I combined assignments from Computational Approaches to Typography and Machine Learning for the Web. For this project, I made an asemic writing generator that responds to user-set gestures.

I was inspired by this work by Cecil Touchon. I wanted to experiment with writing the ‘text’ in a pleasing graphic layout. I broke down the asemic text generation into 4 elements:

• writing strokes
• long writing strokes
• spaces
• new line and carriage return

Each of these actions correspond to a gesture that can be set by the user. The image classification code is based off of this ml5 example of KNN Classification on Webcam Images with mobileNet.

ml5 KNN Classifier documentation

my github repo with code

live demo

Some outputs: