The effect of key and tempo on audio onset detection using machine learning techniques

A sensitivity analysis

Ching-Hua Chuan, Elaine Chew

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In this paper, we explore the effect of musical context on audio onset detection using machine learning techniques. We extract the signal intensity and frequency energy of audio as the attributes of input instances for the machine learning techniques. The audio is synthesized from MIDI files, providing exact information of onset events. We test three state-of-the-art machine learning algorithms, Support Vector Machines (SVM), Neural Networks (NN), and Naïve Bayes (NB) with Ada boosting, for learning and classifying audio onsets. We found that SVMs perform best in general, based on the average of training and 10-fold cross validation errors as the evaluation criterion. We then test the SVM and NN, the two best performing methods, on Bach's Prelude in C major BWV 943, transposed to different keys and time-stretched to various tempi. The error rates ranged from 23.91% (when training set key and tempo equals those of the test set) to 37.22% (when the key is off by four accidentals) and 37.91% (when tempo is 20 beats per minute faster). The results show that audio onset detection performs significantly better when the key and tempo attributes of the test and training sets concur, than when they are different, thus supporting the utility of tempo and key knowledge in designing onset detection systems, or in prescribing confidence statistics to onset detection outcomes.

Original languageEnglish (US)
Title of host publicationISM 2006 - 8th IEEE International Symposium on Multimedia
Pages805-810
Number of pages6
DOIs
StatePublished - Dec 1 2006
Externally publishedYes
EventISM 2006 - 8th IEEE International Symposium on Multimedia - San Diego, CA, United States
Duration: Dec 11 2006Dec 13 2006

Other

OtherISM 2006 - 8th IEEE International Symposium on Multimedia
CountryUnited States
CitySan Diego, CA
Period12/11/0612/13/06

Fingerprint

Sensitivity analysis
Learning systems
Support vector machines
Neural networks
Learning algorithms
Statistics

ASJC Scopus subject areas

  • Computer Networks and Communications

Cite this

Chuan, C-H., & Chew, E. (2006). The effect of key and tempo on audio onset detection using machine learning techniques: A sensitivity analysis. In ISM 2006 - 8th IEEE International Symposium on Multimedia (pp. 805-810). [4061263] https://doi.org/10.1109/ISM.2006.149

The effect of key and tempo on audio onset detection using machine learning techniques : A sensitivity analysis. / Chuan, Ching-Hua; Chew, Elaine.

ISM 2006 - 8th IEEE International Symposium on Multimedia. 2006. p. 805-810 4061263.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Chuan, C-H & Chew, E 2006, The effect of key and tempo on audio onset detection using machine learning techniques: A sensitivity analysis. in ISM 2006 - 8th IEEE International Symposium on Multimedia., 4061263, pp. 805-810, ISM 2006 - 8th IEEE International Symposium on Multimedia, San Diego, CA, United States, 12/11/06. https://doi.org/10.1109/ISM.2006.149
Chuan, Ching-Hua ; Chew, Elaine. / The effect of key and tempo on audio onset detection using machine learning techniques : A sensitivity analysis. ISM 2006 - 8th IEEE International Symposium on Multimedia. 2006. pp. 805-810
@inproceedings{b2864a6c30a44226abbcf8bee71ae7eb,
title = "The effect of key and tempo on audio onset detection using machine learning techniques: A sensitivity analysis",
abstract = "In this paper, we explore the effect of musical context on audio onset detection using machine learning techniques. We extract the signal intensity and frequency energy of audio as the attributes of input instances for the machine learning techniques. The audio is synthesized from MIDI files, providing exact information of onset events. We test three state-of-the-art machine learning algorithms, Support Vector Machines (SVM), Neural Networks (NN), and Na{\"i}ve Bayes (NB) with Ada boosting, for learning and classifying audio onsets. We found that SVMs perform best in general, based on the average of training and 10-fold cross validation errors as the evaluation criterion. We then test the SVM and NN, the two best performing methods, on Bach's Prelude in C major BWV 943, transposed to different keys and time-stretched to various tempi. The error rates ranged from 23.91{\%} (when training set key and tempo equals those of the test set) to 37.22{\%} (when the key is off by four accidentals) and 37.91{\%} (when tempo is 20 beats per minute faster). The results show that audio onset detection performs significantly better when the key and tempo attributes of the test and training sets concur, than when they are different, thus supporting the utility of tempo and key knowledge in designing onset detection systems, or in prescribing confidence statistics to onset detection outcomes.",
author = "Ching-Hua Chuan and Elaine Chew",
year = "2006",
month = "12",
day = "1",
doi = "10.1109/ISM.2006.149",
language = "English (US)",
isbn = "0769527469",
pages = "805--810",
booktitle = "ISM 2006 - 8th IEEE International Symposium on Multimedia",

}

TY - GEN

T1 - The effect of key and tempo on audio onset detection using machine learning techniques

T2 - A sensitivity analysis

AU - Chuan, Ching-Hua

AU - Chew, Elaine

PY - 2006/12/1

Y1 - 2006/12/1

N2 - In this paper, we explore the effect of musical context on audio onset detection using machine learning techniques. We extract the signal intensity and frequency energy of audio as the attributes of input instances for the machine learning techniques. The audio is synthesized from MIDI files, providing exact information of onset events. We test three state-of-the-art machine learning algorithms, Support Vector Machines (SVM), Neural Networks (NN), and Naïve Bayes (NB) with Ada boosting, for learning and classifying audio onsets. We found that SVMs perform best in general, based on the average of training and 10-fold cross validation errors as the evaluation criterion. We then test the SVM and NN, the two best performing methods, on Bach's Prelude in C major BWV 943, transposed to different keys and time-stretched to various tempi. The error rates ranged from 23.91% (when training set key and tempo equals those of the test set) to 37.22% (when the key is off by four accidentals) and 37.91% (when tempo is 20 beats per minute faster). The results show that audio onset detection performs significantly better when the key and tempo attributes of the test and training sets concur, than when they are different, thus supporting the utility of tempo and key knowledge in designing onset detection systems, or in prescribing confidence statistics to onset detection outcomes.

AB - In this paper, we explore the effect of musical context on audio onset detection using machine learning techniques. We extract the signal intensity and frequency energy of audio as the attributes of input instances for the machine learning techniques. The audio is synthesized from MIDI files, providing exact information of onset events. We test three state-of-the-art machine learning algorithms, Support Vector Machines (SVM), Neural Networks (NN), and Naïve Bayes (NB) with Ada boosting, for learning and classifying audio onsets. We found that SVMs perform best in general, based on the average of training and 10-fold cross validation errors as the evaluation criterion. We then test the SVM and NN, the two best performing methods, on Bach's Prelude in C major BWV 943, transposed to different keys and time-stretched to various tempi. The error rates ranged from 23.91% (when training set key and tempo equals those of the test set) to 37.22% (when the key is off by four accidentals) and 37.91% (when tempo is 20 beats per minute faster). The results show that audio onset detection performs significantly better when the key and tempo attributes of the test and training sets concur, than when they are different, thus supporting the utility of tempo and key knowledge in designing onset detection systems, or in prescribing confidence statistics to onset detection outcomes.

UR - http://www.scopus.com/inward/record.url?scp=46249111659&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=46249111659&partnerID=8YFLogxK

U2 - 10.1109/ISM.2006.149

DO - 10.1109/ISM.2006.149

M3 - Conference contribution

SN - 0769527469

SN - 9780769527468

SP - 805

EP - 810

BT - ISM 2006 - 8th IEEE International Symposium on Multimedia

ER -