Robust voice activity detection algorithm for estimating noise spectrum

Kyoung Ho Woo, Tae Young Yang, Kun Jung Park, Chungyong Lee

Research output: Contribution to journalArticle

105 Citations (Scopus)

Abstract

A new voice activity detectlon (VAD) algorithm is proposed for estimating the spectrum of car noise in which noise is filtered out in the frequency domain. The proposed algorithm uses the log energy parameters which are composed of two parts in the critical band. The algorithm detects the noise period by applying two adaptive thresholds to each part. Using the noise period we can reliably estimate the time-varying noise characteristics. The advantage of the proposed technique is that it can prevent incorrect detections caused by unvoiced or nasal sounds with high frequency components being covered by car noise with low frequency components. The algorithm is suitable for real time implementation with one microphone. Also, a speaker-independent speech recognition system has been implemented for car navigation using a fixed point Oak DSP system, which incorporates the proposed VAD algorithm. The system enhanced the recognition rates for 12 isolated command words to 94.52% compared with the 80.7% of the baseline recognizer.

Original languageEnglish
Pages (from-to)180-181
Number of pages2
JournalElectronics Letters
Volume36
Issue number2
DOIs
Publication statusPublished - 2000 Jan 20

Fingerprint

Acoustic noise
Microphones
Speech recognition
Navigation
Railroad cars
Acoustic waves

All Science Journal Classification (ASJC) codes

  • Electrical and Electronic Engineering

Cite this

Woo, Kyoung Ho ; Yang, Tae Young ; Park, Kun Jung ; Lee, Chungyong. / Robust voice activity detection algorithm for estimating noise spectrum. In: Electronics Letters. 2000 ; Vol. 36, No. 2. pp. 180-181.
@article{ea76fc889f91425286f8256f44a9e518,
title = "Robust voice activity detection algorithm for estimating noise spectrum",
abstract = "A new voice activity detectlon (VAD) algorithm is proposed for estimating the spectrum of car noise in which noise is filtered out in the frequency domain. The proposed algorithm uses the log energy parameters which are composed of two parts in the critical band. The algorithm detects the noise period by applying two adaptive thresholds to each part. Using the noise period we can reliably estimate the time-varying noise characteristics. The advantage of the proposed technique is that it can prevent incorrect detections caused by unvoiced or nasal sounds with high frequency components being covered by car noise with low frequency components. The algorithm is suitable for real time implementation with one microphone. Also, a speaker-independent speech recognition system has been implemented for car navigation using a fixed point Oak DSP system, which incorporates the proposed VAD algorithm. The system enhanced the recognition rates for 12 isolated command words to 94.52{\%} compared with the 80.7{\%} of the baseline recognizer.",
author = "Woo, {Kyoung Ho} and Yang, {Tae Young} and Park, {Kun Jung} and Chungyong Lee",
year = "2000",
month = "1",
day = "20",
doi = "10.1049/el:20000192",
language = "English",
volume = "36",
pages = "180--181",
journal = "Electronics Letters",
issn = "0013-5194",
publisher = "Institution of Engineering and Technology",
number = "2",

}

Robust voice activity detection algorithm for estimating noise spectrum. / Woo, Kyoung Ho; Yang, Tae Young; Park, Kun Jung; Lee, Chungyong.

In: Electronics Letters, Vol. 36, No. 2, 20.01.2000, p. 180-181.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Robust voice activity detection algorithm for estimating noise spectrum

AU - Woo, Kyoung Ho

AU - Yang, Tae Young

AU - Park, Kun Jung

AU - Lee, Chungyong

PY - 2000/1/20

Y1 - 2000/1/20

N2 - A new voice activity detectlon (VAD) algorithm is proposed for estimating the spectrum of car noise in which noise is filtered out in the frequency domain. The proposed algorithm uses the log energy parameters which are composed of two parts in the critical band. The algorithm detects the noise period by applying two adaptive thresholds to each part. Using the noise period we can reliably estimate the time-varying noise characteristics. The advantage of the proposed technique is that it can prevent incorrect detections caused by unvoiced or nasal sounds with high frequency components being covered by car noise with low frequency components. The algorithm is suitable for real time implementation with one microphone. Also, a speaker-independent speech recognition system has been implemented for car navigation using a fixed point Oak DSP system, which incorporates the proposed VAD algorithm. The system enhanced the recognition rates for 12 isolated command words to 94.52% compared with the 80.7% of the baseline recognizer.

AB - A new voice activity detectlon (VAD) algorithm is proposed for estimating the spectrum of car noise in which noise is filtered out in the frequency domain. The proposed algorithm uses the log energy parameters which are composed of two parts in the critical band. The algorithm detects the noise period by applying two adaptive thresholds to each part. Using the noise period we can reliably estimate the time-varying noise characteristics. The advantage of the proposed technique is that it can prevent incorrect detections caused by unvoiced or nasal sounds with high frequency components being covered by car noise with low frequency components. The algorithm is suitable for real time implementation with one microphone. Also, a speaker-independent speech recognition system has been implemented for car navigation using a fixed point Oak DSP system, which incorporates the proposed VAD algorithm. The system enhanced the recognition rates for 12 isolated command words to 94.52% compared with the 80.7% of the baseline recognizer.

UR - http://www.scopus.com/inward/record.url?scp=0033903480&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033903480&partnerID=8YFLogxK

U2 - 10.1049/el:20000192

DO - 10.1049/el:20000192

M3 - Article

AN - SCOPUS:0033903480

VL - 36

SP - 180

EP - 181

JO - Electronics Letters

JF - Electronics Letters

SN - 0013-5194

IS - 2

ER -