Scientists at TU Darmstadt have developed an adaptive “human-in-the-loop” system for synthesizing fonts that respond to individual readers and change to maximize their reading speed.
The appearance of text affects readability. For example, dyslexia-friendly fonts tend to be sans-serif, widely spaced, and without underlining, italics, and other variations to avoid an overcrowded appearance.
While some fonts are considered more readable than others, everyone has their own font preferences and needs.
The team of researchers from the TU Darmstadt Center for Cognitive Science began by investigating how text appearance affects readability, with a view to creating fonts that can be customized to improve readability.
Once they developed a technique for assessing reading speed, they had to develop a method of synthesizing new fonts to work alongside.
They accomplished this using a machine learning algorithm, which learned the structure of fonts by analyzing 25 popular and classic fonts, including Helvetica, Times New Roman, Garamond, and Futura.
The system was able to use these fonts to generate an infinite number of new fonts that could be any intermediate form of the 25 fonts, for example halfway between Garamond and Times New Roman.
They built a 3D rendering of the original fonts; in this infinite space, all points represent a possible font that is a combination of the original fonts.
The researchers applied Bayesian Optimization to develop their adaptive font. This starts with an a priori uncertainty about a volume in the font space, and selects consecutive points in this volume (representing potential fonts), for which the user’s reading speed is evaluated by means of a task.
Once the reading speed for that font is known, the algorithm reduces the uncertainty and moves the edges closer to the ideal font. The researchers call their system “AdaptiFont”.
Volunteers – all native speakers of German – were given text to read during an hour-long experiment, using this algorithm to optimize reading speed by changing the font. They read a total of 95 texts in various synthesized fonts from a German children’s encyclopedia Wiki, chosen for their simplicity.
The researchers showed that their algorithm did indeed generate new fonts to increase the reading speed of individuals. While all readers ended up with their own personalized fonts that they liked best, this font may not be optimal in all situations.
“AdaptiFont can therefore be understood as a system that dynamically and continuously creates fonts for an individual while reading, maximizing the reading speed in our moment,” said Professor Constantin Rothkopf, a cognitive science expert at TU Darmstadt.
“This may depend on the content of the text, whether you are tired or perhaps using other display devices.” The researchers recently presented AdaptiFont to the scientific community at the Conference on Human Factors in Computing Systems, and have since filed a patent application for the system.