PPTX Chinese to English Translator

Problem

Chinese PPTX decks often contain text boxes, tables, chart labels, and Chinese annotations embedded in images. Translating slide by slide can make terminology, tone, product names, and company names inconsistent, while Chinese text in screenshots, flow diagrams, or illustrations is easily missed, leaving the final deck in a mixed-language state.

How It Works

The skill treats the PPTX as one complete document:
- extract_content.py extracts all text, formatting metadata, and images into a structured content.json.
- It then reads the full content, identifies the business, academic, or technical register, and builds a glossary to keep product names, abbreviations, and domain terms consistent.
- It emits translations.json and overlays.json: the first covers shape and table text, the second covers Chinese text inside images.
- apply_translations.py writes English back into the document while preserving font size, color, bold, italic, and alignment, expanding text boxes when overflow is likely.
- overlay_image_text.py locates Chinese regions in images, covers them with white rectangles, renders the English replacement, and replaces the original image.

Limits

Best for Chinese-to-English office, technical, or academic presentations. English often runs 20%–40% longer than Chinese, so layout, line spacing, and chart space still need review; EMF and WMF raster or vector formats may not be processed; shapes without text frames, unrecognized areas, or failed items are skipped, and the completion report warnings should be checked.