Hélène Papadopoulos was born in Paris, France, in 1983. She graduated in 2006 from ENSEA (École Nationale Supérieure de l'Électronique et de ses Applications), a leading French Engineering School specialized in Electronics, Signal Processing, Computer Science and Communication. The same year, she received the M.Sc. degree in Image, Signal Processing and
Artificial Intelligence from the University of Cergy-Pontoise. She received her Ph.D. in computer science from the University Paris VI (IRCAM), France in 2010. She is currently a research and teaching fellow at the University Paris Sud/Supelec, in the Laboratory of Signals and Systems. In parallel to her scientific studies, she pursues musical studies. Her research interests include signal processing, machine learning, music perception and cognition, music content processing, classification and music information retrieval.

AG 2, AG 4, MMCI   Info

AG Audience

English

60 Minutes

Saarbrücken

Musical signals are highly structured in terms of harmony and rhythm. We present a new technique for the joint estimation of the chord progression, the downbeats and the local keys from an audio file. We intend to show that integrating knowledge of mutual dependencies between chords, key and metrical structure allows enhancing the estimation of these musical attributes. For this, we propose a specific topology of hidden Markov models that enables modelling chord dependency on metrical structure and musical key. This model allows us to consider pieces with complex metrical structures such as beat addition, beat deletion or changes in the meter. It is evaluated on a set of popular and classical music pieces that present various metrical and tonal structures.

Thorsten Thormählen (for Meinard Müller)

+49 681 9325-417

--email hidden

passcode not visible

logged in users only