Know what a subtitle conversion preserves
The supported subset and local file behaviour are explicit.
UTF-8, cue-only files
Import UTF-8 SRT, cue-only WebVTT or a CueHarbor JSON project, up to 1 MB. A project supports 2,000 cues and a timeline from zero to 100 hours. Individual cue text is limited to 4,000 characters and cannot contain blank lines, because blank lines separate cue blocks.
WebVTT NOTE, STYLE, REGION and header metadata blocks are rejected rather than silently discarded. Inline karaoke timestamps inside cue text are also rejected, because shifting only cue boundaries would leave them at their old times. This is a focused cue editor, not a full WebVTT document processor.
Conversion details
WebVTT cue identifiers and settings are stored and included in VTT output. SRT output omits those fields, normalises timestamps and regenerates numeric cue indices. The WebVTT header is normalised to WEBVTT. Newline conventions are normalised. Stored body markup is preserved as text in the output, while the workspace preview never executes or styles it.
Keep the editable project
JSON preserves editable cues, settings, identifiers, title and text-density threshold. It does not save the temporary preview position or current search. No account, server upload, autosave or AI service is involved. Downloads are generated locally and can contain your subtitle text; review them before sharing.
For format background, read the Library of Congress description of SRT and MDN's WebVTT format reference. These independent references do not imply endorsement.