One reasoning surface
The stable v1 model gives AI one predictable cue structure for SRT and WebVTT today and provides the extension boundary needed for future broadcast, binary, and bitmap formats.
The real reason Cueson exists
Make media content directly readable to AI.
A huge amount of what people say, read, watch, and archive is trapped behind subtitle formats that machines cannot treat as ordinary text. Some are simple text files. Some are dense XML. Some are old binary broadcast records. Some are literally pictures of words. Cueson is designed to turn that fractured universe into one predictable Cue JSON surface where the words, timing, speakers, provenance, layout, and other useful context are directly visible.
Current v1.0.0 release boundary: Cueson provides stable Cue JSON, SubRip, and WebVTT encode, restore, render, conversion, validation, inspection, and completion workflows in the published, independently verified v1.0.0 release. Its documentation and immutable released schemas are published at cueson.io. OCR and additional format families remain future work.
Media content
Text subtitles, styled captions, broadcast files, bitmap subtitles, and transcription-derived timed speech.
Cueson
Parse, preserve, and normalize supported text formats today; add derived plaintext through OCR for future bitmap formats while retaining the original source.
Direct AI reasoning
Search, summarize, compare, classify, extract facts, build graphs, and reason about what the media actually contains.
No information hostage
The plain text is easy to consume, while original bytes, names, timing, metadata, styling, graphics, and provenance remain available instead of being discarded.
No downstream format maze
Current SRT and WebVTT codecs already let downstream tools use Cue JSON instead of separate parsers, and future codecs can join that same contract without creating another downstream format maze.
Target format map
The subtitle universe Cueson is designed to open up
SRT and WebVTT are the stable format contracts in the published v1.0.0 release. The other families remain future directions and show why the common Cue JSON model matters long term, especially when the original content is binary or image-based rather than plain text.
v1
Future
Future + OCR
Text
Binary
Bitmap
| Format | Where it comes from | What it is used for | What is actually inside | What Cueson unlocks |
|---|---|---|---|---|
|
SRT
.srt
Textv1
|
Simple timed text Originally created around the SubRip subtitle-ripping tool, then adopted almost everywhere as a simple subtitle interchange format. |
Downloaded subtitles, media players, video editing, localization handoffs, archives, and subtitle sidecars. | Plain text arranged into numbered cues with start and end times. Very little standardized styling or metadata. | The words and timing become direct Cue JSON fields that an AI can read immediately. |
|
WebVTT
.vtt
Textv1
|
Web timed text Designed for timed text on the web and used with HTML video and audio. |
Web captions, subtitles, chapters, speaker labels, descriptions, cue positioning, and time-aligned metadata. | Timed text plus cue settings, speakers, regions, notes, styles, inline timing, and richer structure than SRT. | Cueson exposes the text, timing, speakers, tokens, and structure without forcing the AI to understand WebVTT syntax. |
|
TTML
.ttml / .xml
TextFuture
|
XML timed text A W3C timed-text family built for richer professional caption and subtitle interchange. |
Streaming platforms, localization, accessibility, broadcast-to-streaming workflows, and professional caption exchange. | XML describing timed text, layout, regions, styles, metadata, language, and presentation behavior. | Cueson can reduce a complex XML document to a predictable semantic cue model while retaining the original structure. |
|
IMSC
.ttml / .xml
TextFuture
|
Streaming timed text A constrained TTML profile created specifically for interoperable internet media subtitles and captions. |
OTT and streaming delivery where platforms need a predictable professional timed-text profile. | TTML-based text, timing, styling, regions, language, and presentation information under a tighter interoperability profile. | The same Cue JSON surface can expose what was said and when, while preserving IMSC-specific presentation details. |
|
SMPTE-TT
.xml
TextFuture
|
Broadcast timed text A professional timed-text profile from the Society of Motion Picture and Television Engineers. |
Broadcast caption interchange and movement between legacy television caption systems and newer timed-text workflows. | XML timed text plus broadcast-oriented metadata and mechanisms for preserving caption information that simpler formats cannot represent. | Cueson makes the human-language content accessible without flattening away the broadcast information that may still matter. |
|
ASS
.ass
TextFuture
|
Styled subtitle script Advanced SubStation Alpha, strongly associated with Aegisub and sophisticated fansubbing workflows. |
Anime, fansubs, karaoke, signs, elaborate styling, precise positioning, typography, and visual subtitle effects. | Script sections, styles, events, dialogue, positioning, override tags, karaoke timing, and drawing or effect instructions. | The dialogue becomes plain structured text for reasoning while the highly expressive visual script remains preserved. |
|
SSA
.ssa
TextFuture
|
Styled subtitle script SubStation Alpha, the predecessor to ASS. |
Legacy styled subtitles, fansubs, positioned dialogue, and older subtitle-authoring workflows. | Scripted subtitle events, styles, dialogue, timing, and positioning directives. | Older SSA collections become machine-readable through the same common cue model used for newer formats. |
|
EBU-STL
.stl
BinaryFuture
|
Legacy broadcast exchange A European Broadcasting Union subtitle exchange standard designed for professional television workflows. |
Broadcaster handoff, television subtitling, localization vendors, transmission workflows, and long-term media archives. | Binary records containing subtitle text, timing, character-set information, positioning, and broadcast metadata. | Cueson turns an opaque legacy binary file into directly inspectable text and timing while retaining the exact binary source. |
|
EBU-TT
.xml
TextFuture
|
Modern broadcast timed text A newer European Broadcasting Union timed-text family based on XML. |
Modern broadcast and online subtitle exchange, accessibility workflows, archives, and professional distribution. | Structured XML containing timed text, metadata, styling, language, and professional broadcast semantics. | Cueson gives AI the readable content immediately without making downstream systems carry an EBU-specific XML parser. |
|
PGS / SUP
.sup
BitmapFuture + OCR
|
Blu-ray bitmap subtitles Presentation Graphic Stream subtitles used heavily on Blu-ray and related disc media. |
Disc subtitles where the subtitle is authored as a rendered visual image rather than editable characters. | Timed bitmap graphics, palettes, positioning, and presentation instructions. The file may contain no text characters at all. | This is where Cueson becomes especially powerful: OCR can derive real text for AI reasoning while the original subtitle images remain authoritative. |
|
VobSub
.idx + .sub
BitmapFuture + OCR
|
DVD bitmap subtitles A common representation created when subtitles are extracted from DVDs. |
DVD subtitle extraction, archival sidecars, media-player playback, and migration from disc media. | A paired format: IDX carries timing, language, palette, and indexing information while SUB carries the actual subtitle images. | Cueson can keep the pair together, preserve the original imagery, and layer OCR text on top so AI can reason about what the subtitles actually say. |
|
SAMI
.smi / .sami
TextFuture
|
Legacy synchronized text Microsoft Synchronized Accessible Media Interchange, created for synchronized accessibility and multimedia captions. |
Older Windows Media collections, synchronized captions, subtitles, and audio-description workflows. | HTML-like timed text with synchronization markers, style classes, language information, and caption content. | Cueson can make older media collections readable through the same modern machine-facing representation. |
The preservation half
Cueson is not supposed to shred every format into generic text. It keeps a machine-friendly interpretation and a lossless source envelope together. That means an AI can see the text immediately while software can still recover or inspect the original file and its native details.
The OCR half
PGS and VobSub illustrate the bigger idea. A human sees subtitles on screen, but the file may contain only images. OCR converts those visual words into structured plaintext that AI can reason over, while Cueson keeps the original subtitle graphics intact as the authoritative evidence.
The payoff: the content of media stops being locked behind whatever file format happened to carry it. Once represented as Cue JSON, the actual language inside the media becomes a directly addressable data source for AI, search systems, analytical software, archives, and applications that have never heard of SRT, TTML, ASS, PGS, or VobSub.