Skip to content
English
  • There are no suggestions because the search field is empty.

How do I get started to create live transcripts, subtitles and translations?

Step-by-Step Guide to Setting Up Live Transcripts, Subtitles, and Translations in Deep Live Hub


1. Create a New Workflow:

  • Login to your Deep Live Hub account.
  • Click the red "+ Workflow" button and choose "Live Stream" from the top menu.
  • Select one of the configuration templates: "Custom," "YouTube," or "Twitch." If you are not streaming directly to YouTube or Twitch, choose the Custom template.
  • Enter a name for your workflow in the "Name" field and click "Finish" to confirm.

2. Configure the Workflow:

  • After creating the workflow, you’ll be directed to the Workflow Configuration Page. You’ll see tabs like "Overview," "Process," "Outputs," "Streams," and "Settings" at the top.
  • Navigate to the Process tab.

3. Set Up Translation and ASR (Automatic Speech Recognition):

  • In the Process tab, select your translation languages by choosing from the dropdown menu. Click on a language to select it, and click again to deselect. Transcripts and subtitles for the input language are generated automatically.
  • If you’ve created dictionaries in your account, select one under the Process page to enhance accuracy.
  • Choose an ASR configuration to set the processing time (between 3-20 seconds) for generating transcripts. Longer processing times yield better grammar and translation accuracy but cause a delay in output.
    • Enabling "Partials Mode" will deliver the fastest possible results, with corrections sent at the end of the processing time (Note: Partials only work for transcripts, not translations).
  • Optionally, enable a live editor by toggling the option slider to the right.

Click "Save" after each configuration change.

ASR Configuration
The ASR Configuration allows you to control how long the AI processes audio before generating transcripts and translations. You can set the processing time between 3-20 seconds. Longer times result in better grammar and translation accuracy but add a delay to the output.

Partials Mode delivers faster, real-time results with corrections sent at the end of the full processing time (only available for transcripts, not translations).

Example:

  • If you choose a 5-second processing time, the AI will process the audio for 5 seconds before generating a transcript, offering a quicker response but with less context for accuracy.
  • A 20-second processing time gives the AI more context, resulting in higher-quality transcripts and translations, but with a slight delay.
  • In Partials Mode, the AI delivers immediate transcript results and sends corrections later if needed.

4. Set Up Output Configurations:

  • Navigate to the Outputs tab. You can create a new output configuration from the left column.
    • By default, an SRT Configuration and a Video Output Pull Configuration are provided:
      • The SRT Configuration creates an SRT file containing transcripts and translations.
      • The Video Output Pull Configuration generates a video stream with subtitles (if "Burned-In" or "Closed Captions" are enabled).
    • To configure Burned-In or Closed Captions, click on Video Output Pull Configuration and adjust the settings in the right column.
    • You can also create a Video Output Push Configuration to stream content to a content delivery network or video platform. This can also done as multiple streams with a variety of settings.
    • Create a Website Output Configuration to display live transcripts and translations on an auto-generated website.
    • A Text Configuration enables real-time text streams (transcripts or subtitles) for HLS-Realtime output.

Stream Output Types:

  • SRT Configuration generates subtitle files.
  • Video Output Pull adds subtitles to the video.
  • Video Output Push streams to external platforms.
  • Website Output displays live transcripts and translations on an auto-generated website.
  • Text Configuration delivers real-time transcripts and subtitles as a text stream.

5. Collect URLs and Stream Keys:

  • Go to the Overview tab to collect the Ingress URLs (RTMP, WebRTC) and the stream key for your input.
  • If you’ve enabled a live editor, the editor’s URL and PIN will be available under the Processing menu.
  • Egress URLs, including the latest SRT, RTMP, HLS, and HLS-Realtime links, as well as the link and PIN for the live viewer, are available under the Output URL menu.

6. Stream Management:

  • View a log of past and current streams in the Streams tab.
  • To disable or delete a workflow, go to the Settings tab. Note that disabling an endpoint won’t disrupt active streams—stop the stream in your encoder first to make the workflow inactive.