LinkedIn ad video captions: one English SRT, and it must process

When you tell LinkedIn's Videos API you are uploading captions for an ad video, caption processing has to succeed or the video is not eligible to serve. Each video takes one caption file and only English captions are supported. A feed video is more forgiving and can still serve if its captions fail.

What does LinkedIn say about the caption file?

In initializeUploadRequest, uploadCaptions defaults to false. When true, the response includes a captionsUploadUrl. The page says each video can include only one caption file and only English-language captions are supported. For an Ads Video, caption processing must succeed for the video to be eligible for serving, while a feed video can still serve if caption processing fails. The upload example sends SRT with a metadata form part and a file part, and a 201 means success.

Does Sume give you an SRT?

Not from what I could confirm. Inline avatar captions are burned into the final MP4 after generation, using styles such as slam or punch, and the Video captions docs say SRT uploads are unsupported as input. I found no SRT output in the docs, so I will not claim one.

That means two separate things: burned-in captions help silent autoplay on any platform, while LinkedIn's caption file is a different asset you supply yourself. If you want both, make the burned-in version in Sume and write the SRT in your own tooling from the same script.

What about non-English avatar videos?

LinkedIn's one English file rule is a constraint on the caption file, not on the video. If your avatar speaks another language, burned-in captions remain the way to show on-screen text, and the file upload would not match the audio. For an ad, skip uploadCaptions entirely rather than risk a failed processing run that blocks serving.

• Ad video, English speech: upload one English SRT and confirm processing before serving.
• Ad video, other language: burn captions in Sume and leave uploadCaptions false.
• Feed video: a caption failure will not stop it serving, but check anyway.

Inline captions from Sume

The captions object on the avatar request is optional. Stage failures soft-fail: the job can still succeed with a clean video_url and captions.status=failed. Estimated duration above 60 seconds is rejected for inline captions.

{
  "captions": {
    "enabled": true,
    "style": "slam",
    "language": "auto"
  }
}