Guide

Media & Text · Local Timing Engine

Subtitle Converter & Time Shifter

Convert SRT and WebVTT (.vtt) subtitles, shift timestamps by milliseconds, and correct PAL/NTSC frame-rate drift (25fps to 23.976fps) live in your browser.

Time ShiftFPS Presets
Detected Cues3
Output FormatVTT
First Cue Start00:00:01.000
Last Cue End00:00:11.250
Source Subtitles (SRT / WebVTT)
Live Converted Output (.vtt)263 chars

Live conversion ready (3 cues · shift 0 ms · 25→25 FPS).

SUBTITLE TIMING & FORMATS

Fix out-of-sync captions and convert SRT ↔ WebVTT cleanly.

Constant Delay vs. Frame-Rate Drift

If captions are consistently 1.5 seconds late from start to finish, use the Time Shift control (-1500 ms). If captions start in sync but drift gradually further apart over an hour, your subtitle track was timed for a different frame rate—use the PAL 25 ↔ NTSC 23.976 / Film 24 FPS presets.

Web-Ready HTML5 Video Tracks

HTML5 <track kind="subtitles"> elements require WebVTT (.vtt) formatting with dot-separated milliseconds and a WEBVTT header. Convert legacy .srt files in one click.

Frequently Asked Questions

Are my subtitle files uploaded to a server?

No. All SRT/WebVTT parsing, millisecond shifting, and frame-rate rescaling run locally in your browser with zero uploads.

How does frame-rate (FPS) drift correction work?

Each cue timestamp is multiplied by (Source FPS ÷ Target FPS) to match the playback speed of your target video encode, and then the millisecond offset is applied.

Can I shift subtitles earlier (negative offset)?

Yes. Enter a negative millisecond value (such as -500 or -1200) or drag the slider left, as long as no resulting cue shifts before 00:00:00.000.

What is the main difference between SRT and WebVTT timestamps?

SRT uses a comma before milliseconds (00:01:02,500) and requires numeric cue indices, whereas WebVTT starts with a WEBVTT header and uses a period (00:01:02.500).

What are the supported file size and cue limits?

Files up to 1 MB and 5,000 cues are supported. WebVTT NOTE, STYLE, and REGION blocks are rejected so output remains clean and deterministic.