Understanding TraDoc

Features

Capabilities available in the current TraDoc product.

Documents

  • Import EPUB, PDF, DOCX, Markdown and plain-text files.
  • Extract segments with reconstruction metadata.
  • Export a partial preview as soon as at least one segment is translated.
  • Reconstruct the final document in the same format as the source.
  • Produce an additional reflowable EPUB from a completed PDF.

Translation

  • Configurable source and target languages.
  • Adjustable segment size, temperature and concurrency.
  • Editable system prompt.
  • Optional glossary injected into the translation context.
  • Removal of reasoning blocks such as <think>.
  • Configurable retries and recovery behavior in the engine.

Progress tracking

  • Projects persisted in SQLite.
  • PENDING, RUNNING, PAUSED, COMPLETED and error states.
  • Live events over SSE.
  • Pause, resume, retry failed segments and delete a project.
  • Source/translation inspector with segment navigation.

Configuration

  • Independent settings for each provider.
  • Model discovery when the remote API supports it.
  • Connection test and translation sandbox.
  • Presets stored in the browser.
  • Hardware requirements estimator.

Proofreading

The backend can clone a completed translation into a proofreading project and includes a dedicated editorial prompt. The feature remains conditional on ENABLE_PROOFREADING and is disabled by default.

Understanding TraDoc

TraDoc's purpose, scope and the components involved in a translation.

Architecture

React frontend, FastAPI API, Python engine, SQLite and persistent storage.