Handwritten Text Recognition Sandbox

This is a “playground” to explore how different handwritten text recognition models work with different types of documents. Using FromThePage’s “AI-Assist” feature you can get an overlay of the machine generated text over each page image, making it easy to see where AI gets it right and where it doesn’t. Using FromThePage’s “AI Text” button you can explore seeding transcription with AI generated text. These features will evolve over time as we gather input from transcribers and institutions.

There are facets on the right side that let you filter by era, language, and model.

Leave a comment on the page with your thoughts, feedback, challenges, frustrations — we read and value every one.

You’re welcome to explore any of the documents. If you want to try transcribing text, click “sign up to transcribe” at the top.

We’re currently looking for 10 development partners for FromThePage's AI-Assist. We know we're going to need help to refine how humans and machines can work together to transcribe text, so we *really* value your willingness to explore with us.

Here are the details:

- 6 month subscription

- up to 50,000 pages

- access to state of the art handwritten text recognition technology

- 1-2 feedback calls required -- informal, but we want to build the best partnership between humans and AI possible, and

that's going to take more perspective than just ours!

- $1500 (valued at $6800 -- at $5300 discount!)

Because of the time required to support and work with folks on this new technology, we're only accepting 10 development partners.

‌

What to expect

We'd have an initial zoom call so we -- together -- can select an initial set of documents that will be successful in this program. We'll triangulate between:

- your organization's goals and priorities

- your documents' appropriateness for HTR

- how interested volunteers will be in the material

(Don't worry -- we'd help you figure all this out!)

‌

You'd be able to choose between promoting your project to our cadre of volunteer transcribers or working with students, researchers or staff. If you're working with students, researchers or staff the project could be private; public volunteers would need to be a public project.

‌

Questions? Drop Sara a note at sara@fromthepage.com or schedule a Zoom call.

March 29, 2024: Added two structured data examples, the Music Hall Accounting Ledger Cash Book and the Carnegie Hall Booking Ledger