Caption a video over 10 minutes: Sume's 600 s row limit
Sume's veed-subtitles row accepts clips of 2 to 600 s. A 25-minute video needs three parts, each billed at least 60 s. Splitting plan inside.

A video longer than 10 minutes cannot go through Sume's veed-subtitles row in one job: duration runs from 2 to 600 seconds. Split a 25-minute video into three parts of at most 600 s, caption each, and join them. Each part is billed for at least 60 seconds, which only matters for tiny tail pieces.
The limit is Sume's, not VEED's. The repo notes that VEED itself takes up to 30 minutes, and that Studio's speech-to-text (STT 1.0) stops at 10, so the row stops there too.
Splitting plan
Keep every part at or under 600 s and avoid a very short last part, because of the 60 s floor. The table shows common lengths.
Do the cutting with a tool that does not re-encode more than it must, and keep the audio track continuous across parts so the speech at each seam is complete. After the jobs finish, download the captioned parts, join them in order and spot-check the seams. Because every part is a separate job, a failure in one does not touch the others: re-run only that part.
| Video length | Parts | Part lengths | Billed seconds in total |
|---|---|---|---|
| 9 min | 1 | 540 s | 540 |
| 12 min | 2 | 360 s + 360 s | 720 |
| 25 min | 3 | 500 s x 3 (with 1500 s total) | 1500 |
| 10 min 20 s | 2 | 310 s + 310 s | 620 |
| 10 min 5 s (605 s) | 2 | 300 s + 305 s | 605 |
Why even splits beat one long part and a stub
Splitting a 605 s video into 600 s plus 5 s works, but the 5 s stub bills 60 s: 660 billed seconds against 605. Two near-equal halves cost exactly the video's length. Choose cut points at a pause in speech so no word is cut and the subtitles join cleanly.
Style and the 4K rule apply per job, so use the same captions.style, position and resolution on every part to keep the look identical across the cuts.
With your own transcript
If you already have an SRT, send it as captions.srt in each part with times rebased to that part's start. VEED then skips its own transcription. Omit it and VEED transcribes each clip, which is simpler but gives you no chance to edit the lines first. The SRT field has a length ceiling far above 10 minutes of speech, so it is never the limit.
For editing lines before the burn, see auto-generate subtitles, edit, then burn. Styles and their billing classes are in the 30 VEED styles. The router is documented at the Video Router page.
Sources
Related posts
More in Media tools
- Free in-browser video upscaler (10 files, 5 min) vs an upscale API
A free upscaler added batch mode on Oct 11: 10 files, 100 MB, 5 minutes each, in-browser. When Sume's video upscale API fits better, and what it costs.
- Hailuo H3 Regen on Sume: regenerate one H3 clip at 2K
hailuo-h3-regen re-prompts a clip you made with MiniMax H3 and returns 2K. The source must be 768p H3 output, 4.5-15 s with audio. Read 2026-10-13.
- Short clip lipsync: Kling bills 5 s steps, LatentSync 40 s
A 6 s clip on Sume Kling Lipsync bills 10 s; a 12 s clip on LatentSync bills 40 s. The billed seconds for both rows, with a table.
- Lip sync an existing video to new audio: 8 Sume video rows
Eight Video Router rows re-sync a clip you already have to new audio: video_url plus audio_url, 2 to 60 s on most, 9 s on InfiniteTalk Video.
Written by Sume