Subtitles vs Closed Captions [2026]: 3 Types, $1.99/min

Subtitles, closed captions and open captions all put words on screen, but they differ in who controls them. Subtitles are a separate text track that assumes the viewer can hear the audio. Closed captions are a text track that also describes sound, and the viewer can switch them off. Open captions are burned into the picture, so nobody can.

That one difference decides your file format, your cost per video and how much room you must leave around a lower third. It also decides whether fixing a typo takes ten seconds or a full re-render. YouTube’s own blog reports that microdramas passed 6.5 billion views in the first half of 2026, with watch time up more than 50% year over year. Serialized vertical video like that forces creators to choose between a switchable track and text baked into the frame, often several times a week.

This comparison of subtitles vs closed captions vs open captions sets the three types against each other on specs, file formats, benchmarks and 2026 pricing from Rev, Submagic, Captions and Adobe. If you only need an editor pick, our guide to auto captions in five editors covers that side by side. Everything below is checked against vendor pages, release notes and the US federal rule text as of November 4, 2026.

Subtitles vs Closed Captions vs Open Captions: The Short Answer and 13 Spec Differences

Use closed captions (a sidecar track the viewer can toggle) as your default. Add open captions only for feeds where autoplay runs muted and you need the words to show every time. Use subtitles when the goal is translation for viewers who hear the audio but do not speak the language.

The labels blur in everyday speech, and platforms add to the confusion. YouTube Studio, for example, uses one creator-facing label, Subtitles, for the page where you add both subtitles and closed captions, according to YouTube’s help documentation as of February 2026. So the words on the button tell you little. What matters is what the text contains, where it lives and whether the viewer can turn it off.

Here is the working definition set for planning a project:

• Subtitles: timed text of the dialogue, usually in the viewer’s language or a translation. It leaves out most sound information.
• Closed captions: timed text of the dialogue plus meaningful non-speech audio such as music cues, sighs or an alarm. The viewer controls visibility.
• Open captions: timed text rendered into the video pixels at export. No toggle, no separate file, and no way for the player to restyle it.

The table below puts the three side by side. The regulatory row comes from the eCFR text of 47 CFR Part 79, which is the federal rule for captioning covered TV programming.

Subtitles: Translation and Sound-On Viewing

The subtitles vs closed captions question comes down to sound. Subtitles exist to carry meaning across a language gap. The viewer hears the speech, hears the music and the footsteps, and needs only the words. That is why a subtitle track can skip a line such as a door slam or a phone buzz and still do its job.

In practice you will deliver subtitles as a sidecar file. SubRip (SRT) is the plain option, and WebVTT is the web-native one. Both hold numbered or timed cues and text, and both let a platform show or hide the result. You upload one file per language, and the viewer picks from a menu.

Rev lists translated subtitles at $6.49 to $15.99 per minute depending on language, which is the clearest published price for human translation in the sources checked. Multiplied across a whole video, that adds up for every extra language. That is why most small channels translate only their top few videos by hand and let a tool handle the rest.

Translation tooling moved quickly this year. Adobe’s Premiere page advertises caption translation inside the editor, and Adobe’s beta announcements reportedly describe translation into 17 languages. Treat that count with care, because Adobe’s own pages give a different number for the shipping feature. Captions, the AI editing app, and Submagic both sell translation as part of their plans too. Submagic’s help centre reportedly prices translation at 1 extra credit per video on top of the 3-credit upload cost.

Subtitles also have a hearing-impaired cousin, often written SDH (subtitles for the deaf and hard of hearing), which adds sound cues to a subtitle file. If you see that term in a delivery spec, read it as closed captions delivered in a subtitle file. The distinction matters when a distributor’s spec asks for one or the other by name.

For creators, the practical test is simple. Play the video with the sound off and read only the text. If you lose the plot, you have subtitles. If you keep it, you have captions. Run that test on your last three uploads and you will learn which type you have been publishing without meaning to.

Where subtitles fall short is accessibility. A viewer who cannot hear the audio gets no information about a laugh track, a warning tone or off-screen speech. If any part of your audience depends on text for sound information, plain subtitles are the wrong tool.

Closed Captions: Sound Cues, Toggles and What the FCC Rule Covers

Closed captions carry the dialogue and the sounds that change how a scene reads. A viewer with the sound off, in a loud gym or on a train, gets the same story as a viewer who hears everything. The closed part means the text stays hidden until the viewer asks for it.

YouTube shipped a good example in December 2025. Expressive Captions arrived for English-language videos across devices. The feature uses AI to show the intensity of speech in capital letters, along with sighs, gasps and environmental noise. YouTube’s announcement page confirms the feature remains available.

For creators, the lesson is that closed captions are a viewer-side layer. You supply the track, and the platform decides how to draw it. You can improve the track by writing sound cues into it, but you cannot force a font or a screen position.

What the US federal rule says

The federal text is narrower than most creators assume. Under 47 CFR 79.4(a)(1), the term video programming means programming provided by, or comparable to programming from, a television broadcast station, and the definition expressly excludes consumer-generated media. So a typical YouTube tutorial or stream VOD does not fall under that internet captioning rule.

The same regulation says, in section 79.1(e)(2), that “Open captioning or subtitles in the language of the target audience may be used in lieu of closed captioning”. That sentence is why a TV distributor can accept a burned-in version in some cases. It does not tell you what a platform or a client will accept, so read the delivery spec first.

Two definitions in the rule help when you talk to a broadcaster. Live programming is video shown on television substantially simultaneously with its performance. Near-live programming is performed and recorded less than 24 hours before it first airs. Prerecorded programming is everything else. If a client asks which category your show falls into, those three lines settle it.

Where closed captions go wrong

Platform bugs. YouTube’s Community support pages reported in October 2026, with an update on November 2, that some creators could not upload or edit secondary-language subtitle files (.srt and .sbv) in the new Studio layout. Users also reported intermittent caption failures between September 22 and November 3. These are user reports, not confirmed official changes, but they argue for keeping a local copy of every caption file you publish.

Open Captions: Burned-In Text and Lower-Third Clearance

The closed captions vs open captions choice is mostly a question of control. Open captions are part of the image. You choose the font, the weight, the outline, the colour and the position, and every viewer on every device sees exactly that. The trade is permanence. If you misspell a guest’s name, you fix it by re-rendering the whole video.

The format earns its place on feeds that autoplay without sound. A viewer scrolling past your clip decides in a second or two whether to stop, and readable text in the frame helps that decision. It also survives re-uploads, screen recordings and shares to apps that strip sidecar tracks, because the words travel inside the pixels.

Single-word styles are the current fashion. Adobe added Single Word caption generation in Premiere 26.3 in June 2026, which times each word to the audio for a fast, social-video look. Adobe’s own what’s-new page describes customizing the style and exporting the captions exactly as they appear in the preview. Whether that suits your content depends on your pacing, since a slow interview rarely benefits from one word at a time.

The lower-third problem

Open captions compete with everything else you put on screen. A name strap, a channel logo bug or a sponsor tag all want the same bottom band. If you build lower thirds, plan the collision before you export. Our walkthrough on how to make lower thirds covers sizing and timing, and the rule that carries over here is simple: when a lower third is on screen, move the captions rather than stacking them on top.

On 9:16 video the problem sharpens. Platform buttons, usernames and descriptions sit over the bottom and right edges of the frame. A caption that looks fine in your editor can vanish under an interface layer on a phone. Check placement in each platform’s preview and leave margin on both sides.

Styling limits you should know

Open captions cannot adapt. A viewer with low vision cannot enlarge them, and a viewer who finds them distracting cannot hide them. If your audience includes people who depend on text, publish a closed-caption track alongside the burned-in version. Many creators keep both: a captioned cut for the feed and a clean master with a sidecar file for accessibility and search.

Caption File Formats: SRT, WebVTT and the Adobe Premiere List

Formats matter because they decide where a caption file can travel. SRT is the plain, widely accepted option, holding cue numbers, timestamps and text. WebVTT is the web video format, and it adds cue settings such as position and limited styling. YouTube’s creator tools accept SRT and, per Community reports, SBV for subtitle uploads.

Adobe’s supported-formats page, dated October 12, 2026, lists the caption and subtitle formats Premiere imports. The list shows how far caption files stretch from the web into broadcast.

CEA-708 sits underneath much of this. It is the digital TV caption data standard, and Premiere 26.5, released in September 2026, improved how it imports wide-aspect CEA-708 captions. The same release added preferences for the default fonts Premiere uses for text, subtitles and closed captions under Preferences, then Graphics. If you work on broadcast deliverables, that release matters. If you only publish to the web, you will likely never touch it.

Working rule: keep one clean SRT as the master for every video. It converts to WebVTT and most other formats with little fuss, and it opens in a plain text editor if a platform breaks. Save it with a language code in the file name, use UTF-8 text, and never edit it in a word processor that adds smart quotes.

Frame rate matters less to sidecar files than to burned-in text, because SRT stores times rather than frames. Even so, a caption file timed against a 30 fps edit can drift if you re-cut the video at a different rate, so retime after any change to the master. If you build the burned-in version from the same timeline, the problem never appears, which is another reason to treat the edit project as the source of truth.

SBV deserves a mention only because YouTube users report it in connection with upload problems. If a tool asks for SBV and you only have SRT, convert instead of retyping. Converters are common, and the text rarely needs changes beyond the timestamp format.

Audience Data, Accuracy Claims and Per-Minute Costs: Benchmarks From 8 Sources

The benchmarks below come from audience research and vendor pages. None is an independent lab test of caption accuracy, so read the accuracy numbers as vendor claims.

What viewers do with captions

The audience numbers explain why captions are no longer niche. These five statements come from research collected in VoxBooster’s 2026 caption statistics roundup and Cablecast’s blog, and each one is quoted word for word with its original year.

Statista’s U.S. figures put it this way for 2023: “41% of U.S. adults used subtitles or closed captions when watching television shows or movies in August 2023.” YouGov found, in a 2023 survey reported by VoxBooster, that “63% of U.S. adults under 30 preferred subtitles turned on when watching television in a language they knew, compared with 38% of adults overall.”

Age splits the habit sharply. In BBC research (the cited page does not give a year), “80% of viewers aged 18–35 used subtitles while watching television, compared with less than 25% of viewers aged 56–75.” The Audience Agency for Stagetext reported in 2023 that “42% of respondents in the United Kingdom cited concentration as a reason for using captions or subtitles.”

The research base is older than the platform boom. Gernsbacher’s 2015 review in Policy Insights from the Behavioral and Brain Sciences states that “More than 100 empirical studies have documented that captioning a video improves comprehension, attention, and memory for the video.” Most of these figures date from 2023 or earlier, so treat them as direction rather than a snapshot of November 2026.

Accuracy claims and cost benchmarks

Vendor accuracy claims depend on audio quality, accents and jargon, and none of these pages publishes a test set. For channel names, product terms and anything legal or medical, plan a human review pass either way. Rev’s human price of $1.99 per minute also gives you a floor for what a professional pass costs.

Pricing: What Captions Cost in November 2026

Pricing falls into three models: pay per minute for humans, subscribe per month for AI apps, or get captions bundled inside an editor. The table lists current plan names and prices from the vendors’ own pages as of November 4, 2026.

A few notes that the table cannot carry. Captions renamed its plans in September 2026, so older reviews that mention Pro or Scale describe the same tiers under new names. Rev also lists a permanent free plan with 45 minutes of AI transcription per month and AI transcription at $0.25 per minute, though those two figures come from a single page. Mark Studios’ October 15 guide describes YouTube’s automatic captions as free and available in more than 60 languages, which matches how most creators use them.

Cost per video, worked out

Dividing plan price by video cap shows what a full month of use costs per video. This assumes you use every slot.

Compare that with a human pass at $1.99 per minute. A 60-second Short costs $1.99 through Rev, while Submagic Pro is $39 per month for 40 videos, or $23 per month billed annually. A 12-minute tutorial costs far more by human, and at that length it can make sense to run AI first and pay a human only for the final proofread.

Tool by Tool: YouTube Studio, Premiere, Submagic, Captions and Rev

Each tool sits in a different spot on the closed-versus-open spectrum. This section covers what the vendor pages show about caption handling, with the newest versions as of today.

YouTube Studio

YouTube Studio is the default home for sidecar tracks. According to YouTube’s help documentation as of February 2026, you open Subtitles, choose a language and add captions by uploading a timed or untimed file, auto-syncing a transcript, typing by hand or using automatic speech recognition. There is also a Subtitle Editor channel role, so you can hand caption work to a colleague without sharing full access.

Studio produces closed captions, not open ones. Burned-in text is not a Studio caption-track feature, so you add it in your editor before upload. For running coverage of YouTube feature changes, 9to5Google’s YouTube guide keeps a news trail worth bookmarking.

Adobe Premiere

Premiere is the tool that handles all three types in one place. It generates captions with speech-to-text, imports SRT and other caption files, exports caption tracks, and renders styled captions into the picture. Adobe released Premiere 26.2 in April 2026, when Color Mode entered public beta, then 26.3 in June with Single Word captions, and 26.5 in September with the font preferences and CEA-708 import work.

Adobe’s own pages use both Premiere and Premiere Pro, and nothing checked confirms a formal rename, so expect both names in search results. Community threads from October 2026 say Premiere still lacks a fully featured native animated or karaoke-style caption workflow, which is why plugins stay popular. Adobe’s Firefly documentation also warns that some captions may not transfer when you send a video from Firefly to Premiere.

Submagic and Captions

Submagic and Captions are built for the open-caption look. Both generate styled captions fast and render them into the export. Submagic’s pages describe AI Auto Captions with translation and custom emoji as add-ons, plus newer tools such as IRIS, which the company introduced as an AI agent in an October 28, 2026 tutorial. Captions describes itself as an AI video editor whose caption tools cover subtitles, closed-caption-style text and open captions burned into exports. Its newest avatar model, Mirage Avatar X, arrived on July 28, 2026, though that matters little for caption work.

Rev

Rev sits on the human side. Its pages describe human and AI transcripts, captions and subtitles, including burned-in open captions and foreign-language subtitles. Rev also announced a redesigned mobile app for legal work on November 2, 2026, a reminder that its newest product work is aimed at legal teams more than at video creators.

Real-World Examples: Eight Caption Choices

Real choices look different from the tidy definitions. Here are eight cases, five drawn from the products above and three common scenarios.

1. YouTube Expressive Captions (December 2025). A viewer-side closed-caption layer that adds tone and sound cues to English videos without any work from the uploader. It shows what the closed-caption model can do when the platform owns the rendering.
2. Premiere 26.3 Single Word captions (June 2026). An open-caption style, because the one-word-at-a-time look is meant to be rendered into the export. Adobe built it for social-video styling.
3. Microdramas on YouTube Shorts (first half of 2026). YouTube reports more than 6.5 billion views for the format. Vertical and serialized, it is a natural home for burned-in text, with a sidecar track as backup.
4. Scenario: a 14-minute software tutorial. Dialogue-heavy, watched on desktop, searched by title. A corrected SRT uploaded to YouTube serves it best, since viewers can toggle it and the transcript stays editable.
5. Scenario: a gaming highlight reel. Loud game audio, fast cuts, a facecam. Burn in short open captions for the commentary and keep them clear of the facecam and the HUD. Add a sidecar track on the long-form upload.
6. Rev burned-in captions (November 2026). Rev’s current pages describe burned-in open captions and foreign-language subtitles next to its caption files. A creator who wants a human-checked track and a captioned cut can order both from one vendor and compare the proofs side by side.
7. Submagic AI Auto Captions (2026). The product page claims 99% accuracy across 123 languages and sells a Starter plan at $19 per month for 15 videos. It is the open-caption path in its cheapest form, because the styled text lands in the export.
8. Scenario: a multilingual product demo. Ship one clean master and one SRT per language. Burning a different language into every export would multiply your render time and your storage.

Cases 1 and 4 show closed captions at work, cases 2, 3 and 7 show burned-in text, and cases 5 and 6 sit in the middle, where a team pays for human review of a track or an export. None of the eight needed a single caption type for every platform.

Notice what repeats. Long videos with a searchable life lean on tracks. Short videos on muted feeds lean on pixels. Translation always favors tracks, because one master plus several small text files beats several full video exports.

Use-Case Recommendations

Match the type to where the video will live and who will watch it.

1. YouTube long-form tutorials and interviews: Upload a corrected closed-caption track. Skip burned-in text unless the footage is noisy. Review the auto-generated version line by line before publishing.
2. Shorts, Reels and TikTok: Use open captions positioned inside the platform’s visible area. Preview in the app, because interface layers cover the bottom and right edges.
3. Livestream VODs and gaming: Publish the clean VOD with a sidecar track. Cut highlights with short burned-in captions for social.
4. Corporate training and compliance: Deliver closed captions in a standard format such as SRT or DFXP, and check the client’s spec for sound-cue requirements. If the work is for covered US broadcast programming, check 47 CFR Part 79 with your distributor.
5. Multilingual channels: Keep an English master and one subtitle track per language. Pay a human for the languages that carry revenue, and use AI for the rest.
6. Podcast video and talking-head clips: Use single-word or two-line open captions on the social cut, and a full transcript-based track on the long version.
7. Film festival and broadcast delivery: Follow the delivery spec exactly. Expect SCC, STL or another broadcast format from the Adobe list rather than a plain SRT.

If you run more than one of these workflows, the split is easy to automate. Export the clean master first, publish the SRT with it, and treat every burned-in cut as a derivative that you can re-render from the project at any time. That order protects you when a platform changes its caption rules, because the text and the picture never get locked together in your only copy.

Budget also shapes the choice. A solo channel posting weekly can run AI captions on a Submagic or Captions plan and spend the saved money on a human pass for the three videos a year that matter most. A small studio with clients should price a human pass into each quote at $1.99 per minute and up, so the client sees accuracy as a line item rather than a surprise.

Frame by Frame: Building Open Captions That Clear the Lower Third

This build assumes a 1080×1920 vertical cut in Premiere with a speech-to-text pass already run. The same logic works in any editor. The point is to treat captions like any other graphic, with an in, a hold and an out.

1. Set the sequence. Create a 1080×1920 sequence at your source frame rate. Do not convert frame rate here.
2. Generate the transcript. Run speech-to-text, then read it against the audio. Fix names and product terms before you style anything.
3. Choose the default font. In Premiere 26.5, set default fonts for text, subtitles and closed captions under Preferences, then Graphics. Pick a heavy sans-serif with strong contrast.
4. Pick the caption layout. Use two short lines for dialogue, or Single Word captions for fast clips. Break lines at natural phrase boundaries.
5. Set the in point. Start each caption on the first frame of its first spoken word. Late captions feel wrong faster than early ones.
6. Set the hold. Keep each caption on screen through its last word plus a short tail. If a line is too fast to read, split the speech or trim the wording.
7. Set the out point. Clear the caption before a hard cut so text never carries across two shots.
8. Check lower thirds. Wherever a name strap or logo bug appears, move the captions up or hold them until it clears. The lower-thirds and titles hub collects the sizing and timing guides.
9. Preview on a phone. Export a test and play it inside the target app. Look at the bottom third and the right edge for collisions with buttons.
10. Export twice. Make the clean master, then the captioned cut. Export the SRT from the same project.

Render settings

Migration Guide: Moving Between Caption Types

Most creators start with auto captions on one platform and later need something else, and the closed captions vs open captions split is usually what forces the move. These four moves cover the common cases.

From open captions to a sidecar track

You need the text back. Export the transcript or the caption track from your original project, not from the burned-in video. If the project is gone, re-run speech-to-text on the clean audio and correct it. Save it as SRT, upload it as a closed-caption track, and keep the burned-in version only for feeds that need it.

From YouTube auto captions to a corrected SRT

Download the auto-generated track from Studio, edit it in a text editor or your NLE, and upload the corrected file as a replacement. Given the Community reports of upload problems with secondary-language files, keep the SRT in your project folder before you upload.

From a sidecar file to open captions

Import the SRT into your editor, style it, and render. Premiere imports SRT directly, and the caption track is editable on the timeline until you export. Check every cue against lower thirds before you commit.

From platform captions to your own files

Some apps generate captions you can edit only inside the app. If you plan to reuse the text, copy it out into a plain-text file before the platform changes its interface. YouTube’s reports of upload trouble in the new Studio layout show how a workflow can change under you. A local SRT you control costs nothing and survives any layout change.

From one language to many

Translate the master SRT after you lock the picture, since every cut change moves the timecodes. Then review the result with a native speaker for names and idiom. Human translation runs from $6.49 to $15.99 per minute at Rev, so budget it for your highest-traffic videos first. For burned-in versions, plan one export per language and name files by language code.

Pros and Cons of Each Caption Type

The pros and cons of subtitles vs closed captions vs open captions come down to control, cost and time.

Subtitles

• Pros: cheap to add per language, easy to swap, works well when viewers hear the audio.
• Cons: omit sound cues, rely on platform support, give the viewer no accessibility for sound.

Closed captions

• Pros: viewer control, sound information, editable after publishing, searchable text.
• Cons: platform sets the look, bugs can block uploads, not every app accepts a sidecar file.

Open captions

• Pros: always visible, full style control, survives re-uploads and screen recordings.
• Cons: cannot be turned off, hard to fix, one export per language, collides with lower thirds and app UI.

The cons list carries one more cost that people forget, which is time. A burned-in cut adds a render pass to every video, and a second language adds another. A sidecar file adds a few minutes of review. Across a year of weekly uploads, that difference adds up to dozens of render hours.

Verdict: Closed Captions as the Default, Open Captions as an Add-On

In the subtitles vs closed captions decision, and again in closed captions vs open captions, closed captions win for most creators in November 2026 on cost and flexibility. One corrected SRT serves every long video, converts to other formats, and costs nothing to fix. The numbers back that up: Submagic Pro is $39 per month billed monthly for 40 videos, and a human pass is $1.99 per minute, so a clean track is cheap compared with re-rendering.

Open captions win on short vertical video, where muted autoplay is the default and the viewer never opens a menu. Premiere’s June 2026 Single Word captions and the caption-first design of Submagic and Captions show where the style is heading. Pair them with a sidecar file whenever you can, so accessibility and search do not depend on pixels.

Subtitles win only on translation for viewers who can hear. If you publish in several languages, build one master and several small text files.

One caution applies to every route. Auto-generated text is a draft, and the vendors say as much by selling human review next to it. Rev’s human tier at 99%+ accuracy and its AI tier at 96%+ show the gap that the vendors themselves price in. For a channel where names and product terms carry the story, that gap is worth a proofreading pass.

The takeaway: write the SRT first, burn in captions second, and let the platform’s toggle handle everyone who needs it.

Before you render

What is the difference between subtitles and closed captions?

Subtitles carry the dialogue for viewers who can hear the audio, often in a translated language. Closed captions carry the dialogue plus meaningful sound cues for viewers who cannot rely on audio. Both are text tracks the viewer can switch off. YouTube Studio uses the label Subtitles for the page where you add either one.

What are open captions?

Open captions are rendered into the video at export, so every viewer sees them and nobody can turn them off. You control the font, colour and position. The cost is that any fix needs a re-render. The US federal rule says open captioning or subtitles in the language of the target audience may be used in lieu of closed captioning for covered programming.

Does the FCC require captions on my YouTube videos?

The internet captioning rule in 47 CFR 79.4 covers programming provided by, or comparable to, a TV broadcast station and excludes consumer-generated media. A typical creator video falls outside it. Clients, platforms and other laws can still require captions, so check each delivery spec.

Which file format should I use for captions?

SRT is the safest default because platforms and editors accept it widely. WebVTT suits web players that need positioning. Broadcast work may call for SCC, DFXP, STL or another format on Adobe’s supported list. Keep one clean SRT as your master.

How much do captions cost in 2026?

Rev lists human captions from $1.99 per minute and translated subtitles from $6.49 to $15.99 per minute. Submagic plans run $19, $39 and $69 per month, or $12, $23 and $41 billed annually. Captions’ Basic plan is $9.99 per month, and Premiere’s single-app plan is reported at $22.99 per month.

Should I use open or closed captions on Shorts, Reels and TikTok?

Use open captions for muted autoplay feeds, placed inside the visible area and away from interface buttons. Preview in each app before posting. If the platform accepts a caption track as well, add one for accessibility.

Can I trust auto-generated captions?

Treat them as a first draft. Rev claims 96%+ accuracy for its AI transcription and 99%+ for human captions, and Submagic claims 99% for its AI captions, but these are vendor claims without a published test set. Names, jargon and accents cause most errors, so read every line against the audio before you publish.