DATA_USER_INFO event: The MEDIA_DATA_USER_INFO event identifies participants whose audio, video, screen share, transcript, or chat data the service has observed. It includes user_id, user_name, and participant_uuid. The service sends it automatically in the EVENT_UPDATE message, so no subscription is needed. Zoom Meetings only. For details, see Participant user info.
  • Audio source: MEDIA_DATA_AUDIO includes a content.source field in JSON and protobuf. A value of 1 means microphone audio, and 2 means shared-content audio. Use it to tell screen-share audio apart from microphone input. For details, see AUDIO_DATA_SOURCE.
  • Original audio parameters: Meeting audio keeps its 48 kHz sample rate and 16 kHz frequency range by default. Apps that need 48 or 32 kHz audio keep the same frequency range. Apps that need 16 kHz audio see no change. Telephony and special device audio keeps its native sample rate and frequency range. For details, see Handling media data.
  • Audio message boundaries: For RTP and raw Opus, G711, or G722 content, each MEDIA_DATA_AUDIO message contains one audio frame. For raw L16 content, messages contain variable-length frames up to 128 KiB. For details, see Message boundaries.

TLS version requirement

RTMS requires TLS 1.2 or TLS 1.3 on all WebSocket connections. Connections that use an older TLS version are rejected during the TLS handshake. SRT media streams aren't affected. Make sure your client connection requests use one of the supported versions. For the supported cipher suites, see TLS requirements.