Handwriting Synthesis | 手写笔迹合成

Implementation of Handwriting Synthesis with Recurrent Neural Networks








Please "Enter text" first

Preview Image








    Note:
  1. This website is only for demonstration and research purpose, please do not use it for commercial or any illegal purpose.
  2. This is a reproduction of sjvasquez/handwriting-synthesis on GitHub by Sean Vasquez.
  3. This is a implementation of the handwriting synthesis experiments in the paper Generating Sequences with Recurrent Neural Networks by Alex Graves.
  4. The valid character set is {'?', 'U', 'M', 'r', 'o', '1', '5', ',', '2', '8', 'w', 'v', 'x', 'B', 'I', '4', 'W', 'a', 'p', 'q', 's', '!', 'T', 'Y', 't', 'C', 'j', ' ', 'K', 'P', 'f', 'b', '(', '3', '"', 'G', 'V', 'k', 'm', '7', '6', 'd', 'D', "'", '.', 'O', '#', '9', 'S', 'F', ':', 'J', 'L', '\x00', 'h', 'n', 'i', ')', '0', 'E', '-', 'z', ';', 'g', 'y', 'e', 'R', 'A', 'u', 'N', 'l', 'H', 'c'}
  5. The length of input characters is limited up to 75 characters.
  6. The speed of synthesis is low and limited by the performance of the server.