ViMuSe - a video-based AI music recommendation engine to improve creative efficiency and diversity.

ViMuSe - 基于视频的AI音乐推荐引擎，可提高创作效率和多样性。

• 批准号: 10104871
• 金额: $ 18.23万
• 依托单位: PYRSOS AI RESEARCH LTD
• 依托单位国家: 英国
• 项目类别: Collaborative R&D
• 财政年份: 2024
• 资助国家: 英国
• 起止时间: 2024 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=10104871
• 关键词: ViMuSe; video; based; AI; music

This project aims to develop an AI tool which will instantly recommend music tracks that would go well with a specific video clip (i.e. movie trailer). Our proposed recommendation engine will be able to dynamically recommend music tracks that are well-suited to the video content, even if the user does not have a specific music track in mind.This project unites Pyrsos' outstanding research capabilities with DAACI's existing music catalogue and software platform infrastructure to build cutting edge machine learning technology and make it available to a broad audience. It builds upon state-of-the art representation learning for multimedia and advances it by training a neural network that can map video and music excerpts into a shared semantic space. Extensive user testing will be conducted during the project to evaluate the technology and provide feedback.

该项目旨在开发一种人工智能工具，可以立即推荐与特定视频剪辑（即电影预告片）相匹配的音乐曲目。我们提出的推荐引擎将能够动态地推荐非常适合视频内容的音乐曲目，即使用户心中没有特定的音乐曲目。该项目将Pyrsos卓越的研究能力与DAACI现有的音乐目录和软件平台基础设施结合起来，构建尖端的机器学习技术，并将其提供给广泛的受众。它建立在最先进的多媒体表示学习的基础上，并通过训练一个可以将视频和音乐摘录映射到共享语义空间的神经网络来推进它。在项目期间将进行广泛的用户测试，以评估技术并提供反馈。