Haar-like filtering based speech detection using integral signal for sensornet

Jun Nishimura, Tadahiro Kuroda

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

5 Citations (Scopus)

Abstract

Speech detection using haar-like filtering is proposed as a new and very low calculation cost method for sensornet applications. The simple haar-like filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. To further decrease the calculation cost, the use of intermediate signal representation called "integral signal" is proposed. Our method yielded speech/nonspeech classification accuracy of 97.44% for the input length of 0.1s. Compared with high performance feature extraction method MFCC (Mel-Frequency Cepstrum Coefficient), the proposed haar-like filtering can be approximately 93.71% efficient in terms of the total amount of add and multiply calculations while capable of achieving the error rate of only 2.56% relative to MFCC.

Original languageEnglish
Title of host publicationProceedings of the 3rd International Conference on Sensing Technology, ICST 2008
Pages52-56
Number of pages5
DOIs
Publication statusPublished - 2008
Event3rd International Conference on Sensing Technology, ICST 2008 - Tainan, Taiwan, Province of China
Duration: 2008 Nov 302008 Dec 3

Other

Other3rd International Conference on Sensing Technology, ICST 2008
CountryTaiwan, Province of China
CityTainan
Period08/11/3008/12/3

Fingerprint

Feature extraction
Costs

Keywords

  • Haar-like filtering
  • Integral signal
  • Sensornet
  • Speech detection

ASJC Scopus subject areas

  • Software
  • Electrical and Electronic Engineering

Cite this

Nishimura, J., & Kuroda, T. (2008). Haar-like filtering based speech detection using integral signal for sensornet. In Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008 (pp. 52-56). [4757072] https://doi.org/10.1109/ICSENST.2008.4757072

Haar-like filtering based speech detection using integral signal for sensornet. / Nishimura, Jun; Kuroda, Tadahiro.

Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008. 2008. p. 52-56 4757072.

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

Nishimura, J & Kuroda, T 2008, Haar-like filtering based speech detection using integral signal for sensornet. in Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008., 4757072, pp. 52-56, 3rd International Conference on Sensing Technology, ICST 2008, Tainan, Taiwan, Province of China, 08/11/30. https://doi.org/10.1109/ICSENST.2008.4757072
Nishimura J, Kuroda T. Haar-like filtering based speech detection using integral signal for sensornet. In Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008. 2008. p. 52-56. 4757072 https://doi.org/10.1109/ICSENST.2008.4757072
Nishimura, Jun ; Kuroda, Tadahiro. / Haar-like filtering based speech detection using integral signal for sensornet. Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008. 2008. pp. 52-56
@inproceedings{dc9e4356ea0440bf93ce464884505347,
title = "Haar-like filtering based speech detection using integral signal for sensornet",
abstract = "Speech detection using haar-like filtering is proposed as a new and very low calculation cost method for sensornet applications. The simple haar-like filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. To further decrease the calculation cost, the use of intermediate signal representation called {"}integral signal{"} is proposed. Our method yielded speech/nonspeech classification accuracy of 97.44{\%} for the input length of 0.1s. Compared with high performance feature extraction method MFCC (Mel-Frequency Cepstrum Coefficient), the proposed haar-like filtering can be approximately 93.71{\%} efficient in terms of the total amount of add and multiply calculations while capable of achieving the error rate of only 2.56{\%} relative to MFCC.",
keywords = "Haar-like filtering, Integral signal, Sensornet, Speech detection",
author = "Jun Nishimura and Tadahiro Kuroda",
year = "2008",
doi = "10.1109/ICSENST.2008.4757072",
language = "English",
isbn = "9781424421770",
pages = "52--56",
booktitle = "Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008",

}

TY - GEN

T1 - Haar-like filtering based speech detection using integral signal for sensornet

AU - Nishimura, Jun

AU - Kuroda, Tadahiro

PY - 2008

Y1 - 2008

N2 - Speech detection using haar-like filtering is proposed as a new and very low calculation cost method for sensornet applications. The simple haar-like filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. To further decrease the calculation cost, the use of intermediate signal representation called "integral signal" is proposed. Our method yielded speech/nonspeech classification accuracy of 97.44% for the input length of 0.1s. Compared with high performance feature extraction method MFCC (Mel-Frequency Cepstrum Coefficient), the proposed haar-like filtering can be approximately 93.71% efficient in terms of the total amount of add and multiply calculations while capable of achieving the error rate of only 2.56% relative to MFCC.

AB - Speech detection using haar-like filtering is proposed as a new and very low calculation cost method for sensornet applications. The simple haar-like filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. To further decrease the calculation cost, the use of intermediate signal representation called "integral signal" is proposed. Our method yielded speech/nonspeech classification accuracy of 97.44% for the input length of 0.1s. Compared with high performance feature extraction method MFCC (Mel-Frequency Cepstrum Coefficient), the proposed haar-like filtering can be approximately 93.71% efficient in terms of the total amount of add and multiply calculations while capable of achieving the error rate of only 2.56% relative to MFCC.

KW - Haar-like filtering

KW - Integral signal

KW - Sensornet

KW - Speech detection

UR - http://www.scopus.com/inward/record.url?scp=63049090172&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=63049090172&partnerID=8YFLogxK

U2 - 10.1109/ICSENST.2008.4757072

DO - 10.1109/ICSENST.2008.4757072

M3 - Conference contribution

AN - SCOPUS:63049090172

SN - 9781424421770

SP - 52

EP - 56

BT - Proceedings of the 3rd International Conference on Sensing Technology, ICST 2008

ER -