Skip to main content

New Tech Can Copy And Mimic Your Handwriting

By September 1, 2016March 2nd, 2023Blog, Technology News

rsz_new_tech_canResearchers at University College London have created a fascinating algorithm that can take a sample of hand-written text, and imitate it with eerie similarity. At first glance, that might seem like a silly, or even trivial thing to spend time on, but make no mistake. The problems being solved here are far from trivial.

As an experiment, try writing the same sentence ten times. See how much variability there is in the words, letters and spacing? Not only does the spacing between the words in the sentence vary from one to the next, but so does the spacing between individual letters. The shape can vary markedly from one iteration to the next, too.
Fortunately, this is the kind of problem that machine learning is especially adept at solving, as the researchers have recently indicated.

As to how good their algorithm is…not even the authors of the samples could tell the difference, when comparing their own handwriting with sentences generated by the computer.

The algorithm works by marking up each letter, and every punctuation mark, analyzing them in detail, and examining the variance between duplicated letters to create a baseline. At that point, if the user types in text into the box provided, they’ll get back the same words, written in their own hand.

In terms of practical application, one of the things the team is already doing is finding samples of long dead famous people, and reproducing their works in their own handwriting. So far, the team has found viable samples for Sir Arthur Conan Doyle and former President Abraham Lincoln.

In terms of practical application, it has enormous potential for the disabled, and Amazon’s floral delivery service is currently investigating its use. While your company may not have a specific need for this, it’s a fascinating case study in the capabilities of machine learning.

Chris Forte

Chris Forte, President and CEO of Olmec Systems, has been in the MSP workspace for the past 25 years. Chris earned his Master’s Degree from West Virginia University, graduating Magna Cum Laude. He was a past member of the Entrepreneurs’ Organization, a current member of the New Jersey Power Partners and Executive Association of New Jersey, where he has previously served on its board of directors. In his spare time, Chris enjoys traveling with his family. He also admits to being a struggling golfer and avid watcher of college football and basketball. He currently lives in Boonton Township, NJ with his wife, two daughters, son, and black lab Luna.