PDFs back into Word and Excel files you can work with
Instead of drawing the page into Word, Tuowen rebuilds the document itself: paragraphs, headings, tables, contents, headers and footers are real Word structure, ready to edit, restyle and hand on. Digital PDFs and scans alike, all offline on your own computer.
Every feature for 7 days of use after installing · Windows 10 / 11 · macOS (Apple silicon) · Chinese interface
The film is in Chinese.
Many PDF-to-Word files look just like the original until you change something. Tuowen aims for a Word file that looks as if someone had carefully typeset it.
| Common PDF-to-Word problems | What Tuowen does | |
|---|---|---|
| Body text | Text boxes hold the page together, every line is a paragraph, one edit throws it all off | Real paragraphs: sentences across lines and pages joined into one, laid out with styles |
| Headings and contents | Headings are bold plain text; the contents are a row of dots | Heading styles for headings; the contents become a Word TOC field you can click and update |
| Tables | A table across pages becomes one table per page; tables inside cells fall apart | Tables across pages joined into one; nested tables, three-line tables and merged cells restored |
| Headers and footers | Headers and page numbers left in the body, out of place on every page | Real headers and footers, sectioned like the original, page numbers counted from the same start |
| Scans | No telling right from wrong in the recognised text: check every character | Uncertain characters highlighted in yellow; seals and handwritten signatures kept as images |
| Bills to Excel | A drawing of the page: merged cells everywhere, numbers stored as text | Tables of the same kind joined into one data table, numbers as numbers, formulas generated |
We test on engineering tenders, contracts, technical specifications, design calculations, standard drawing sets and science textbooks, and reconvert all of them after every change.
01 · Structure
Sentences that run across lines and pages become one paragraph again. Chapters get Heading 1 and articles Heading 2, so the Navigation pane shows the whole outline. No text boxes holding the layout together: change a word and the rest of the text simply flows on.
02 · Contents, headers and footers
The original table of contents becomes a Word TOC field: click an entry to jump, and “Update table” refreshes only the page numbers. Headers and footers follow the original’s sections, with left, centre and right parts aligned, and page numbers start where the original starts counting.
03 · Tables
Tables are built from the ruling lines in the PDF. Continued tables are joined back into one and repeated header rows removed; nested tables, three-line tables with only horizontal rules, and tables without side borders are all recognised, and merged cells come back as they were.
04 · To Excel
Two ways to lay it out. As in the original: one worksheet per table, with titles, units, notes and signature rows, and amounts, totals, tax and summaries written as formulas. As a data table: tables of the same kind joined across pages, broken rows rejoined and numbers stored as numbers, ready to filter and sum. A bill of several thousand pages, checked item by item against the costing software’s own file, came out with nothing missing.
05 · Scans
Built-in offline text recognition straightens skewed pages and turns sideways tables upright, then rebuilds paragraphs and tables as usual. Characters it is unsure of are highlighted in yellow; red seals and cross-page seals are kept as transparent images and handwritten signatures as the original picture. Fonts that copy out as gibberish and text turned into outlines are recovered as well.
06 · Technical documents
In scanned textbooks, standards and design calculations, formulas, worked examples and chemical equations can be recognised as editable Word equations, graded by confidence: A checked, B to be proofread (yellow), C for reference only (orange), with the doubts written in comments. Fix the uncertain ones against the image of the original. A separate compilation can list each formula and figure with its number and page. The recognition model is a separate download; without it conversion works as usual and formulas stay as images.
07 · Two layouts
Closest to the original: font sizes, line breaks and page breaks follow the source as closely as possible, for checking and archiving. Best layout: one font, size, line spacing and first-line indent throughout, with text flowing naturally across lines and pages, ready to rewrite.
Opening a file counts every page and tells you roughly how long it will take. For a long document, convert the first 3 pages first and go on when you like the result.
Stop halfway and the pages done so far are written out. With formula recognition on, the Word file comes first and the formulas are recognised in the background, then saved as a second copy.
Two-column layouts become Word columns; sheets in standard drawing sets map page to page, with frames and title blocks fixed in place.
The “/” on fill-in lines, the per-mille sign “‰” and the diameter sign “Ø” are corrected by their shapes, fill-in underlines are kept, check boxes become Word symbols, and text watermarks are dropped instead of ending up in the body.
Drafting and answering tender documents, reusing technical specifications and contract clauses.
Bills of quantities and estimates turned into Excel files that calculate and sum.
Design specifications, technical standards and lecture notes back into editable Word, formulas included.
Contracts, variation orders and all kinds of scans turned into electronic documents.
Tuowen is getting its final polish. Once it is finished, you can buy it here. Until then, download it and use every feature free for 7 days of use.
Buying for an organisation, want to know the price in advance, or have a file that does not convert well and would like us to look at it? Write to gouqingyu@163.com.
Try itTuowen is improving quickly, and the PDF that came out wrong for you may be exactly what we need. 「反馈问题」 (Report a problem) on the result page makes a feedback folder; zip it and send it through the feedback page, and we will get back to you.
Send feedbackTuowen is built and tested on Chinese documents, and its interface is in Chinese. Text recognition reads Chinese and English, but our test documents are Chinese, so please try your own files during the trial.
PDFs exported from software convert best. Scans, PDFs that copy out as gibberish and PDFs with text turned into outlines are recovered with the built-in offline text recognition. Each page is diagnosed when the file is opened, and the result lists the pages where recognition was used.
No recognition is 100% accurate. Tuowen marks characters it is unsure of in yellow in Word, and the cells in Excel, so you only need to check what is highlighted instead of comparing every character. Editable formulas are graded A / B / C, with B and C shaded and commented; the image of the original is the one to trust.
Recognising formulas needs a model of about 400 MB (Apache-2.0 licence) that is not part of the installer; it is a separate download on the download page. Install it when you convert technical scans; without it conversion works as usual and formulas stay as images. One file for Windows and macOS; when you tick the formula option, the app tells you which folder it goes in. Formula recognition is a Pro feature.
Standard converts to Word and Excel and recognises scans, with formulas kept as images. Pro adds formula recognition (formulas, worked examples and chemical equations as editable Word equations) and drawing list extraction (a command-line tool for now). Lots of technical scans: Pro. Mostly contracts, tenders and bills: Standard is enough.
Keys are bound to a computer. Click the badge in Tuowen’s top right corner: the panel shows the machine code. Enter it on the order page to get that computer’s key, then paste the key into the same panel. To change computers, enter the new machine code on the order page; each seat can be moved twice a year.
After installing, every feature is yours for 7 days of use (only days you open a file count, within 30 days). After the trial or a yearly license ends, Tuowen still opens but converts only the first 5 pages of each file, without formula recognition, with a note at the top of the document. Renew and activate the new key. A yearly plan renewed early starts the day after the old one ends.
No. The text and formula recognition models are installed on your computer, everything is processed offline, and the app does not connect to the internet.
Tuowen writes standard Word (.docx) and Excel (.xlsx) files that Microsoft Office and WPS both open and edit. There is a Windows version and a macOS version for Apple silicon; the Mac app is not notarised by Apple yet, so the first start needs one visit to System Settings → Privacy & Security, as the download page explains.
