Live webinar captions now, burned-in highlight clips after

Sume does not offer live captions. For a webinar, use a real-time transcriber during the event, and use Sume afterwards to burn captions onto the short highlight clips you post. A 90-minute webinar with MAI-Transcribe-2-Streaming at its $0.54 per hour introductory price is $0.81, and ten highlight clips at $0.20 each add $2.00, so $2.81 in total.

The split is the honest one: live work needs a streaming service, and finished-clip work is what Sume's caption route is built for.

The live half

Microsoft's launch post says MAI-Transcribe-2-Streaming emits first partial hypotheses "in just over 100ms" in 60 languages with automatic language detection, and the Learn page lists the Realtime API and the Azure Speech SDK as integration paths. It is a public preview without an SLA (read 2026-10-08), so show the captions as a convenience, not as an accessibility guarantee, unless you have a human fallback.

The recorded half

After the event, pick up to ten moments, cut each to 60 seconds or less and host the files at public HTTPS URLs. Sume's caption route burns captions on a clip of up to 60 seconds for $0.20. The recording of a webinar is rarely a single short clip, so the 60-second ceiling means you caption highlights, not the full session.

For the full-length transcript, Sume STT can transcribe the recording in 600-second slices at about $0.01 a minute, which is $0.90 for 90 minutes (nine slices at 10 cents).

Request for one highlight

Use language as a hint if the webinar is not in English, and check the job state before you resubmit a clip.

curl -X POST https://api.sume.com/v1/video-captions \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_url": "https://example.com/webinar-highlight-03.mp4",
       "language": "en", "style": "punch"}'

Picking highlights

Choose clips that work on their own: a clear question and a clear answer, a demo moment or a customer-facing announcement. Aim for 30 to 55 seconds so that the clip stays within the 60-second caption limit with room for a trimmed head and tail. A transcript with word timings makes this fast; search it for the phrases from your run sheet.

Budget the clips separately from the live service. If a 90-minute webinar yields ten clips you will post, the caption bill is $2.00 whether the live captions were perfect or not, and a post-event pass also lets you fix names that the live service got wrong.