Live Transcription

Live transcription uses automatic transcription to convert speech of all active talkers in a video session into text in real-time. It is helpful when participants have limited language proficiency or difficulty in hearing.

Live transcription is enabled by the host, like recording or screen sharing. Participants see the live transcript as subtitles under or to the side of the video. Full transcription of a session can be saved when the session is complete. The transcription process starts when the first participant requests it and stops when the last participant opts out of it. The live transcript is sent to the endpoints that have subscribed for it through notifications.

Note: Live transcription is a subscription-based service. Connect to Sales or Account Managers for subscription details.

Configuring Live Transcription Settings in a Room

Live transcription is automatically enabled in all video rooms if the subscription of this service is enabled. However, you can disable it for a particular room. You can also improve the transcription quality of the service by using the following room level settings:

• live.transcription.language: String. Required. Language Code. The primary language of communication. For more information, see Language Codes [#language-codes).

• live_transcription.auto_transcribe: Boolean, true or false. Default is false. Change this setting to true if you want the transcription process to start automatically when the video session starts. The live transcript is saved in a file, which is made available through the post-session API.

• live_transcription.enable: Boolean, true or false. Default is true. If the live transcription service is subscribed, it is enabled by default. Change this setting to false if the transcription is not needed in a room.

In addition, you must be familiar with the process of creating a video room and obtaining the transcription file after the video session. For more information, see

• Creating a Video Room

• Obtaining the Post Session Transcription File

Below is an example JSON payload for room definition with live transcription.

Example:

{
  "name": "HLS Trial",
  "owner_ref": "XOXO",
  "settings": {
    "description": "HLS Trial",
    "mode": "group",
    "scheduled": false,
    "adhoc": false,
    "duration": 30,
    "moderators": "1",
    "participants": "2",
    "audiences": 6,
    "auto_recording": false,
    "quality": "SD",
    "live_transcription": {
        "language": "english_us",
        "auto_transcribe": true,
        "enable": true
    } 

  }
}

Note: You are not permitted to define a room with live transcription settings if you have not subscribed to this service.

Methods and Notifications

In addition to auto start settings, live transcription can also be started or stopped using the SDK method calls.

When the SDK method is called to start the live transcription, it initiates the live transcription and subscribes to the feed to receive the live transcription. If the transcription process is already started in a video room, the SDK method subscribes to the process.

When a user executes the command to stop live transcription, the user is unsubscribed from the feed and does not receive the transcribed content after this point. When the last subscribed user stops subscription, the transcription process stops in the room.

If the transcription is automatically started at the room-level, the process does not stop when the last user stops. It is only stopped at the end of the session.

The methods and notifications for each SDK are individually documented. For implementation details, refer to the SDK documentation for your platform:

• Web SDK
  
• Android SDK
  
• iOS SDK

Language Codes

Use the following language codes while defining a room for live transcription and while using the startLiveTranscription() method of the supported SDKs: