Tavus disclosure_type controls vs Sume authored caption cues
Tavus added disclosure_type, verbal_disclosure and visual_disclosure for EU AI Act use. Sume has no such setting; authored caption cues can burn a line of text.
Tavus's changelog lists EU AI Act controls: disclosure_type, verbal_disclosure and visual_disclosure on the PAL, plus a per-conversation policy of "eu". Sume documents no disclosure setting. What it does have is authored caption cues that burn a line of text onto a video without running speech-to-text, which you can use for a visible line you write yourself.
This is not legal advice, and the Tavus entry is undated in the snapshot read 2026-10-01. Sume facts are from Video captions and Generate avatar video.
What does the Tavus entry say?
The entry says the controls apply the disclosure automatically and switch Raven-1 to limited emotion recognition when PAL fields are left on auto. It describes settings on a live conversation system. Sume's docs list no equivalent disclosure setting.
How do authored cues work on Sume?
Standalone captions take a public HTTPS video_url. Pass cues (or segments) with text, start and end to burn authored overlay text; this skips speech-to-text. Without cues, captions need audible speech, and a silent clip fails as caption_no_speech with next_action: use_overlay_captions.
curl -X POST https://api.sume.com/v1/video-captions \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: caption-cue-001" \
-d '{"video_url":"https://example.com/clip.mp4",
"cues":[{"text":"Your own line of text here","start":0,"end":4}]}'What differs for inline captions on avatar video?
| Route | Text source | Notes |
|---|---|---|
POST /v1/video-captions | cues, script_text or speech-to-text | Takes an existing public video URL |
Inline captions on talking-video | The spoken script text | Rejected above 60 seconds estimated; no separate billed caption job |
What should I be careful about?
Whether a burned-in line meets any rule is for your counsel and the rule text. Sume's docs do not describe a provenance or watermark feature. For a related first-frame approach, see the AI label compose post.
Sources
Related posts
More in Developers
- Tavus recording storage key_template vs Sume media URLs
Tavus can write recordings to S3, GCS or Azure Blob. Sume returns finished files as media.sume.com artifact URLs, delivered by webhook or job result.
- Temporal Activity ID policies are not a Sume Idempotency-Key
Temporal's Conflict and Reuse policies dedupe Activity IDs inside Temporal. They never reach Sume, so a paid submit still needs its own Idempotency-Key.
- Temporal Maximum Attempts 1 and a paid Sume submit
Standalone Activities default to at-least-once retries. Maximum Attempts 1 gives at-most-once, but a Sume Idempotency-Key lets you keep retries safely.
- Threads API: wait about 30 seconds before publishing a container
Threads suggests waiting on average 30 seconds after creating a media container before threads_publish. Poll the container status, like a Sume job poll.
Written by Sume