Methods & limits
Subtitles are not one string. They are an ordered set of timed intervals.
Each cue’s start, end, and text are the core data, so conversion and sync return to one explainable model.
Explicit SRT and WebVTT support
The tool promises UTF-8 SRT and WebVTT only. It does not claim full support for complex VTT styles, regions, notes, or ASS, SSA, and LRC syntax.
Dual anchors are linear correction
Two anchors determine both starting offset and time rate for continuous drift. Inserted or removed video segments create jumps that need segmented editing.
Text extraction does not rewrite
Subtitle-to-text removes structure and optional tags only. It performs no recognition, translation, summary, or polishing that could silently change the words.