OnlyText

How to Use OnlyText

1) Load one or more transcript files (.docx, .txt, or text-based .pdf).
2) Click “Create OnlyText TXT.”
3) Use the preview menu to compare the original input with the cleaned TXT output.
4) Choose the export mode and click Export. ZIP is the default; individual downloads are available as an option.

Filename rules
• All outputs are .txt files.
• Existing .txt input filenames are preserved exactly.
• Non-TXT inputs, such as .docx or .pdf files, are exported with the same base filename and a .txt extension.
• Output filenames never include the original .docx or .pdf extension.

Transcript-start rule
• OnlyText looks for the first true transcript speaker label followed by a colon, such as FLEMING:, Jewell:, Malcolm Jewell:, Interviewer:, Respondent:, or Speaker 1:.
• It scores likely speaker starts by looking at the label shape, repeated labels, nearby speaker turns, and whether the label is followed by transcript text.
• Everything before that first speaker label is removed.

Cleanup rules
• Removes title page and front matter before the first detected speaker label.
• Removes standalone page numbers, roman numerals, page break markers, and common repeated header/footer lines.
• Removes repeated short lines that occur multiple times, which helps with exported headers and footers.
• Preserves transcript speaker labels and paragraph breaks.

Notes
• All processing occurs locally in the browser.
• DOCX conversion uses Mammoth.js to extract text from Word files.
• PDF conversion uses PDF.js to extract embedded text from text-based PDFs. Scanned/image-only PDFs are not supported.
• Exports are always UTF-8 plain text. No encoding choice is required.
• Results should be reviewed before deposit or publication.