TechCrunch记者亲身参与了一项AI视频生成试验，由创业公司Synthesia为其创建了个人数字虚拟形象。1记者前往Synthesia位于纽约的办公室，通过提供照片和语音样本，Synthesia团队在数天内完成了多个虚拟形象的制作。1这些虚拟形象包括个人虚拟头像和能与用户进行互动的虚拟分身，这也是Synthesia首次为记者创建此类数字虚拟形象。1
该交互式虚拟形象采用了语音转文本、代理语言模型、文本转语音和视频模型等多种技术组件。1该虚拟形象经过特殊训练，仅能回答关于记者撰写的风险投资欺诈故事的问题。1这次体验引发了记者对AI虚拟形象在新闻业和企业沟通中的潜在应用前景，以及相关伦理问题的深入思考。1
Synthesia今年早些时候达到了40亿美元的估值，去年年度经常性收入超过1亿美元。1
A TechCrunch reporter visited Synthesia's New York office to create a personal digital avatar powered by artificial intelligence, marking the first time the AI video generation startup has built such a representation for a journalist 1. The process involved providing photographs and voice samples, which Synthesia's team used to construct multiple virtual personas, including a static avatar and an interactive version capable of engaging with users 1.
The interactive avatar was engineered using a combination of speech-to-text technology, a language model agent, text-to-speech conversion, and video generation—taking several days for Synthesia's team to complete 1. The resulting virtual representation was trained specifically to answer questions about the reporter's coverage of venture capital fraud, demonstrating a focused application of the technology 1. This development comes as Synthesia continues its rapid growth, having achieved a $4 billion valuation earlier this year and generating over $100 million in annual recurring revenue last year 1.
评论
还没有评论，欢迎留下第一条。