Why a caption file gets rejected
Captions fail more deliveries to US public television than picture or sound, and almost never because the words are wrong. They fail on nine things, and every one of them is visible before you send.
1. The format is not the one they can read
For a sidecar file, NETA takes .scc and nothing else — not .srt, not .vtt, not .mcc. Those formats are fine for the web and for review; they are not broadcast caption files. Embedded captions are the other accepted route, and they have their own rule, below.
2. The file name does not match the video
show.mxf needs show.scc, character for character. Spanish captions are show_SP.scc, and when a programme has both English and Spanish, both go as sidecars with nothing embedded. An automated system pairs the two files by name; if the name is off, there are no captions at all as far as it is concerned.
3. The caption file starts on a different timecode than the master
A master that starts at 01:00:00;00 needs a caption file that starts at 01:00:00;00. Export the captions from a sequence that starts at zero and every caption lands an hour away from its line.
4. Drop-frame and non-drop are mixed
At 29.97 fps, drop-frame and non-drop-frame count the same frames with different labels. A caption file written in one against a master in the other drifts by about three and a half seconds per hour: fine at the top of the programme, a line and a half late by the end. It is the failure most often described as “the captions go out of sync towards the end”.
5. Embedded 708 without the 608 bytes
Where captions are carried inside the file, they must be 708 captions carrying 608 compatibility bytes, in a SMPTE 436M ancillary data track. A file with 708 only is not accepted. This is not a preference — older decoding paths and station equipment read the 608 stream, and if it is absent the captions disappear on part of the audience’s screens.
6. Timing that drifts away from the voice
Captions should land within about a second of the spoken word. Drift comes from cutting picture after the captions were written, from a conform that moved a scene, or from authoring against a rough cut. It is invisible in a caption editor and obvious on the picture.
7. Reading speed
A caption that is correct and on time can still be unusable if it is on screen for eight frames. The US convention for captioned programming is to keep the reading speed within what an average viewer can follow, which in practice means giving each caption a minimum on-screen time and breaking long sentences across captions instead of cramming them.
8. Too many lines at once
One to three lines on screen at a time. Two is what reads best, and three is the ceiling, not the target. Four-line blocks arrive when a caption is converted from a subtitle format without re-breaking the text.
9. Placement over something that matters
Captions must never sit over name keys, on-screen text, faces or important action. The fix is a caption moved to the top of the frame while the lower third is up, and back down afterwards. It is the last thing anyone checks and the first thing a viewer notices.
What a caption audit gives you
For the free audit, you get whether the file conforms and how many issues there are, by type. The paid audit adds the timecoded list — every caption that needs moving, retiming or re-breaking, with the timecode and the reason, so the fix is a task and not an investigation. If you would rather not touch it, I can author the English captions in-house and deliver them with a signed caption validation statement.
Have the caption file checked before it ships
The first Delivery QC Audit is free — one free audit per production company, one file up to 60 minutes — with a written report in two business days, every item measured or checked on the picture.
Read next
Written by Cleared for Air. The FCC caption rules and the caption style references are linked under Resources. Your distributor’s own requirements are the ones that count; check them before you deliver. © 2026 Cleared for Air.