🧹 Text Cleaner

The Text Cleaner applies the options you select to the text in the input box. Its default pass removes web addresses, email addresses, HTML markup, many emoji and invisible characters, then normalizes spacing and common quotation marks. Each option is independent, so you can keep email addresses while stripping HTML, or preserve emoji while fixing whitespace. Character counts below both boxes make it easier to see how much was removed before you copy the result.

This is deterministic cleanup, not a rewriting or fact-checking system. It does not summarize, translate, correct grammar, decide which URL is meaningful, or understand whether an email address belongs in a spoken script. The optional special-character filter is intentionally broad and can remove symbols used in formulas, code, pronunciation guides, or specialist notation. Keep that option off unless the output is plain prose and inspect technical material carefully.

Use the side-by-side boxes as a review step. Names, initials, product codes, dates, currency, mathematical symbols, and abbreviations often need manual pronunciation edits even after markup is gone. Removing a URL may also join words that were separated only by the address, while removing an email can erase context such as a support contact. Read the sentence around each important removal and restore a spoken replacement when the information should remain audible.

The input and output exist in the current page rather than a cloud editing account. Clean sends the selected rules through local JavaScript; Copy writes the result to the browser clipboard; Send to TTS places the chosen text in session storage and opens the OfflineTTS workspace in another tab. The cleaner itself does not invoke a speech model. Once you send text into TTS, the selected engine and language determine model downloads and whether non-English Kokoro phonemization is required.