Consider recording the media before real-time processing.
Scan the object via taobao streaming and make it dance.
Transplant lolita pictures to bilibili.
Share dialogs/info from soul/qq/wechat.
Repurpose a wide range of streaming platforms.
use ocr to filter out text info
find new title from danmaku or comments
我们都知道,视频主要由画面和音频组成。但还有一个元素同样包含了巨大的信息量 —— 字幕。结合现有的自然语言处理模型,我们便能实现对话抽取式的自动剪辑。PS: 软件图标是我妹妹Hannah设计滴~
项目开源地址:https://github.com/COLOR-SKY/DialogueExtractor
学业之余我会陆续更新工具教程。