Voice adaptive speech translation for video content

视频内容的语音自适应语音翻译

• 批准号: 104698
• 金额: $ 32.83万
• 依托单位: PAPERCUP TECHNOLOGIES LIMITED
• 依托单位国家: 英国
• 项目类别: Feasibility Studies
• 财政年份: 2019
• 资助国家: 英国
• 起止时间: 2019 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=104698
• 关键词: Voice; adaptive; speech; translation; video

"There are hundreds of millions of TV shows, movies, and digital videos across VoD, TV catch-up services and the web. However, we estimate that 99% of all that video content is shackled to a single language. Papercup provides a unique solution to automatically generate new vocal content in other languages, while maintaining the qualities of the original speaker's voice in a new language. Papercup aims to be the de-facto solution for translating video content, and to unshackle the hundreds of millions of videos that exist in the world today so that all of the world's content can be available in any language.A key differentiator for content creators is their personality and the way in which they articulate themselves. If creators were to use a generic voice from, e.g., Google or Microsoft to create content in other languages, they would be unable to differentiate themselves effectively. In other words, there are 15 million videos uploaded to the major distributions channels every day; if all of these creators used the same voices, they would essentially all sound the same.Over the past few years, video platforms have grown at an accelerated rate (there are over 200 video on-demand platforms in the US alone). Using Papercup's technology, creators of videos, whether a YouTube clip or Hollywood film, will be able to automatically reproduce their content in any language, while retaining the actors' or speakers' voice characteristics across any platform. By contrast, tech giants, which typically have their own video platforms (e.g., Amazon Video, Google's YouTube), cannot offer creators the same value proposition of being platform-agnostic."

“VoD、电视回放服务和网络上有数亿部电视节目、电影和数字视频。然而，我们估计 99% 的视频内容都受限于单一语言。Papercup 提供了一种独特的解决方案，可以自动生成其他语言的新声音内容，同时保持新语言中原始说话者的声音质量。Papercup 旨在成为翻译视频内容的事实上的解决方案，并摆脱束缚。 当今世界上存在数以亿计的视频，因此世界上所有的内容都可以以任何语言提供。内容创作者的一个关键区别在于他们的个性和表达自己的方式。如果创作者使用谷歌或微软等公司的通用声音来创建其他语言的内容，他们将无法有效地使自己脱颖而出。也就是说，每天有1500万个视频上传到各大发行渠道；如果全部 如果这些创作者使用相同的声音，那么他们听起来基本上都是一样的。 在过去的几年里，视频平台加速增长（仅在美国就有超过 200 个视频点播平台）。使用 Papercup 的技术，视频创作者（无论是 YouTube 剪辑还是好莱坞电影）将能够自动以任何语言复制其内容，同时在任何平台上保留演员或演讲者的声音特征。由 相比之下，科技巨头通常拥有自己的视频平台（例如亚马逊视频、谷歌的 YouTube），无法为创作者提供与平台无关的相同价值主张。”