LISTEN is a joint laboratory launched by Télécom Paris with Valeo, Bruitparif and Music World Media, its aim is to develop adapted artificial intelligence systems in order to extract from sound signals (whatever their nature: music, speech, common environmental sounds, etc.) the fundamental characteristics that are useful for understanding them.
This fast-growing field has many applications, ranging from the analysis of complex acoustic scenes to the analysis of musical content (Music Information Retrieval or MIR), including predictive maintenance for industry.
The joint laboratory will focus its research work on five fundamental issues: frugal learning based on scarce data, multi-view, multi-task & distributed learning, model-based deep learning, self-supervised learning, and finally the learning of (deep) generative models.
The Télécom Paris team at the heart of the project includes the six members of the ADASP (Audio Data Analysis and Signal Processing) research group, which is part of the S²A (Signal, Statistics and Learning) affiliated with Telecom Paris’ in-house research laboratory: the LTCI. As Slim Essid, head of the joint laboratory, points out: « LISTEN intends to position itself at the cutting edge of research in machine listening at international level, with a high potential impact both for the industrial sector concerned and for public players. »
The laboratory has planned its work over a three-year cycle and aims to support several theses on the subject.
« The joint laboratory is one of the most successful forms of collaboration between Télécom Paris and players in the industrial world. LISTEN’s partners have understood that research is the fuel for innovation, and they dare to invest over long periods of time, particularly in research », stresses Sylvain Lamblot, Director of development and partnerships at Télécom Paris.