Low cost speech detection using Haar-like filtering for sensornet

Jun Nishimura, Tadahiro Kuroda

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

13 Citations (Scopus)

Abstract

Haar-like filtering based speech detection is proposed as a new and very low calculation cost method for sensornet applications. The simple haarlike filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. Our method yielded speech/nonspeech classification accuracy of 96.93% 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 85.77% efficient in terms of the amount of add and multiply calculations while capable of achieving the error rate of only 3.03% relative to MFCC.

Original languageEnglish
Title of host publicationInternational Conference on Signal Processing Proceedings, ICSP
Pages2608-2611
Number of pages4
DOIs
Publication statusPublished - 2008
Event2008 9th International Conference on Signal Processing, ICSP 2008 - Beijing, China
Duration: 2008 Oct 262008 Oct 29

Other

Other2008 9th International Conference on Signal Processing, ICSP 2008
CountryChina
CityBeijing
Period08/10/2608/10/29

Fingerprint

Costs
Feature extraction

ASJC Scopus subject areas

  • Signal Processing
  • Software
  • Computer Science Applications

Cite this

Nishimura, J., & Kuroda, T. (2008). Low cost speech detection using Haar-like filtering for sensornet. In International Conference on Signal Processing Proceedings, ICSP (pp. 2608-2611). [4697683] https://doi.org/10.1109/ICOSP.2008.4697683

Low cost speech detection using Haar-like filtering for sensornet. / Nishimura, Jun; Kuroda, Tadahiro.

International Conference on Signal Processing Proceedings, ICSP. 2008. p. 2608-2611 4697683.

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

Nishimura, J & Kuroda, T 2008, Low cost speech detection using Haar-like filtering for sensornet. in International Conference on Signal Processing Proceedings, ICSP., 4697683, pp. 2608-2611, 2008 9th International Conference on Signal Processing, ICSP 2008, Beijing, China, 08/10/26. https://doi.org/10.1109/ICOSP.2008.4697683
Nishimura J, Kuroda T. Low cost speech detection using Haar-like filtering for sensornet. In International Conference on Signal Processing Proceedings, ICSP. 2008. p. 2608-2611. 4697683 https://doi.org/10.1109/ICOSP.2008.4697683
Nishimura, Jun ; Kuroda, Tadahiro. / Low cost speech detection using Haar-like filtering for sensornet. International Conference on Signal Processing Proceedings, ICSP. 2008. pp. 2608-2611
@inproceedings{1896d0533b384059a25ca34c7e9771b3,
title = "Low cost speech detection using Haar-like filtering for sensornet",
abstract = "Haar-like filtering based speech detection is proposed as a new and very low calculation cost method for sensornet applications. The simple haarlike filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. Our method yielded speech/nonspeech classification accuracy of 96.93{\%} 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 85.77{\%} efficient in terms of the amount of add and multiply calculations while capable of achieving the error rate of only 3.03{\%} relative to MFCC.",
author = "Jun Nishimura and Tadahiro Kuroda",
year = "2008",
doi = "10.1109/ICOSP.2008.4697683",
language = "English",
isbn = "9781424421794",
pages = "2608--2611",
booktitle = "International Conference on Signal Processing Proceedings, ICSP",

}

TY - GEN

T1 - Low cost speech detection using Haar-like filtering for sensornet

AU - Nishimura, Jun

AU - Kuroda, Tadahiro

PY - 2008

Y1 - 2008

N2 - Haar-like filtering based speech detection is proposed as a new and very low calculation cost method for sensornet applications. The simple haarlike filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. Our method yielded speech/nonspeech classification accuracy of 96.93% 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 85.77% efficient in terms of the amount of add and multiply calculations while capable of achieving the error rate of only 3.03% relative to MFCC.

AB - Haar-like filtering based speech detection is proposed as a new and very low calculation cost method for sensornet applications. The simple haarlike filters having variable filter width and shift width are trained to learn appropriate filter parameters from the training samples to detect speech. Our method yielded speech/nonspeech classification accuracy of 96.93% 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 85.77% efficient in terms of the amount of add and multiply calculations while capable of achieving the error rate of only 3.03% relative to MFCC.

UR - http://www.scopus.com/inward/record.url?scp=67249135264&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=67249135264&partnerID=8YFLogxK

U2 - 10.1109/ICOSP.2008.4697683

DO - 10.1109/ICOSP.2008.4697683

M3 - Conference contribution

AN - SCOPUS:67249135264

SN - 9781424421794

SP - 2608

EP - 2611

BT - International Conference on Signal Processing Proceedings, ICSP

ER -