Musical Instrument Identification using Multiscale Mel-frequency Cepstral Coefficients

Bob L. Sturm, Marcela Morvidone, Laurent Daudet

Publikation: Bidrag til tidsskriftKonferenceartikel i tidsskriftForskningpeer review

16 Citationer (Scopus)
520 Downloads (Pure)

Resumé

We investigate the benefits of evaluating
Mel-frequency cepstral coefficients (MFCCs) over several time scales
in the context of automatic musical instrument identification
for signals that are monophonic but derived from real musical settings.
We define several sets of features derived from MFCCs
computed using multiple time resolutions,
and compare their performance against other features
that are computed using a single time resolution,
such as MFCCs, and derivatives of MFCCs.
We find that in each task --- pairwise discrimination,
and one vs. all classification --- the features involving
multiscale decompositions perform significantly better than
features computed using a single time-resolution.
OriginalsprogEngelsk
TidsskriftProceedings of the European Signal Processing Conference
Sider (fra-til)477-481
Antal sider5
ISSN2076-1465
StatusUdgivet - 2010
BegivenhedEUSIPCO 2010 - Aalborg, Danmark
Varighed: 23 aug. 2010 → …

Konference

KonferenceEUSIPCO 2010
LandDanmark
ByAalborg
Periode23/08/2010 → …

Fingerprint

Musical instruments
Derivatives
Decomposition

Citer dette

@inproceedings{96ce2be9d56f43748b0ea99b64ce65f2,
title = "Musical Instrument Identification using Multiscale Mel-frequency Cepstral Coefficients",
abstract = "We investigate the benefits of evaluating Mel-frequency cepstral coefficients (MFCCs) over several time scalesin the context of automatic musical instrument identificationfor signals that are monophonic but derived from real musical settings.We define several sets of features derived from MFCCs computed using multiple time resolutions,and compare their performance against other featuresthat are computed using a single time resolution,such as MFCCs, and derivatives of MFCCs.We find that in each task --- pairwise discrimination, and one vs. all classification --- the features involvingmultiscale decompositions perform significantly better thanfeatures computed using a single time-resolution.",
author = "Sturm, {Bob L.} and Marcela Morvidone and Laurent Daudet",
year = "2010",
language = "English",
pages = "477--481",
journal = "Proceedings of the European Signal Processing Conference",
issn = "2076-1465",
publisher = "European Association for Signal Processing (EURASIP)",

}

Musical Instrument Identification using Multiscale Mel-frequency Cepstral Coefficients. / Sturm, Bob L.; Morvidone, Marcela; Daudet, Laurent.

I: Proceedings of the European Signal Processing Conference, 2010, s. 477-481.

Publikation: Bidrag til tidsskriftKonferenceartikel i tidsskriftForskningpeer review

TY - GEN

T1 - Musical Instrument Identification using Multiscale Mel-frequency Cepstral Coefficients

AU - Sturm, Bob L.

AU - Morvidone, Marcela

AU - Daudet, Laurent

PY - 2010

Y1 - 2010

N2 - We investigate the benefits of evaluating Mel-frequency cepstral coefficients (MFCCs) over several time scalesin the context of automatic musical instrument identificationfor signals that are monophonic but derived from real musical settings.We define several sets of features derived from MFCCs computed using multiple time resolutions,and compare their performance against other featuresthat are computed using a single time resolution,such as MFCCs, and derivatives of MFCCs.We find that in each task --- pairwise discrimination, and one vs. all classification --- the features involvingmultiscale decompositions perform significantly better thanfeatures computed using a single time-resolution.

AB - We investigate the benefits of evaluating Mel-frequency cepstral coefficients (MFCCs) over several time scalesin the context of automatic musical instrument identificationfor signals that are monophonic but derived from real musical settings.We define several sets of features derived from MFCCs computed using multiple time resolutions,and compare their performance against other featuresthat are computed using a single time resolution,such as MFCCs, and derivatives of MFCCs.We find that in each task --- pairwise discrimination, and one vs. all classification --- the features involvingmultiscale decompositions perform significantly better thanfeatures computed using a single time-resolution.

M3 - Conference article in Journal

SP - 477

EP - 481

JO - Proceedings of the European Signal Processing Conference

JF - Proceedings of the European Signal Processing Conference

SN - 2076-1465

ER -