What this tool does
Downloaded subtitle files often carry things they should not. The cleaner strips them in one rules-based pass:
- Fansub watermarks, URLs and ad entries
- SDH captions (like [footsteps], (music)) and speaker labels
- ASS effects, HTML tags, control characters
- Duplicate entries, empty entries, punctuation and spacing mess
Upload and detect
Drag in a file or click to upload — SRT, VTT, ASS, SSA and TXT are supported — or paste subtitle text directly. Encoding is auto-sniffed: if UTF-8 decoding produces replacement characters, it falls back to GBK or Big5, and the page shows "Encoding fixed as {encoding}". If you upload dual-track bilingual subtitles (two identical timelines, one Chinese and one English), they are merged into one bilingual track with an "original on top / translation on top" switch.
Configuring the 13 rules
The interface has four groups with 13 checkboxes: Tags and formatting (ASS effects, HTML tags, control characters, renumber after sorting), SDH and sound effects (SDH captions, speaker labels, pure lyric lines), Watermarks and junk entries (watermarks and ads, duplicate entries, empty entries), Text fixes (punctuation and spacing, all-caps repair, merge short lines). Only "lossless and certain" items are on by default; items that may drop content (lyric lines, caps repair, merge short lines) are off — tick them when needed. Each item has an ⓘ hover tip, and any checkbox change re-runs instantly.
Change diff and undo
Results are not overwritten directly: every change is listed in a diff table with a rule tag (why this entry was edited or deleted). Want one back? Click "Restore" on that row; "Restore all" brings everything back in one click, and clicking it again re-applies the full cleaning. The stats bar shows counts for original / cleaned / deleted / modified / restored.
Preview and export
Toggle between "Cleaned / Original" views. In the original view, deleted entries are struck through and marked "Deleted", and the exported file matches the preview exactly. Export formats: SRT (default), VTT, ASS (available when the source is ASS, with the style section preserved as-is), and plain-text TXT without timelines. The filename gets a `.clean` marker automatically.
FAQ
Will it delete normal dialogue by mistake? The watermark rule first strips watermark lines (keeping dialogue lines inside bilingual entries) and only deletes the whole entry if traces remain; SDH removal only touches bracketed content with SDH keywords that is 40 characters or shorter — ordinary parenthetical dialogue stays.
How are pure lyric lines distinguished from lyrics inside dialogue? Only entries where every single line is wrapped in ♪ music notes or markers like [music] are deleted; dialogue containing lyrics is kept.
Could speaker labels hit Chinese text? Only an all-caps alphanumeric name at line start followed by a colon (like `GEORGE:`) matches; Chinese colon sentences never do.
Can I clean multiple files at once? No — one file per run. Download the result, then upload the next one.
Privacy
Cleaning runs entirely in your browser; no subtitle content is ever sent to a server.
Related tools and guides
For a short walkthrough focused on removing watermarks and ads, see remove subtitle ads and watermarks; run your file through the subtitle quality check before handing it over; to change formats use format conversion.