The localization of beat instants is often based on some low-level rhythmic feature extracted from the audio signal (usually an onset detection function), as well as the tempo path that is estimated from it. Going from such descriptors to a sequence of beat instants requires a tracking strategy that models how the beat evolves and offers a good trade-off between stability and responsivity to changes in timing. In the literature this is often performed using dynamic programming methods. In this article we focus on this latter stage of beat tracking and propose a novel strategy based on an efficient generation and joint steering of multiple trackers (paths). This solution is shown to lead to improved computational efficiency with respect to dynamic programming methods, as confirmed by a first set of experiments. In a second set of experiments the proposed method is compared with a broader set of state-of-the-art solutions, though relying on different rhythmic descriptors and beat-tracking strategies, in order to offer a more general assessment of our solution.

Multipath beat tracking

DI GIORGI, BRUNO;ZANONI, MASSIMILIANO;SARTI, AUGUSTO
2016-01-01

Abstract

The localization of beat instants is often based on some low-level rhythmic feature extracted from the audio signal (usually an onset detection function), as well as the tempo path that is estimated from it. Going from such descriptors to a sequence of beat instants requires a tracking strategy that models how the beat evolves and offers a good trade-off between stability and responsivity to changes in timing. In the literature this is often performed using dynamic programming methods. In this article we focus on this latter stage of beat tracking and propose a novel strategy based on an efficient generation and joint steering of multiple trackers (paths). This solution is shown to lead to improved computational efficiency with respect to dynamic programming methods, as confirmed by a first set of experiments. In a second set of experiments the proposed method is compared with a broader set of state-of-the-art solutions, though relying on different rhythmic descriptors and beat-tracking strategies, in order to offer a more general assessment of our solution.
2016
AES
Music; Engineering (all)
File in questo prodotto:
File Dimensione Formato  
JAES-D-16-00024HR.pdf

Accesso riservato

: Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione 977.75 kB
Formato Adobe PDF
977.75 kB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/1003014
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 13
  • ???jsp.display-item.citation.isi??? 7
social impact