How SRT files work
Players read SRT cues in time order even if numbering slips, but broken numbering still confuses many tools. Timing uses milliseconds, overlaps annoy viewers, and blank lines separate cues.
You will learn how a cue behaves from start to end time, why overlaps matter, and how to edit without corrupting structure. You will also see why comma milliseconds are not optional style.
Generate a sample SRT from any clear audio clip using our free tool, open it beside this article, and compare line by line. Concrete examples beat abstract rules.
Free tool
Rather do it than read about it? Upload an audio or video file and get a timed .srt back.No signup, no payment, up to 100 MB per file.
Step-by-step guide
Step 1: Understand cue numbering
Starts at one and increments. Some tools renumber for you. If numbers jump or duplicate, fix before export; picky importers choke on messy indices.
Step 2: Parse the time line
Two timestamps in HH:MM:SS,mmm format. Milliseconds matter when dialogue is fast or when cues sit close together.
Step 3: See how text attaches
Multiple lines show as stacked lines in the cue. Keep total reading time realistic for mobile viewers.
Step 4: Learn player display rules
Show at start, hide at end unless another cue already began. Overlapping times can make players flicker between lines.
Step 5: Avoid overlaps
They can flicker or stack awkwardly. Shorten end times slightly to leave breathing room between cues.
Step 6: Edit in small chunks
Large find-replace risks breaking arrows or commas. After bulk edits, import into a player for a smoke test.
Step 7: Save backups
Text files are easy to break and easy to restore if you planned. Keep _v2 copies before risky experiments.
Tips for better subtitles
- Milliseconds matter for rapid dialogue.
- Waveform editors help when timing feels tight.
- Keep a test player open while editing raw text.
- Sort by time when in doubt.
- Document your tool’s renumber behavior.
- Never assume silence means no cue.
Common mistakes
- Using dots instead of commas in milliseconds Some tools accept both; many do not.
- Empty cues A cue with a number and a time but no text makes some players blink a blank box on screen. Delete them rather than leaving placeholders behind.
- Out-of-order times Cues are expected to run forward in time, and a block that starts before the previous one ended confuses strict parsers. Sort before exporting, not after a problem appears.
- Editing without preview Syntax errors in a subtitle file are invisible in a text editor and obvious in a player. Load the file once before you send it anywhere.
FAQ
Why is there a comma before the milliseconds?
It is what the SubRip format specified, and parsers were written to match. A dot is WebVTT syntax: tolerant players accept it, strict ones reject the cue, so a file with dots can load and then quietly show nothing.
What happens when two cues overlap?
Behaviour depends on the player: some stack both lines, some drop one, some flicker between them. None of that is what you intended, so end each cue before the next one starts, even when the speech genuinely overlaps.
How long should a cue stay on screen?
Roughly one to six seconds. Under a second the eye cannot catch it; over six it sits there long after it has been read. Keeping lines short is what makes both ends of that range comfortable.
Conclusion
SRT behavior is predictable once you respect cue structure and time order. Preview early and often.
Generate timed text from audio, then refine cues with confidence.
When something looks impossible, suspect syntax first: commas, blank lines, and cue order solve most mysteries.
Now on your own file
The quickest way to finish this guide is to look at the format on a file of your own: upload it, correct any line against the audio, download the .srt.Your upload and the subtitles are deleted automatically about an hour after the conversion.