Features for audio and music classification

By Martin F. McKinney, Jeroen Breebaart. International Society for Music Information Retrieval Conference, 2003

418 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Predicting the Geographical Origin of Music (2014, background): “The most straightforward way of representing geographical position is via latitude and 1550-4786/14 $31.00 © 2014 IEEE DOI 10.1109/ICDM.2014.73 1115 longitude.”
Speaker identification and clustering using convolutional neural networks (2016, background): “Despite being the state-of-the-art approach and well-working industry standard, this approach in principle has major shortcomings: Using MFCC feature vectors, the all-purpose answer for all audio analysis tasks [8], no specific voice-related characteristics of the speech signal despite the gross spe”
Audio-visual classification and detection of human manipulation actions (2014, methodology): “of work where they have been used to classify a broad set of different sound classes [35], [40].”
A Large-Scale Evaluation of Acoustic and Subjective Music-Similarity Measures (2004, background): “Although many features have been proposed for music analysis, such as spectral centroid, band-width, loudness, and sharpness (McKinney and Breebaart 2003), in this article we concentrate on features derived from Mel-Frequency Cepstral Co-ef(cid:142) cients (MFCCs).”
Semantic Annotation and Retrieval of Music and Sound Effects (2008, methodology): “We compute dynamic Mel-frequency cepstral coefficients (dMFCCs) from each half-overlapping, mediumtime (∼743 msec) segment of music audio[8].”