Skip to Main content Skip to Navigation
Master thesis

Clustering incrémental de signaux audio

Maxime Sirbu 1, 2, 3
1 MuTant - Synchronous Realtime Processing and Programming of Music Signals
Inria Paris-Rocquencourt, UPMC - Université Pierre et Marie Curie - Paris 6, IRCAM, CNRS - Centre National de la Recherche Scientifique
2 Repmus - Représentations musicales
STMS - Sciences et Technologies de la Musique et du Son
Abstract : This report aims to study different methods of online clustering, mainly applied to audio signals. We will first detail the state-of-the art algorithms for clustering, as well as the theory behind them. Then we will extend this methods to incremental clustering, and present different online algorithms. These algorithms are based on the hidden markov models, which are classic art representations of data and hidden states in signal processing, and hidden semi-markov models, which extend them to a semi-markov representation of the states. We present this within the context of audio segmentation – the task of segmenting audio sources in homogenous chunks – and classification – the task of identifying these chunks – applied to audio event detection. We will also set an experimental protocol, with a view to evaluate them and compare the result to a state-of-the art algorithm for the same task.
Complete list of metadata

Cited literature [36 references]  Display  Hide  Download

https://hal.inria.fr/hal-01196455
Contributor : Maxime Sirbu <>
Submitted on : Wednesday, September 9, 2015 - 6:09:41 PM
Last modification on : Friday, January 8, 2021 - 2:04:38 PM
Long-term archiving on: : Monday, December 28, 2015 - 11:24:43 PM

Identifiers

  • HAL Id : hal-01196455, version 1

Citation

Maxime Sirbu. Clustering incrémental de signaux audio. Apprentissage [cs.LG]. 2015. ⟨hal-01196455⟩

Share

Metrics

Record views

323

Files downloads

295