Development of Next-Generation Music Recognition Algorithm for Content Monitoring
Lead Participant:
BREATHE MUSIC LIMITED
Abstract
**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.
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.
Lead Participant | Project Cost | Grant Offer |
|---|---|---|
| BREATHE MUSIC LIMITED | £349,997 | £ 244,998 |
Participant |
||
| QUEEN MARY UNIVERSITY OF LONDON | £148,038 | £ 148,038 |
People |
ORCID iD |
| Jeff Francis (Project Manager) |