Broadcast engineering

Everything you ever wanted to know about captions, subtitles and DVB

A visual guide to caption formats, regional standards and channel settings across SRT, MPEG-TS, HLS, DASH, RTMP and social platforms.

22 min read Updated 19 November 2026

Start with the viewer and the delivery path

Your contribution monitor shows captions. The television receiver displays a different language, and the mobile player has no subtitle menu. A feed can carry valid captions while a downstream conversion, manifest or player makes them unavailable.

This guide follows captions and subtitles from authoring to the screen. It covers the major broadcast and streaming families, their regional variations and the settings engineers need to specify at each boundary. Older proprietary formats and individual national receiver profiles still require their own delivery specifications.

Technical review: 19 November 2026. Compatibility below means a documented carriage method exists, subject to the stated conditions. It does not certify a particular encoder, Evrideo product, social account or receiver. The diagrams are illustrative engineering models, not screenshots of Evrideo settings.

1. Meaning

Language, dialogue, speaker identity and meaningful sounds. What must the audience understand?

2. Representation

Text, caption commands or bitmap graphics. Which decoder can interpret the information?

3. Delivery

Embedded data, a transport-stream PID or a separate subtitle track. How does it reach the device?

4. Presentation

Track selection, timing, fonts and placement. What does the viewer actually see?

Follow all four layers. A transport connection can be healthy while caption presentation fails.

Caption compatibility at a glance

Find your caption format, check the delivery column, then confirm the destination profile. Countries show common use, not exclusive support or legal requirements.

Documented carriageDocumented carriageProfile-dependent carriageProfile-dependent carriageConvert or use a custom pathConvert or use a custom path

On small screens, scroll the table sideways. Format names stay visible.

Caption format, delivery and regional usage comparison
Caption / subtitle formatCommon countries / marketsMPEG-TSSRT carrying MPEG-TSHLSDASHRTMP / RTMPS
CEA / CTA-608USA, CanadaDocumented carriageDocumented carriageDocumented carriageDocumented carriageProfile-dependent carriage
CEA / CTA-708 (native)USA, CanadaDocumented carriageDocumented carriageDocumented carriageProfile-dependent carriageProfile-dependent carriage
DVB bitmapUK, Europe, AustraliaDocumented carriageDocumented carriageConvert or use a custom pathConvert or use a custom pathConvert or use a custom path
Teletext subtitlesEurope, Australia (legacy)Documented carriageDocumented carriageConvert or use a custom pathConvert or use a custom pathConvert or use a custom path
DVB TTMLDVB markets (receiver-specific)Documented carriageDocumented carriageProfile-dependent carriageProfile-dependent carriageConvert or use a custom path
ARIB STD-B24Japan; national ISDB variantsDocumented carriageDocumented carriageConvert or use a custom pathConvert or use a custom pathConvert or use a custom path
WebVTTWorldwideConvert or use a custom pathConvert or use a custom pathDocumented carriageProfile-dependent carriageConvert or use a custom path
IMSC timed textGlobal OTT; ATSC 3.0 marketsProfile-dependent carriageProfile-dependent carriageProfile-dependent carriageDocumented carriageConvert or use a custom path
SubRip (.srt file)WorldwideConvert or use a custom pathConvert or use a custom pathConvert or use a custom pathConvert or use a custom pathConvert or use a custom path
Burned-in textWorldwideDocumented carriageDocumented carriageDocumented carriageDocumented carriageDocumented carriage

A tick identifies a documented delivery route, not a guarantee that every encoder or player supports it. A cross means the source format needs conversion or a custom integration in the delivery profiles summarized here; it does not mean the transport can never carry those bytes.

Profile-dependent carriage
  • SRT is payload-agnostic. Here it carries MPEG-TS unchanged, so those two columns match. SRT does not convert captions.
  • Starred ticks need the exact encoder, codec, packaging and player profile checked. HLS uses IMSC text in fragmented MP4; DASH WebVTT uses a supported MP4 mapping. DASH native 708 needs a separate interoperability check: the cited DASH-IF caption profile covers 608 compatibility data in a 708 wrapper. RTMP caption support depends on the receiving platform.
  • DVB TTML uses EN 303 560 carriage. Moving compatible IMSC text between MPEG-TS and OTT requires profile validation and repackaging; conversion is needed if the source profile does not fit. An arbitrary TTML, EBU-TT or SMPTE-TT file is not automatically a valid stream.
  • Brazil and other ISDB countries use national profiles. The Japan example does not certify those receivers. SDH, translation and forced narrative describe a track’s purpose, so their delivery follows its format.
  • Burned-in text is part of the video picture on every path. It cannot be switched off or selected as a separate language track. A SubRip file must first be converted, embedded or rendered.

Technical basis: Apple HLS · DASH-IF · DVB · DVB TTML · Teletext · EBU · ATSC · ARIB · SRT · YouTube Live.

Captions, subtitles, SDH and forced narrative

In North American usage, captions normally include dialogue plus information such as speaker identification, music and significant sounds. Translation subtitles often assume the audience can hear the soundtrack. In other markets, “subtitles” also covers accessibility text. Record the actual language and purpose rather than relying on the label alone. W3C explains these accessibility distinctions.

Translation subtitles

Illustrative programme frame
The train leaves in 5 minutes.

Dialogue translated for the audience.

Accessible captions / SDH

Illustrative programme frame
[STATION ANNOUNCEMENT]
The train leaves in 5 minutes.
[WHISTLE BLOWS]

Dialogue, speaker context and important sounds.

Same scene, different editorial purpose. SDH means subtitles for deaf and hard-of-hearing audiences. The examples are invented, not captured programme content.
  • Closed: a selectable caption/subtitle service. The viewer can usually switch it off.
  • Open or burned-in: text rendered into picture pixels. It survives delivery as video, but cannot be turned off or independently restyled.
  • Forced narrative: selected translations of foreign dialogue, signs or plot-essential text, associated with an audio-language experience. It is not a replacement for a full accessibility track.
  • Same-language subtitles: dialogue text for comprehension or language learning; check whether sound descriptions are included.
  • Live, prepared and near-live: different authoring workflows, each needing synchronization and quality checks.
  • Pop-on, roll-up and paint-on: presentation behaviours, especially familiar in 608 workflows. They describe how captions appear, not their language or network protocol.

Audio description and signing are separate access services. They complement captions and need their own production and delivery plans.

The format families engineers encounter

FamilyWhat it carriesWhere it fits
CEA / CTA-608Legacy caption text and display commands; CC1–CC4 services with a constrained character repertoire.Originated in NTSC line 21. Digital systems can preserve 608-compatible data; a digital feed need not contain analogue video.
CEA / CTA-708Digital caption service blocks, richer presentation and multiple services.ATSC 1.0 and embedded digital workflows. Native 708 and 608 compatibility data inside a digital carriage are different payloads.
DVB subtitlesUsually compressed bitmap objects assembled into display regions.EN 300 743 broadcast services, carried in MPEG-TS. The receiver overlays the image when selected.
Teletext subtitlesCharacter cells, page selection and restricted styling.Legacy services, including teletext transported in DVB MPEG-TS under EN 300 472.
TTML / IMSCXML timed text with defined timing, styling and layout profiles.Professional interchange and OTT. IMSC 1.3 is text-only; earlier IMSC versions also specify image profiles.
DVB TTMLTTML documents with DVB-specific transport and timing rules.EN 303 560. Distinct from the bitmap DVB subtitle family. Receiver support must match the service profile.
WebVTTUTF-8 cues with start/end times and optional positioning, regions and voice information.Browser and HLS subtitle workflows; DASH can use supported WebVTT mappings.
EBU-TT / EBU-TT Live / EBU-TT-DRelated TTML profiles for authoring, live contribution and distribution.Specify the exact profile. A live contribution document is not automatically a valid delivery segment.
SMPTE-TTSMPTE timed text and conversion guidance for broadcast caption data.Professional interchange. Select the target delivery profile explicitly rather than assuming every TTML file is interchangeable.
ARIB STD-B24Japanese broadcast caption coding, including character and presentation mechanisms.ISDB workflows. Preserve language-specific features and check the target national profile.

CEA and EIA names persist in equipment menus even where CTA is the current standards-body name. Treat them as a clue to the required specification and revision, not proof of full compatibility. ATSC’s legacy carriage guidance also describes Caption Distribution Packets (CDPs) used in professional interfaces: 608 compatibility and native 708 payloads must be checked independently.

Primary references: EBU broadcast recommendations, W3C IMSC, WebVTT, EBU-TT-D, EBU-TT Live, SMPTE timed text and ARIB STD-B24.

Files arriving from the content supplier

SubRip .srt files usually provide basic timed text. They are unrelated to the SRT network transport. SCC stores 608 caption commands; EBU STL is a legacy broadcast subtitle exchange format. TTML, WebVTT, ASS/SSA, DVD VobSub and Blu-ray PGS may also appear in libraries. ASS/SSA can contain elaborate styling; VobSub and PGS are image-based formats.

A filename extension does not establish the language, accessibility role, source edit, frame rate or delivery suitability. Preserve the original and create a checked delivery derivative. The Library of Congress describes SubRip’s basic structure and implementation-dependent formatting.

A global standards map

North America

Expect 608 compatibility and 708 in established ATSC 1.0 workflows. ATSC 3.0 uses the IMSC-based A/343 caption system. Confirm the receiver generation.

Europe and DVB markets

DVB bitmap and teletext services coexist with TTML-based delivery. OTT and hybrid TV profiles add their own requirements; DVB-T/T2, DVB-S/S2 and DVB-C describe delivery systems, not subtitle languages.

Japan and ISDB markets

ARIB STD-B24 is central to Japanese broadcast caption workflows. Character repertoire, downloaded glyphs, ruby annotations and writing direction deserve explicit conversion tests.

Brazil and Latin America

ISDB-T-derived and other broadcast systems coexist across the region. Brazil has local ABNT profiles and a TV 3.0 caption standard; do not apply a Japanese receiver profile or one continent-wide rule automatically.

Asia, Middle East and Africa

Broadcast systems and language needs vary by country and operator. Specify the receiver profile, script and delivery contract. A label such as DTMB, DVB or ISDB alone cannot define the subtitle acceptance test.

Australia, New Zealand and global OTT

Check each broadcaster or platform specification, including legacy teletext/DVB paths. Internet delivery commonly uses WebVTT or IMSC, but a television app can support a narrower profile than a desktop browser.

This is a standards-family map, not a map of legal obligations or a claim that every operator in a region uses the same format.

ATSC A/343 specifies the ATSC 3.0 caption system. ARIB documents the Japanese family, while SBTVD’s TV 3.0 guidance covers Brazil’s newer profile. EBU R 110 describes teletext, DVB bitmap and DVB TTML approaches.

For global services, test Arabic and Hebrew bidirectional text, CJK line breaking, ruby and vertical writing where required, accented Latin characters and font availability. Keep language, script, regional variant and accessibility role as separate metadata. Regulatory obligations vary by service and jurisdiction; obtain a current compliance review for each launch market.

Which settings belong to the channel or output?

A channel establishes editorial policy. Each distribution output maps that policy into the destination’s technical contract. The player then applies viewer preferences within its capabilities. Use this ownership model to avoid changing a global setting to fix one endpoint.

Channel / programme

One approved language and accessibility policy

Programme edit ID · source audio language · available text languages · SDH / translation / forced roles · live caption source · QC status · fallback owner

Broadcast output

Target standard, video codec, service numbers, subtitle PID, DVB page IDs, teletext page, language descriptors and timing.

OTT output

HLS / DASH packaging, subtitle profile, track groups, language and role signalling, segment timing, CDN access and player targets.

Social output

Platform ingest protocol, accepted caption carriage, language limit, account/event settings and live/replay validation.

Viewer / receiver

Select, synchronize and render

Preferred language · captions on/off · supported styling · safe placement · font rendering

The arrows show responsibility moving through the chain. Per-output settings can differ while the programme’s approved language and meaning remain consistent.

Store enough metadata to automate safely

For each text asset, record the matching programme version, language, role, source format/profile, timing basis, frame rate where relevant, offset, duration, rights, revision and QC result. For live captions, add source ownership, expected delay, health monitoring and recovery behaviour.

Each output needs an explicit mapping, for example English SDH → 708 service 1 on one destination, an English WebVTT rendition on another. Do not treat a 708 service number, a teletext page number and a DVB composition page ID as interchangeable language identifiers. Track selection must remain deterministic when a programme has two English services with different purposes.

Compare SRT, MPEG-TS, HLS, DASH and RTMP

Read this as a carriage matrix. “Possible” requires compatible encoders, muxers, packagers and players. Preserving bytes, converting a caption format and presenting it to the viewer are three separate capabilities.

Layer / outputCaption and subtitle pathWhat to configure and prove
SRT transportPayload-agnostic. Commonly carries MPEG-TS containing embedded 608/708 or separate DVB / teletext streams. SRT itself performs no caption conversion.Choose the carried container and retain its required streams. Test the receiving gateway; good packet delivery does not prove caption decode.
MPEG-TS containerVideo can contain embedded caption data; separate elementary streams can carry DVB bitmap, teletext or profile-specific timed text.PMT signalling, PID selection, language/type descriptors, page IDs and timestamps. Demux/remux filters must retain auxiliary streams.
HLS deliveryEmbedded 608/708 in supported video carriage; separate WebVTT or IMSC text subtitles under Apple’s profile. Raw DVB bitmap PES is not a WebVTT rendition.EXT-X-MEDIA entries, matching groups on variants, language/accessibility/forced flags and cue-to-media timing. Test the actual OS and player.
MPEG-DASH deliverySeparate timed-text AdaptationSets, commonly IMSC in MP4; WebVTT mappings and embedded 608/708 are possible where the profile/player supports them.MPD language, Role and Accessibility descriptors, codec/profile, segment timeline and Period transitions. A decoder is still required.
RTMP / RTMPS ingestCaption-bearing video or a destination-supported metadata convention such as onCaptionInfo. Generic subtitle files are not automatically converted into these messages.Encoder and receiving-platform agreement. RTMPS adds TLS protection, not additional caption formats or languages.
Burned-in videoCaption pixels follow any compatible video path, including the transports above.Create the appropriate language rendition and verify crop/safe areas. The viewer loses independent selection and styling.

Basis: the SRT implementation’s transport scope, DVB carriage, Apple’s HLS profile, DASH-IF text guidance and Twitch’s documented RTMP conventions. These sources describe different layers, so a universal yes/no “caption support” tick would be misleading.

Inside a DVB transport stream

Illustrative MPEG-TS service

PMT: the receiver’s component directory

Lists component PIDs and the descriptors needed to identify their purpose.

Video PID 256

Compressed picture. Embedded caption data, when used, belongs to the relevant video carriage.

Audio PID 257

Programme audio. Language and accessibility metadata must describe the actual audio service.

Subtitle PID 300

DVB subtitle PES packets. Example: English accessibility subtitles; composition page 1, ancillary page 1.

Subtitle PID 301

A second service. Example: French translation with its own descriptor/page mapping.

PID numbers and page IDs are examples only. One subtitle PID can also contain multiple services; the descriptors and page IDs identify the correct one. A PID is a packet identifier, not a language code.

EN 300 743 organizes subtitle displays using pages, regions, objects and colour information. A renderer builds the overlay and presents it at the signalled time. Receiver setup needs the subtitle component, ISO language code, subtitling type and composition/ancillary page mapping. See the DVB subtitle specification.

Bitmap DVB subtitles can represent many scripts because the receiver displays graphics. Converting those graphics to editable text requires a text source or OCR with review. A remux can preserve the subtitle PID, but cannot turn its images into WebVTT merely by changing the output URL.

Teletext adds page and magazine selection; DVB TTML has its own document and timing rules. EN 303 560 defines that TTML path. Keep the three families distinct in procurement and testing.

A conversion plan from one channel to many screens

  1. 1. Acquire and identifyMatch the caption source to the exact programme edit. Read language, accessibility role, timestamps and source profile.
  2. 2. Normalize and reviewBuild a checked timed-text master where practical. Preserve speaker cues, sound descriptions, placement intent and meaningful styling.
  3. 3. Render or encode per outputProduce the agreed 608/708, DVB bitmap, WebVTT or IMSC derivative. Keep output-specific mappings explicit.
  4. 4. Package and signalEmbed, multiplex or segment the derivative. Publish the language/role metadata that lets the receiver discover it.
  5. 5. Decode at the destinationSelect the service on a real receiver/player. Compare words, timing and placement against the approved source.
Conversion is an editorial and technical process. A text master simplifies derivatives, but cannot recover information already lost in a legacy or image-only source.

Map supported features deliberately. A rich TTML document can lose positioning or styling when reduced to a simpler target. A 608 export may need character substitutions and line reflow. Preserve the source and document each accepted loss instead of overwriting the master.

Live captioning adds authoring delay before transport and player buffering. Measure cue timing against decoded audio/video at the destination. A faster video path can expose late captions even when the caption source has not changed. ASR and translation need language-specific review, especially for names, numbers, overlapping voices and meaningful non-speech sounds.

Example: one programme, three output contracts

A hypothetical broadcaster has an English SDH master and a French translation. Its broadcast affiliate requests DVB bitmap subtitles; its app requests HLS WebVTT; its YouTube event accepts one embedded caption track. Create and test those derivatives separately. Keep the French app track even if the social event cannot accept it through the chosen ingest path.

HLS and DASH: make the track discoverable

For HLS, the multivariant playlist identifies subtitle and closed-caption groups. Each relevant video variant must reference the right group. Language and role metadata drive menus and automatic selection; they do not create missing cue data. Apple’s accessibility profile includes specific characteristics for SDH and requires compatible subtitle packaging. Apple’s requirements and RFC 8216 define the details.

The following is an illustrative playlist fragment, not a complete playable manifest. Both renditions belong to the same subtitle group; each relevant video variant needs SUBTITLES="subs". Actual playlists, timing and segments must exist.

#EXT-X-MEDIA:TYPE=SUBTITLES,GROUP-ID="subs",NAME="English SDH",LANGUAGE="en",AUTOSELECT=YES,DEFAULT=NO,FORCED=NO,CHARACTERISTICS="public.accessibility.transcribes-spoken-dialog,public.accessibility.describes-music-and-sound",URI="en-sdh/index.m3u8"
#EXT-X-MEDIA:TYPE=SUBTITLES,GROUP-ID="subs",NAME="English forced",LANGUAGE="en",AUTOSELECT=YES,DEFAULT=NO,FORCED=YES,URI="en-forced/index.m3u8"

For embedded captions, use the corresponding TYPE=CLOSED-CAPTIONS and service mapping rather than a subtitle URI. WebVTT segment timing must relate to the video timeline; X-TIMESTAMP-MAP provides that relationship in the applicable HLS carriage.

DASH signals text through the MPD. Check the selected AdaptationSet, lang, profile/codec and role/accessibility descriptors. Embedded 608/708 caption service signalling is associated with the video AdaptationSet. A player that lacks the decoder can ignore a perfectly valid advertised service. DASH-IF describes these selection rules.

  1. Player → manifestDiscover available languages, roles and supported caption formats.
  2. Player → media endpointRequest the chosen text track and video, or receive captions inside the video samples.
  3. Player → timelineMap cue timestamps onto media playback, including discontinuities and ad boundaries.
  4. Player → screenRender the selected service using supported presentation rules and viewer preferences.
Simplified playback sequence. A separate subtitle track needs its own successful requests; caption-bearing video needs extraction and decoding instead.

Social platforms: verify ingest and playback separately

A platform’s viewer delivery format does not define what its ingest endpoint accepts. Sending a stream to RTMP, or uploading a caption file for VOD, does not establish equivalent live support. Rehearse the exact destination and check the recording separately.

DestinationDocumented routeAcceptance check
YouTube LiveEmbedded 608/708 or a supported HTTP caption-post workflow. For HLS ingestion, Google documents embedded captions with H.264, not HEVC; separate caption posts are another path.Its embedded-caption help currently specifies one caption track. Enable the event’s caption mode, test the intended language and check latency mode restrictions.
TwitchEmbedded caption-bearing video or RTMP onCaptionInfo, with the service selection described in its guide.Test CC1 / NTSC field 1 compatibility and the viewer CC control. A title/tag saying “captions” is not proof of a delivered track.
Facebook LiveMeta documents publisher-supplied live captions. Its public launch guidance is older than this guide.Confirm the current encoder, account and Live Producer path before committing a profile or language count. Do not infer a DVB or WebVTT ingest capability from the player.
LinkedIn LiveLinkedIn’s producer guide describes 608/708 passthrough with supported broadcast tools.The guide is older; validate the current tool/account combination and replay. Uploaded-video caption features do not establish arbitrary live-ingest support.
TikTok LIVEExternal live video: TikTok documents an RTMP server URL and stream key for accounts with access to its external streaming workflow. Confirm that the destination account exposes this option.Selectable captions: the cited ingest instructions do not establish 608/708, onCaptionInfo, DVB, WebVTT or IMSC passthrough. Verify a caption-enabled rehearsal on the actual account and viewer app; use a dedicated burned-in rendition when selectable delivery cannot be verified.
Instagram Live ProducerExternal live video: eligible professional accounts use a stream URL and key from Live Producer. Match the RTMP/RTMPS scheme supplied by the platform and check the incoming preview before starting the broadcast.Selectable captions: external video ingest alone does not establish broadcast-caption passthrough. Verify the current account, viewer app and any saved replay. Treat automatic captions as a separate feature; use burned-in text when a selectable caption path cannot be verified.
Other social / short-form destinationsAutomatic captions, uploaded sidecars, embedded captions and burned-in text are separate product features.Verify live versus upload support, language, device and account. Where only burned-in text is practical, produce an intentionally separate rendition.

Sources: YouTube live captions, YouTube HLS ingest, YouTube latency settings, Twitch, Meta, LinkedIn, TikTok’s external-streaming instructions and Switchboard’s Instagram integration instructions. The TikTok example is regional guidance; check access on your account. The Instagram source documents an integration provider’s workflow. Neither source verifies selectable broadcast-caption carriage. Platform limits can change; repeat the rehearsal after changing codecs, encoders or event settings.

A caption-safe feed for TikTok and Instagram

  1. Keep the channel masterRetain the clean programme feed and original caption services for broadcast and OTT outputs. Make social-specific changes on a separate distribution output.
  2. Prepare the social renditionChoose the destination layout and caption language. If selectable captions are unverified, render the approved text into the picture before encoding. Check portrait crops, interface overlays, line length and timing on a phone.
  3. Connect the live destinationUse the account’s current server URL and stream key. Confirm key validity and any manual preview/start steps. Mobile LIVE access alone is insufficient evidence that an external encoder can connect.
  4. Watch as the audienceCheck the actual live stream, including readability and any available caption controls. Verify the saved replay separately. Burned-in text cannot be switched off, restyled or selected as another language.

Acceptance rule: video arrival, caption data carriage and readable captions at the viewer are separate checks. Record each result. An unverified caption path should remain unverified in the output specification, rather than receiving a compatibility tick.

The failures that survive a green signal monitor

SymptomInvestigateTest that exposes it
Video works; captions disappearA remux dropped the subtitle PID, a transcode stripped caption data, or the caption mapping points to the wrong service.Compare input and output components, then decode the selected service at the destination.
Menu exists; no words appearManifest points to missing segments, CORS/auth prevents access, or advertised captions are absent from the payload.Fetch the advertised resources and inspect cue data while dialogue is present.
Wrong language or duplicate English entriesLanguage metadata, service mapping, SDH/translation role or duplicate source injection.Select every track by name and compare its content and purpose.
Text drifts after an edit or advertWrong programme version, timecode/frame-rate interpretation, timestamp reset or splice mapping.Play through the edit/ad boundary and seek across it on replay.
Text survives a cut for too longA clear/end event was lost, a bitmap display remained active, or cue lifetime crosses the boundary.Test abrupt source changes, silence, fallback slates and return to programme.
Boxes, clipped lines or wrong word orderFont coverage, bidi shaping, unsupported glyphs, line wrapping or safe-area conversion.Use representative scripts and the smallest supported screen, with graphics and lower thirds visible.
Captions fail only on backupDifferent page IDs, service numbers, clocks, metadata or caption source availability.Force a controlled failover and verify the same viewer selection still works.

SSAI deserves its own test case: both programme and ad caption timelines must remain meaningful after stitching. Check silence, gaps, cue clearing and the return to programme. For encrypted delivery, inspect manifests and request health, then validate rendered captions with an authorized playback probe; an HTTP 200 alone cannot establish accessibility.

A practical release checklist

  1. List target markets, receiver generations, apps and social destinations. Obtain their caption delivery specifications.
  2. Identify each programme edit, text language and accessibility role. Separate full SDH from translation and forced narrative.
  3. Record input formats and each conversion or accepted loss. Keep the source asset and QC evidence.
  4. Map per-output PIDs, page IDs, caption services, manifest groups and language/role metadata.
  5. Agree measurable timing, completeness, readability and placement criteria with the receiving platform and editorial team.
  6. Test live entry, language switching, every bitrate rendition, adverts, replay seeking, backup changes and long-running timing behaviour.
  7. Check the actual television, browser and mobile player. Include viewers’ caption settings and relevant language scripts.
  8. Assign alerts, a fallback and an owner when a caption source is unavailable. Retest when an encoder, packager, player or platform changes.

For the operational handoff, read our caption and audio-description checklist. For distribution planning, explore Outpost and discuss your required output profiles with Evrideo. Bring a representative captioned sample and the destination specification so the complete path can be assessed.

References

  1. W3C: media accessibility requirements
  2. ATSC A/79: caption carriage and redistribution
  3. Library of Congress: SubRip subtitle format
  4. ETSI EN 300 743: DVB subtitling
  5. ETSI EN 303 560: DVB TTML
  6. EBU R 110: digital TV and online subtitling
  7. W3C: IMSC Text Profile 1.3
  8. W3C: WebVTT
  9. EBU Tech 3370: live subtitling
  10. EBU Tech 3380: EBU-TT-D
  11. SMPTE: timed text document family
  12. ATSC A/343: captions and subtitles (April 2026)
  13. ARIB STD-B24: Japanese digital broadcast coding
  14. SBTVD: TV 3.0 captioning operational guidance
  15. SRT project: payload-agnostic transport
  16. Apple: HLS authoring specification
  17. RFC 8216: HLS rendition and timing signalling
  18. DASH-IF IOP Part 9: text and caption tracks
  19. YouTube: live caption requirements
  20. YouTube: HLS ingestion
  21. YouTube: broadcast latency and caption settings
  22. Twitch: closed caption delivery
  23. Meta: publisher captions for Facebook Live
  24. LinkedIn: broadcast producer guide
  25. TikTok: external live streaming with OBS
  26. Switchboard: Instagram Live Producer integration
Back to Guides