Wednesday, 29 November 2017

Handwriting Transcription

TNA, the UK National Archives, jumping the gun a little with a report A Year in
Archives 2017, gives examples of progress during the year in the UK.

The one that caught my attention was University College London’s Bentham
Project using a Handwritten Text Recognition (HTR)
programme transkribus.eu/.
If successful HTR would be a breakthrough technology for genealogy. Imagine opening up all those PCC probate records without the need to learn paleography and making them full text searchable. HTR is something that's been worked on for years, but  may finally be coming close to being a practical tool.
That's just as voice recognition has emerged -- see this example of a YouTube video with subtitles produced by voice recognition - click on CC at the bottom right hand side.

3 comments:

Susan said...

This would be brilliant.....fingers crossed.

Flameseh? said...

OMG! This brings back memories of times past. I hope this comes to fruition. It won't be perfect I'm sure, but it'll will reduce a lot of "hair hurt". ;-)

Pierce said...

Even OCR of printed text has problems. A lot of the documents I have tried to read are so bad parts I cannot read, even when I know the context and the individuals that should be mentioned. And then there are the wills over 200 years old that I spend hours trying to comprehend. Maybe it could be programmed to recognize some of the ancient handwriting I've tried to work with.