01
Who it’s for
A translator with a folder of chat screenshots to transcribe.
Translation and immigration desks · live
A folder of chat screenshots, typed up one page at a time.
For a translator with a folder of chat screenshots to transcribe. They typed each one into Word by hand. I built a tool that turns the folder into one Word document, the picture on the left and the text on the right, and that never pays twice for a page it already finished.
Python command-line toolOptional web formLive · private
Chat screenshots6 items · sample
Screenshot → Word
Upload chat screenshots and get back a Word doc with each image and its transcribed text side by side.
Conversions
result.jsonsaved after each image
result.docx0 pages
The Word document builds here: one page per screenshot, the picture on the left, its text on the right.
Waiting for screenshots.Read by the AI service: 0Skipped, already done: 0
A simulation: the screenshots are drawn for this page and the people in them are made up. The reading is faked; the real tool sends each image to an AI service and copies its text word for word. Keys: Enter on the folder adds it, R runs or resumes, S stops.
What it’s for
Four parts, plainly.
02
By hand, until now
They typed each screenshot into Word, one at a time.
03
What I built
It turns a folder of screenshots into one Word document, one page each, with the picture on the left and the transcribed text on the right. If a run stops, it picks up where it left off without paying again for finished images.
04
Where it is
Live and private: a Python command-line tool with an optional web form.
Rough edges
What it doesn’t do.
- Word for wordIt copies the text exactly and does not translate it.
- Keys and batch sizeIt needs an AI service API key, and the hosted web form only takes small batches.
Honest line: the page shows the shape of a run, not the real reading. Finished pages are written to a manifest after each image, which is what makes the resume free.