Title: AI-generated music detection at Deezer

Romain Hennekin, head of research at Deezer

Abstract:

In the last few years, AI music generation models have made it possible to generate realistic music in one click. Anyone can now be a "music artist"... who can massively deliver their art to music platforms.

As this new kind of generated content can possibly compete with real music artists and dilute their revenue, presents a risk of personality rights and copyright infringement, and can be used to flood music platforms, the music industry needs to monitor such content and develop tools to detect AI-generated music tracks.

This talk will give a quick overview of AI-based music generation systems and the threats they pose to the music industry, and present recent deezer works about AI-generated music detection, the current challenges, and the ones to come.

Bio:

Romain Hennequin is head of research at Deezer. With over 10 years of experience in industrial research, Romain has addressed various topics in the fields of Artificial Intelligence ranging from Source Separation to Music Information Retrieval, Recommender Systems, or Graph Mining. He has been involved in several Music-related open-source projects such as the Source Separation library Spleeter.

At Deezer, Romain currently leads a team of researchers working on Automatic Music Description, User Behavior Understanding and Interaction in Natural Language amongst other topics.

Romain graduated in Computer Science from Ecole Polytechnique, UPMC (now Sorbonne Université), and Telecom Paris and earned a PhD in signal processing from Telecom Paris.

After the seminar, Pablo Alonso and Dmitry Bogdanov will present the Essentia Models News:

In this short talk they will review the latest updates related to the development of Essentia. They will cover an overview of the last models integrated into the library, the current research project developed in the BSC infrastructure on self-supervised representation learning, and the evolution on the development of the Essentia API.

Activity in the frame of:

Cátedra UPF-BMAT en Inteligencia Articial y Música (TSI-100929-2023-1). Project funded by Secretaría de Estado de Digitalización e Inteligencia Artificial, and Unión Europea-Next Generation EU