Add your SMI file
Drop one or more .smi files onto the converter above, or browse for them. They load into browser memory only — nothing is uploaded, so there is no size-based pricing and no server queue.
Converting a .smi file to .ttml extracts the document's content from its office package into a portable form. Free, private, and validated — the file never leaves your browser.
Batch files can each use a different output. Nothing uploads for local conversions.
Working inputs include camera RAW, browser-local audio/video, PDF, CBZ/CBR comics, office documents, ebooks, markup, 3D models, structured text, images, and archives.Converting a .smi file to .ttml extracts the document's content from its office package into a portable form. An .smi file is a SAMI caption document: HTML-like markup in which <SYNC Start=…> elements mark the millisecond at which a caption appears, and the caption runs until the next SYNC. Microsoft designed it as an accessibility format, so a single file can carry several languages selected by CSS class.
A .ttml file is a Timed Text Markup Language document: an XML vocabulary in which each <p> element carries begin and end (or dur) attributes plus the caption text. It is the W3C standard that broadcast and streaming systems use as their interchange format for captions and subtitles. For this route the practical draw is a real published standard with formal profiles rather than a de facto convention and expresses styling, layout, and regions that plain caption formats cannot — balanced against far more verbose than SubRip or WebVTT for the same cues, which is worth knowing before you commit a large batch.
The practical trigger for this conversion is usually a mismatch: with .smi, never formally standardised, so real-world files vary in structure. Switching to .ttml buys you a real published standard with formal profiles rather than a de facto convention, which is why it is the better fit for delivering captions to broadcast and streaming platforms that mandate a TTML profile. Because the conversion runs locally, trying it costs nothing but a few seconds of compute on your own machine.
| Aspect | SAMI captions (.smi) | Timed Text Markup Language (.ttml) |
|---|---|---|
| Format type | Text-based — characters and structure, so there is no visual quality loss | Text-based — characters and structure, so there is no visual quality loss |
| How it stores data | the document has an HTML shape: <SAMI> wrapping <HEAD> with a <STYLE> block and <BODY> with the SYNC elements | the root <tt> element declares the TTML namespace and optional ttp: parameters such as frameRate, tickRate, and timeBase |
| Strongest at | captioning legacy Windows Media and Silverlight content | delivering captions to broadcast and streaming platforms that mandate a TTML profile |
| Weak spot | never formally standardised, so real-world files vary in structure | far more verbose than SubRip or WebVTT for the same cues |
| Metadata | style-class and title information describes presentation and language selection, not the cues; conversion keeps timing and text | styling, layout, and region definitions are read past rather than applied: timing and text convert, presentation does not, and no external resource is ever fetched |
Drop one or more .smi files onto the converter above, or browse for them. They load into browser memory only — nothing is uploaded, so there is no size-based pricing and no server queue.
Select .ttml in the output menu next to each file. The menu only offers targets this engine can genuinely produce, so if TTML is selectable, the route is real and validated.
Press Convert. The package is unzipped in memory and its XML content parsed locally; nothing is uploaded.
Each result is signature-checked before the download unlocks, so a failed encode can never masquerade as a valid TTML file. Outputs keep the original filename with the .ttml extension.
There is no visual quality to lose — smi is text-oriented and ttml is text-oriented, so the question is structural fidelity. Text, ordering, and basic structure are preserved; complex layout, embedded objects, and styling beyond the target's model are simplified.
Yes — the .smi file is processed inside your browser tab and never uploaded. The package is unzipped in memory and its XML content parsed locally; nothing is uploaded. Close the tab and the file is gone from memory.
Broadcast and streaming toolchains — including the DFXP, SMPTE-TT, EBU-TT, and IMSC profiles — read and write TTML, and caption editors such as Subtitle Edit export it. Browsers do not. Novus Convert parses TTML locally with document types and external entities refused outright, resolves all five time-expression shapes, and converts to any of the twelve timed-text targets.
It depends on the content: the root <tt> element declares the TTML namespace and optional ttp: parameters such as frameRate, tickRate, and timeBase. Convert one representative file first and compare before batch-processing a large set.
In .smi, style-class and title information describes presentation and language selection, not the cues; conversion keeps timing and text. Re-encoding through the browser pipeline does not carry embedded metadata into the output, which doubles as a privacy scrub — check the exported file if you specifically need tags preserved.
Yes — the reverse route exists as a separate tool. Bear in mind that round-tripping .smi → .ttml → .smi is not a perfect undo when any lossy step is involved; keep your original if fidelity matters.