| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 112.7 KB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
The purpose of this paper is to present a method for the classification of musical pieces based on a language modeling
approach. The method does not require any metadata and is used with raw audio format. It consists in 1) transforming music data into a sequence of symbols 2) building a model for each category by estimating n-grams
from the sequences of symbols derived from the training set.
The results obtained on three audio datasets show that, providing the amount of data is sufficient for estimating the transitions
probabilities of the model, the approach performs very well. The performance achieved
with the ISMIR 2004 Genre classification dataset is, to our knowledge, one of the best published in the literature.
Description
Keywords
Machine Learning Music Information Retrieval