Development of Next-Generation Music Recognition Algorithm for Content Monitoring

**视觉**版权法要求音乐行业跟踪世界上每一首歌曲和每一张唱片的每一次公开表演，以收集和准确分配版税。然而，准确的跟踪只出现在少数场景中。这个数字化转型项目将开发一种新的音乐识别技术(MRT)解决方案，方法是调用新的尖端人工智能技术；下一代MRT算法，能够识别在现场播放或表演的音乐，或非官方在线封面(例如On_YouTube_)。这将反过来允许词曲作者和音乐所有者获得应得的版税。**主要目标**开发一种解决方案，使用“窄”人工智能技术来识别歌曲的替代版本，而不考虑歌曲的播放风格。目前的解决方案擅长识别艺术家的音乐原创录音，还可以识别已注册的唱片封面版本。例如，可以使用类似于Shazam的解决方案来识别《嘿Jude》的录音，只要该录音的“音乐指纹”已经被注册。然而，这些现有技术在大多数情况下无法识别音乐家和表演者在全国数千个现场音乐场所(如酒吧、俱乐部和咖啡馆)进行的现场音乐。其目标是确保现场场地和音乐表演者拥有一种简单、非侵入性的自动化解决方案，以确定应向音乐所有者支付的版税。由于场地上的音乐不被识别，收集协会只估计谁欠什么；特别是规模较小的音乐创作者永远得不到他们有权获得的版税。**主要关注领域**开发一种创新的MRT解决方案，以解决当前音乐识别算法无法解决的领域，即识别歌曲中的版权所有者的内容。这个项目的动机是开发一个MVP解决方案，并定义如何最好地将其推向市场。**它是如何创新的？**_呼吸音乐的音乐识别引擎利用人工智能来开发一种变化莫测的、破坏性的MRT作为识别模型，以取代像_Shazam这样的现有‘录音’监控系统--能够监控和识别录音中的‘录音’和‘内容’。这一创新克服了只能识别和匹配以前“看过”的录制版本的限制。

**Vision**Copyright law requires that the music industry should track every public performance of every song and recording in the world to collect and accurately distribute royalties. However, accurate tracking only occurs in a minority of scenarios.This digital transformational project will develop a novel Music Recognition Technology (MRT) solution by evoking new, cutting-edge AI technologies; a next-generation MRT algorithm to enable identification of music being played or performed in a live venue, or unofficial online covers (e.g. on _YouTube_). This will in turn allow songwriters & music owners to gain the royalties due to them.**Key objectives**To develop a solution using 'narrow' AI technologies to recognise alternative versions of a song irrespective of the style in which it is being played. Current solutions are good at recognising original recordings of music by the artist and also registered recorded cover versions. For example, a recording of "Hey Jude" could be identified using _Shazam_-like solutions as long as a "music-fingerprint" of that recording has been registered. However, these existing technologies cannot in most instances identify live music undertaken by musicians & performers in the thousands of live music venues such as pubs, clubs and cafes across the land. The objective is to ensure that live venues and music performers have an easy, non-intrusive automated solution to identify the royalties due to the music owners. As a result of music in venues not getting recognised, collection societies only estimate who is owed what; the smaller music creators especially are the ones who never get the royalties they are entitled to.**Main areas of focus**On the development of an innovative MRT solution to address the areas where current music recognition algorithms are unable to address, i.e. to identify the rights-holders 'content' within the song. The motivation of this project is to develop an MVP solution and define how this could be best taken to market.**How it is innovative?**_Breathe Music's_ music recognition engine utilises AI to develop a step-change, disruptive MRT as an identification model to replace existing 'recording' monitoring systems like _Shazam_ - capable of monitoring and identifying the 'recording' but also the 'content' within the recording. This innovation overcomes the restrictions of only being able to identify and match only previously "seen" recorded versions.