Answer in brief
A video gets its one chance in silence, so the meaning has to read with the eyes and sound works as amplification. Here are the two kinds of captions, where the line can sit, where free music stops being free, and what the editing packages include.
Video is watched muted, and that is not a guess
Feeds are scrolled on public transport, in queues, next to a sleeping child and in meetings. Sound is rarely switched on in those places, and the video gets its one chance in silence.
What follows is not "sound does not matter" but something else: the meaning has to be readable with the eyes, and sound works as amplification. That order changes the whole edit.
The check is simple. Mute the video, watch it, and try to say what is being offered. If you cannot, the video is built for a condition that usually is not there.
The rest of this piece is how that works technically: how burned-in captions differ from platform ones, what type size survives in a feed, and which of it overlaps with accessibility requirements.
Two kinds of captions and the difference between them
Captions are either burned into the picture or supplied to the platform, which lays them over the video from a separate file or from its own transcription.
Burned-in captions look the same everywhere and do not depend on the viewer's settings. They cannot be switched off, translated automatically or selected as text, but they will definitely be shown.
Platform captions can be turned on and off, some platforms translate them, and search engines read them as text. In exchange you do not control the typeface, the position or the line breaks, and on some devices they will not appear at all.
The sensible arrangement is both layers: a short burned-in line carrying the main point, plus a caption file with the full speech. The video then reads without any settings and stays available to anyone who needs the whole text.
Where captions can sit and where they cannot: The caption line lives in the middle third of the frame,…
The platform's interface takes the edge of the frame for the handle, the description, the buttons and the follow control. A line placed too low ends up underneath those elements and is simply not seen.
So the text lives in the middle third vertically, not along the bottom edge the way it does in cinema. That looks wrong to anyone arriving from horizontal formats, and it is correct.
The second trap is line length. Two lines of a few words read while moving; four lines in small type do not read at all, and the viewer leaves before finishing.
The third is contrast. White text on a bright frame disappears. A plate, an outline or a darkened band under the line exists not for looks but to keep the text readable over any frame.
The type size that survives a feed
Video is nearly always watched on a phone, and often not full screen but inside a feed where it occupies part of the height. Text that looked large in the editing software becomes unreadable there.
Check it on a phone at the size the video will actually be seen, not in the editor. It takes a minute and removes half the notes that would otherwise arrive after publication.
A useful rule is to squint at the line, or hold the phone at arm's length. If the words stop separating, the size is too small, whatever the preview says.
Check digits and Latin characters inside other scripts separately: in some typefaces they are noticeably smaller than the letters and lose legibility first as the frame shrinks.
The first second decides whether anyone reaches the sound
Until the viewer stops, sound does not exist for them. So the first frame has to work in silence and say what the video is about with no line and no voiceover.
That does not mean putting text across the whole screen. It means showing an object, an action or a situation that explains the subject by itself, and captioning it in one short line.
The worst first second is a logo and a title card. They carry no information, and in that time the viewer has already decided whether to scroll on.
If the video opens on speech, that speech needs a caption. A talking head without captions in a muted feed is just a person opening their mouth.
What sound does when it is switched on
Sound does not repeat what the picture already makes clear. It adds what is not in the picture: intonation, rhythm, the feel of a material, silence in the right place.
The most noticeable is edit rhythm tied to the music. A cut landing on the beat reads as care, even when the viewer never works out why.
The second is speech intelligibility. A voice recorded on a built-in microphone in a room with echo is not rescued by processing; it is solved on the shoot with a lapel mic and quiet, not in the edit.
The third is loudness. A video noticeably louder than its neighbours in a feed reads as an advert, and gets muted before it is finished.
Music: where "free" stops
A track from a platform's built-in library is free inside that platform. Moved onto a website, into a deck or into a paid campaign, it stops being covered by that licence.
Hence the common situation: the video sits happily in the feed while the same file draws a claim on the website. That is not a platform failure, it is two different sets of terms on one file.
The practical conclusion: decide where the video will live before choosing the music. If it is going onto a site and into advertising, the track is taken under a licence that permits it.
The VITON13 packages separate this explicitly. Single Reel at $70 includes a licensed track in its scope, while Campaign Edit at $180 does not include licensed music — the licence is purchased by you.
Voiceover: when it helps and when it is in the way
A voiceover looks like a way to explain everything at once, but in a muted feed it does nothing at all, and in a played one it competes with the music for attention.
It is justified where intonation and attitude have to come across: a founder speaking, an explanation of something complicated, a client's story. There the voice adds what text cannot.
It is surplus where the picture already shows everything. A voice narrating what is on screen slows the pace and lengthens the video for nothing.
And in either case the voice needs captions. Otherwise half the audience receives a video with the substance removed.
Accessibility requirements line up with commercial interest
For prerecorded video, WCAG 2.2 requires captions at Level A. It is a formal requirement, and it is also exactly what a person watching without sound needs.
The same guidelines carry requirements for text contrast and for not conveying meaning by colour alone. Both apply directly to a caption line.
It helps to read this as a coincidence rather than an obligation: what makes a video accessible is what makes it legible in a feed. One piece of work, two returns.
It needs no separate budget line. Captions are built into the edit from the start rather than added once the video has been approved.
What the file handover must contain
A finished video is not one file. You need versions for the places it will live, and separately the material those versions can be rebuilt from.
The minimum set: an export for every required format, a version without burned-in captions, and a separate caption file if you intend to upload captions to the platform.
The container matters: MP4 carrying H.264 video and AAC audio opens in current browsers and on phones without installing anything, and that is a sensible default.
And the edit project with its media linked. Without it, the next change is not a revision but a rebuild.
What it costs and what is included
Sound and caption work is not priced separately — it sits inside the editing packages, and what differs between them is volume, not whether captions happen.
Single Reel is $70 for one vertical edit up to 45 seconds, 1-2 working days, one revision round; captions and a licensed track are included, while shooting, scripting and actors are not.
Reels Pack is $90 for up to five edits, 1-2 working days from footage delivery, two rounds on the edit; filming is not included and you supply the footage. Reels Pack Express is $150 for the same scope with a 24-hour turnaround, one working day, one revision round.
Campaign Edit is $180: a hero cut and short derivatives, 3-5 days, two rounds on the cut and one on the grade; licensed music is not included. Monthly Content Engine is $290 per month on a monthly cycle with 30 days notice, and the volume is agreed at the start of each cycle.
The checklist before publishing
Watch the video muted on a phone and try to say what is being offered. If you cannot, the problem is not the viewer.
Check that the caption line does not disappear under the platform interface and stays readable over the brightest frame in the edit.
Match where the video will live against what the music licence allows — before publication, not after the email arrives.
Collect the version without burned-in captions, the separate caption file and the edit project: without them the next revision costs what a new video costs.
Practical checklist
- Watch the video muted on a phone and try to say what is being offered.
- Lift the caption line off the bottom edge into the middle third of the frame.
- Check the line stays readable over the brightest frame in the edit.
- Decide where the video will live before choosing the music.
- Ask for a version without burned-in captions and a separate caption file.
- Collect the edit project with its media linked.
Questions and answers
Why is video watched muted?
Feeds are scrolled on transport, in queues, next to a sleeping child and in meetings, and sound is rarely switched on there. What follows is not that sound is pointless but that the meaning has to read with the eyes, with sound as amplification.
Burned-in or platform captions?
Both. Burned-in captions always show and do not depend on viewer settings, but cannot be switched off, translated or selected as text. Platform captions are read by search engines as text and get translated, but you do not control the typeface or position, and on some devices they never appear.
Where must the caption line not sit?
On the bottom edge: the platform interface takes that part of the frame for the handle, description and buttons, and the line ends up underneath. Text lives in the middle third vertically, in two short lines, with a plate or outline for contrast.
What does an edit with captions cost at VITON13?
Captions sit inside the editing packages and are not priced separately: Single Reel is $70 for one vertical edit up to 45 seconds, 1-2 working days, one revision round; Reels Pack is $90 for up to five edits with two rounds on the edit; Campaign Edit is $180 over 3-5 days. Filming is not included in any of them.
Can we use a free track from the platform's library?
Inside that platform, yes. On a website, in a deck or in a paid campaign that licence does not cover it, and that is where claims come from. Single Reel at $70 includes a licensed track in its scope; Campaign Edit at $180 does not include licensed music — the licence is purchased by you.