Burn feature text onto silent product clips with caption cues

To put feature text on a silent product clip, call POST /v1/video-captions with the clip's public HTTPS URL and a cues array of text, start and end in seconds. Sume skips speech-to-text and burns exactly your copy at those times. Without cues, a clip with no speech fails with caption_no_speech.

Short, readable on-screen text matters when video is the storefront. Best Buy's TikTok Shop debut in late October 2026 leans on shoppable creator videos (read 2026-10-04), so the clip itself has to say what the product is without sound.

Why do silent clips fail the default path?

Speech-to-captions, whether from script_text or from speech-to-text when it is omitted, needs audible speech. The docs mark a silent clip as caption_no_speech with next_action use_overlay_captions. That is a pointer to the cues path, not a dead end.

What are the request options?

Only video_url is required.

What does the call look like?

Keep each card short, and keep end times inside the clip. The call returns a job; read it from the job endpoints.

import os, requests

r = requests.post(
    "https://api.sume.com/v1/video-captions",
    headers={"Authorization": "Bearer " + os.environ["SUME_API_KEY"]},
    json={
        "video_url": "https://example.com/earbuds-silent.mp4",
        "style": "slam",
        "cues": [
            {"text": "Fits in one hand", "start": 0.5, "end": 2.5},
            {"text": "Charging case included", "start": 3.0, "end": 5.5},
        ],
    },
    timeout=60,
)
print(r.status_code, r.text[:300])

What text should you burn?

Only facts you can show or stand behind: what is in the box, a size, a real feature. Do not invent specifications. Check the final frame on a phone screen, since TikTok Shop's interface may cover part of the picture, and confirm its current safe areas on TikTok's pages.