Skip to main content

DOCX to TXT converter

Extract plain text from Word docs

No watermarkFree & unlimited
All processing happens in your browser. No data is sent to any server.

This is the most destructive of the Word conversions, and deliberately so. Mammoth opens the .docx package and returns the raw text of every paragraph, with nothing else attached: no styles, no tags, no markup of any kind. Each paragraph becomes one line in a UTF-8 text file. That makes the output ideal for word counts, diffing, grep, corpus work and anything that chokes on formatting, and useless for reproducing how the document looked. What you get is the words, in reading order, and only the words.

Key facts about DOCX to TXT converter

Key facts about DOCX to TXT converter
ExtractorMammoth extractRawText, which walks the document XML and returns paragraph text with no markup at all.
Line structureOne line per Word paragraph. No wrapping is applied, so a long paragraph is one very long line.
TablesFlattened. Each cell contributes its own line, so column alignment and row grouping disappear entirely.
ListsBullets and automatic numbering are gone, because they are list properties rather than text in the document.
ImagesProduce nothing. Pictures, charts and diagrams leave no placeholder and no caption unless one was typed.
HyperlinksThe visible text is kept, the target URL is not. A linked phrase reads as ordinary words.
Headers and footersNot extracted, so running heads, footer notes and page-number fields do not appear in the output.
EncodingUTF-8 with no byte order mark, so accents, curly quotes and non-Latin scripts survive intact.
Typical size deltaA 1.2 MB illustrated Word report commonly extracts to 20 to 60 KB of text, since the images carry most of the weight.
Input accepted.docx only. The legacy binary .doc format is not read, and password-protected packages cannot be opened.
BatchUp to twenty documents per run; multiple results arrive as one ZIP built inside the page.

What happens to your file

Everything readable in your document is handled locally. The .docx is loaded as an ArrayBuffer, unzipped and parsed by Mammoth in this tab, and the extracted string is wrapped in a text/plain Blob for download. Because plain text is the most exposed form a document can take, it matters that no network request carries it: there is no upload step, no queue and no server-side copy of either the Word file or the text pulled out of it. Closing the tab discards both.

About this tool

  1. 1

    Select the document

    Drop a .docx file on the panel or click to browse. A batch of Word files can be queued together.

  2. 2

    Convert

    Mammoth reads the package and returns the paragraph text, which is wrapped as a UTF-8 text file.

  3. 3

    Download the .txt

    Save the file, or take a multi-document run as a single ZIP assembled locally in the browser.

  4. 4

    Re-add the structure you need

    Because tables and lists arrive flattened, restore any separators in a text editor before feeding the file onward.

Specs & compatibility
Input and outputTakes .docx (the Office wordprocessingml type) and returns the TXT as a blob with the text/plain type, named after the file you dropped in.
EngineMammoth 1.12.0 runs extractRawText over the .docx, which walks the document body and returns the characters with no styling stage in between.
ModesFile mode only: DOCX to TXT has no paste box, so the DOCX arrives through the picker or a drag and drop and the TXT comes back as a download.
Batch and capsTwenty DOCX files per run at 500 MB each; several TXT results are zipped in page memory as converted-txt-files.zip.
Browser and deviceDOCX to TXT needs only the File API, a Blob download and the parser already in the bundle, so every current browser behaves the same and the page keeps working with the network off.
  • Use this when a tool needs clean input: word-frequency scripts, plagiarism and similarity checks, translation memories and version diffs all behave better without Word markup.
  • Because each paragraph is one unwrapped line, open the result with soft wrapping enabled or your editor will show a handful of extremely long lines.
  • Tables lose their shape completely. If a document is mostly tabular, extract it to HTML instead and convert the tables from there.
  • Check the end of the file for footnote text, which is appended after the body rather than sitting where the reference marker was.
  • For a word count you can trust, extract first and count the text file. Word own count includes text boxes and other content that does not reach the output.
  • Feed a report into a script, a search index or a diff tool that expects plain text and cannot parse an Office package.
  • Strip a document down to its wording before pasting into a system that mangles rich text, such as a ticket or a form field.
  • Recover the readable content of a Word file on a machine with no Office installation and no permission to upload it anywhere.
No, and that is the point of this conversion. Plain text has no concept of fonts, alignment, columns, tables or page breaks, so all of it is discarded and only the paragraph text remains, one paragraph per line. If you need the structure, convert to HTML or Markdown instead, both of which keep headings, lists and tables in a form you can read and edit.
Each table cell contains one or more paragraphs, and every one of those paragraphs becomes its own line in the output. The result is the cell contents in reading order, top to bottom and left to right, with nothing marking where a row ended or which column a value belonged to. Documents that are mostly tabular are better converted to HTML.
Yes, and without a byte order mark. Accented characters, curly quotation marks, dashes, and non-Latin scripts such as Greek, Cyrillic or Chinese are stored correctly. Some older Windows tools assume the legacy system code page when no mark is present, so if such a program shows mojibake, tell it explicitly to read the file as UTF-8.
Their text is extracted, but it appears after the main body rather than at the point where the reference marker sat, and the marker itself is not distinguishable from ordinary text. For short documents this is easy to tidy manually; for heavily annotated academic work, the HTML conversion keeps the reference links and is much easier to follow.
Because almost all of a .docx is not text. The package holds formatting definitions, style tables, relationship files and, most of all, embedded images, which routinely account for the majority of the size. Extracting only the words commonly turns a one-megabyte report into a few tens of kilobytes, which is a normal result rather than a sign of missing content.
No. An encrypted .docx cannot be unzipped without its password, and this tool provides no way to supply one, so the conversion stops at the reading stage. Open the document in Word with the password, remove the protection, save an unprotected copy, and run that copy through the converter instead.

Related tools

View all

Updated

We use anonymous analytics to improve ToolChamp. No personal data is stored or sold. Privacy Policy