The new AI transcription feature enables real-time transcription of audio from live productions. These transcriptions can be embedded as EIA-608 via CTA-708 closed captions, making them displayable on most downstream platforms and players. Alternatively, you can use Open Captions burned into the video.
Using AI Transcription
When AI transcription is enabled in a Producer control room, it can be toggled on and off via the new configuration sidebar:
Configuration Options
Switch
A toggle at the top to turn transcription on or off.
Language
Select the language for transcription. Available options include:
English (US)
English (GB)
English (AU)
Spanish
French
Portuguese
German
Danish
Italian
Delay
Defines the target delay for real-time transcription. A longer delay can provide better accuracy by giving the AI more context. Sub-2-second delays have shown excellent results in testing.
Volume Threshold
Helps filter out background noise, improving accuracy in noisy environments.
Custom Dictionary
Allows users to add domain-specific terms, such as company or personal names, for better transcription accuracy.
๐ Note: Changes to any settings during an active transcription session require restarting the transcription or clicking the link in the notification to apply updates.
โ Tip: For the best results, ensure that audio quality is high and background noise is minimized.
Embedding Transcriptions as Closed Captions
Enable transcription via the control room sidebar.
Navigate to the output settings.
Turn on closed captions and select AI TRANSCRIPTION as the source.
๐ Note: Embedding closed captions is separate from other potential applications like open captions and translations, which will be supported in the future.
How to use Open Cations
Ensure that AI TRANSCRIPTION is en
Navigate to the plus button to add OPEN CAPTIONS as a source
Once added to the layer, proceed to the OPEN SETTINGS icon to customise the captions to your preference.
Push the OPEN CAPTION source to PROGRAM to play live.
๐ Note: Open Captions are only available if the AI Transcription is enabled on your Producer. As Open Captions are considered a web source, standard source limits will apply.
โ Tip: Just like text overlays, open captions support full customisation, including font style (with custom fonts), weight, colour, letter spacing, positioning, and shadow effects. Additionally, two new settings allow you to configure:
Line Width โ Sets the maximum number of characters per caption line (except for long single words).
Number of Lines โ Defines how many caption lines appear on screen.
๐ Note: When a speaker change is detected, a new line is created, prefixed with ">>". Open captions function like any other source, allowing positioning, layering, and integration into preview, program, or clean feeds.
By default, captions are positioned at x:10 and y:1070, but adjustments may be needed for resolutions other than 1920x1080.