Humata
ProductivitySearch, summarize and compare uploaded documents with source-linked answers. Humata adds team-oriented document access and page-based plans for ongoing knowledge work.
Questions across filesCited passages
PDFMATHTRANSLATE / READ THE TRANSLATION BESIDE THE ORIGINAL
PDFMathTranslate
An open-source PDF translation project focused on scientific documents and bilingual comparison while preserving layout elements. The current next branch uses BabelDOC and supports local interfaces plus configurable translation services.
Official PDFMathTranslate-next README preview of a scientific paper and its Chinese translation, credited to the project’s BabelDOC demonstration; not a FindGoodAI translation-quality benchmarkExplore the image ↓
What you can do with it
What this tool is for
PDFMathTranslate helps readers translate PDF documents while retaining more of the page structure than a simple copy-and-paste text workflow. Scientific papers are a central use case because formulas, diagrams, references and multi-column layouts matter to comprehension. A bilingual view can make it easier to compare a translated explanation with the original wording.
This guide follows the current PDFMathTranslate-next branch and its pdf2zh-next commands, rather than combining instructions from different generations of the project. The translation engine and the PDF layout engine perform different jobs: one produces translated text, while the other attempts to place it in a useful document structure. Both need review on the actual file.
01
Read a foreign-language paper
02
Prepare a technical reading pack
03
Compare translation engines
04
Create an accessible study workflow
A closer look at the product
Official product image. Click to inspect the details.
Product interface and examples
Key features, practical tasks
Capabilities in practice
- Layout-aware processing: The project is designed to preserve important page elements while replacing or pairing translated text. Inspect columns, headings and reading order on a sample, because a visually complex paper can expose problems that a simple page does not.
- Bilingual comparison: Original and translated material can be compared as part of the reading workflow. This is useful for checking a technical phrase or a cautious claim without repeatedly switching between unrelated files.
- Scientific document focus: The project highlights formulas, charts, contents and annotations as important preservation targets. Treat preservation as an intended capability, not a guarantee that every equation label or figure caption survives every PDF unchanged.
- Configurable translation services: The service guide lists multiple backends with different support levels. Choose based on language quality, data requirements and cost, and keep the backend configuration with the output so a translation can be reproduced later.
- Local model route: The WebUI guide documents using Ollama as a backend. A local model can avoid a hosted translation bill, but available memory, model quality and processing time still limit what is practical on the user's computer.
- Graphical and command-line workflows: The project offers a browser interface around a local process and a command-line route. The interface suits occasional reading tasks, while a recorded command and configuration can help maintain a repeatable document pipeline.
- Deployment choices: Official guides cover desktop-oriented installation and containers. Pick one supported route and keep its configuration together, rather than mixing package names and flags from the original and next branches.
Get started, step by step
- 01
Identify the document
- 02
Choose the current setup
- 03
Translate a sample
- 04
Review and retain provenance
A practical first workflow
A practical first workflow
- Identify the document. Check whether the file is readable text, scanned pages or a mixed layout, and confirm permission to process it. Keep the original unchanged so layout or translation errors can be traced back.
- Choose the current setup. Follow the next branch's installation and interface guide. Select the source and target languages and configure the intended backend without exposing API credentials in shared scripts.
- Translate a sample. Start with a small range containing prose, a figure and an equation. Compare reading order, technical terms and captions before committing resources to a long document.
- Review and retain provenance. Save the bilingual result with the original, backend name and processing date. Manually check passages used in a report or publication, and mark unresolved translation ambiguities in the reading notes.
Example prompt or task: Translate a representative page range of this scientific PDF from English to Simplified Chinese, then compare the original and translated versions for terminology, equation labels, figure captions and missing paragraphs.
This is an editorial starting workflow, not a measured performance result.
Find the right plan for you
Free software
Explore official plans ↗Free software Models, hardware and hosting are separate
The next project is distributed under AGPL v3 and can be run without purchasing the software. A local model backend uses the user's own hardware; a hosted model or translation provider can charge separately or impose a limited free allowance.
What is free and what can cost extra
The next project is distributed under AGPL v3 and can be run without purchasing the software. A local model backend uses the user's own hardware; a hosted model or translation provider can charge separately or impose a limited free allowance.
Do not assume that a public demonstration service is unlimited or that every listed provider remains permanently free. For a recurring workflow, estimate cost from a small representative document and keep the backend choice explicit. Review the project's license before redistributing modifications or operating a modified network service.
A few things worth knowing
Frequently asked questions
Is the software free?
Yes. The current next branch is open source under AGPL v3.
Does that make every translation request free?
No. The selected backend may have fees or limits.
Can a local model be used?
The WebUI guide documents an Ollama route; hardware and model suitability still matter.
Which project generation does this guide use?
PDFMathTranslate-next and its pdf2zh-next workflow.
Will every formula and layout be perfect?
No. Check a representative sample and important pages.
Can I use it for Chinese reading notes?
Yes. Select the supported target language and preserve original technical terms where useful.
Does a translated PDF replace the original citation?
No. Cite the original work and verify claims against it.
The details behind your decision
Explore platforms, inputs and outputs, licensing, and access requirements.
Four useful starting projects
Four useful starting projects
- Read a foreign-language paper: Translate a short section first, then compare the original terminology and the translated explanation. Keep important claims in a reading note with the original sentence so the translation does not become the sole evidence.
- Prepare a technical reading pack: Process documents that your team has permission to use, keeping originals beside translated copies. Add a brief note about the selected engine and the passages that were manually checked.
- Compare translation engines: Use the same page range from one paper and evaluate two backends on terminology, omissions and layout. Focus on the subject matter your readers actually need rather than selecting the most fluent general prose.
- Create an accessible study workflow: Use the translated copy to orient a reader, then return to figures and formulas in the original. A translation can reduce language friction without replacing the specialist knowledge needed to interpret the research.
Platforms, inputs, outputs and setup
Platforms, inputs, outputs and setup
The current branch builds on BabelDOC and provides WebUI, command-line and container routes. Its WebUI documentation describes a local browser address and configurable language settings; opening a browser page does not by itself mean the translation model is local.
Scans, unusual fonts, multi-column reading order and translated text expansion can complicate the result. If a paragraph does not fit cleanly, review the source and output rather than silently using a damaged page. Retain original citations and numeric values when creating notes from the translated document.
Data, permissions and licensing
Data, permissions and licensing
The PDF can be processed through local software while its text is sent to a hosted translation engine. Check both the document-processing location and the configured model endpoint; these are separate parts of the data path.
Use documents you are authorized to process and keep restricted manuscripts out of unapproved services. Translation does not remove copyright or publication restrictions, and an automatically translated paper should not be redistributed as though it were an authorized edition.
Sources, documentation and help
Official sources and review date
Official sources and review date
Reviewed October 3, 2026. Product facts come from the official sources below. Suggested projects, prompts and review methods are FindGoodAI editorial guidance, not measured performance results.
Your experience can help someone choose
COMMUNITY
Sign in to rate and comment
Tried it? Share your experience
Loading experiences…
Manage my comments
Community and moderation
Saved tools are private. Approved comments are public; edits return to moderation. Comment counts include approved comments and replies. Each person has one rating. New ratings are automatically approved; withdrawn, unapproved or invalid ratings do not count. Share real experiences and avoid spam, private information or personal attacks.
For moderation appeals or data requests: support@findgoodai.com
NEXT STEP
Explore PDFMathTranslate ↗Get started with PDFMathTranslate through its official guides
Bring a real task and see how it fits the way you work.
Jump to a section
User experiences