Scholarlink

This study proposes a method for transferring a style aligned with the input sound to character images. By combining ImageBind, a model that embeds sound and image features into the same space, with a Bezier curve-based text representation, the style corresponding to the input sound is effectively transferred to the character images. Experiments demonstrated the successful generation of character images that reflect the sound-aligned style. Additional results with input sounds can also be viewed and heard at the following link: https://squ602.github.io/WaveFontStyler/

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

WaveFontStyler: Font Style Transfer Based on Sound

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kota Izumi,
  • Keiji Yanai

摘要

This study proposes a method for transferring a style aligned with the input sound to character images. By combining ImageBind, a model that embeds sound and image features into the same space, with a Bezier curve-based text representation, the style corresponding to the input sound is effectively transferred to the character images. Experiments demonstrated the successful generation of character images that reflect the sound-aligned style. Additional results with input sounds can also be viewed and heard at the following link: https://squ602.github.io/WaveFontStyler/