Downbeat tracking with multiple features and deep neural networks

Abstract :

In this paper, we introduce a novel method for the automatic estimation of downbeat positions from music signals. Our system relies on the computation of musically inspired features capturing important aspects of music such as timbre, harmony, rhythmic patterns, or local similarities in both timbre and harmony. It then uses several independent deep neural networks to learn higher-level representations. The downbeat sequences are finally obtained thanks to a temporal decoding step based on the Viterbi algorithm. The comparative evaluation conducted on varied datasets demonstrates the efficiency and robustness across different music styles of our approach.

Complete list of metadatas

https://hal.telecom-paristech.fr/hal-02287003
Contributor : Telecomparis Hal <>
Submitted on : Friday, September 13, 2019 - 4:27:45 PM
Last modification on : Sunday, September 15, 2019 - 1:11:40 AM

Identifiers

  • HAL Id : hal-02287003, version 1

Citation

Simon Durand, Juan P. Bello, Bertrand David, Gael Richard. Downbeat tracking with multiple features and deep neural networks. ICASSP 2015, Apr 2015, Brisbane, Australia. ⟨hal-02287003⟩

Share

Metrics

Record views

4