The calligrapher.ai demo version of Calligrapher.ai, which runs in the browser via JavaScript, instantly turns any typed text into handwritten text. After entering a proposal, the site offers 9 different styles, each of which can be adjusted for speed, width and legibility. You can download the text in an SVG file.
The site is particularly interesting because it does not use well-known handwritten fonts – every letter here can look original, even if it is repeated many times.
Calligrapher.ai was created by machine learning researcher Sean Vazquez, who used research from a 2013 paper by Alex Graves of UK-based AI company DeepMind. Calligrapher.ai “draws” each letter as if it were written by a human hand, guided by data obtained from a recurrent neural network (RNN) that was trained on the basis of the IAM On-Line Handwriting Database (a database containing digitized handwriting samples of 22 people ).
Calligrapher.ai’s handwriting synthesis model is tuned to an English-language letter, and users have reported problems reproducing diacritical marks commonly found in other languages.
In 2018, Vasquez provided the code base that provides a demo of the app on GitHub so that it can be adapted to other apps.
With neural networks now working with speech, images, video, printed and now handwritten text, it seems that AI can claim to succeed in all areas of human creativity.
II against copyright. The free Watermark Remover.io tool effectively removes watermarks from stock images
Source: Arstechnica
