Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features

Hasan Almgotir Kadhim; Lok Woo; Satnam Dlay

doi:10.1109/PIC.2015.7489855

Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features

Hasan Almgotir Kadhim^*, Lok Woo, Satnam Dlay

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Citations (Scopus)

Abstract

To simplify the jobs of speaker diarization and speech separation, at first, speech signal should be segregated to two speech formats, dialog and mixture. This paper describes a new algorithm which achieves that first step efficiently. The algorithm is based on Perceptual Linear Predictive feature extraction, optimized k-means and both top-down & bottom-up scenarios. After extracting features of the observation signal, k-means clusters the statistical properties such as variances of the PDF (histogram) of clustered extracted features. k-means is optimized by discounting the worst pattern of clustering step through doing the k-means procedure twice. The feedback loop is necessary for the guiding of the optimized k-means by exploiting the attributes of ordinary k-means. The results of segregation are excellent. The calculated diarization error rate of outputs is very limited.

Original language	English
Title of host publication	Proceedings of 2015 IEEE International Conference on Progress in Informatics and Computing, PIC 2015
Editors	Liang Xiao, Yinglin Wang
Publisher	IEEE
Pages	286-291
Number of pages	6
ISBN (Electronic)	9781467390880
ISBN (Print)	9781467380867
DOIs	https://doi.org/10.1109/PIC.2015.7489855
Publication status	Published - 13 Jun 2016
Externally published	Yes
Event	3rd IEEE International Conference on Progress in Informatics and Computing, PIC 2015 - Nanjing, China Duration: 18 Dec 2015 → 20 Dec 2015

Conference

Conference	3rd IEEE International Conference on Progress in Informatics and Computing, PIC 2015
Country/Territory	China
City	Nanjing
Period	18/12/15 → 20/12/15

Keywords

bottom-up scenario
clustering
dierization error rate
k-means
perceptual linear predictive
segmentation
speaker diarization
speech segregation
speech separation
top-down scenario

Access to Document

10.1109/PIC.2015.7489855

Cite this

@inproceedings{2ed9f8812a4d495f816cc7a4c986e45f,

title = "Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features",

abstract = "To simplify the jobs of speaker diarization and speech separation, at first, speech signal should be segregated to two speech formats, dialog and mixture. This paper describes a new algorithm which achieves that first step efficiently. The algorithm is based on Perceptual Linear Predictive feature extraction, optimized k-means and both top-down & bottom-up scenarios. After extracting features of the observation signal, k-means clusters the statistical properties such as variances of the PDF (histogram) of clustered extracted features. k-means is optimized by discounting the worst pattern of clustering step through doing the k-means procedure twice. The feedback loop is necessary for the guiding of the optimized k-means by exploiting the attributes of ordinary k-means. The results of segregation are excellent. The calculated diarization error rate of outputs is very limited.",

keywords = "bottom-up scenario, clustering, dierization error rate, k-means, perceptual linear predictive, segmentation, speaker diarization, speech segregation, speech separation, top-down scenario",

author = "{Almgotir Kadhim}, Hasan and Lok Woo and Satnam Dlay",

year = "2016",

month = jun,

day = "13",

doi = "10.1109/PIC.2015.7489855",

language = "English",

isbn = "9781467380867",

pages = "286--291",

editor = "Liang Xiao and Yinglin Wang",

booktitle = "Proceedings of 2015 IEEE International Conference on Progress in Informatics and Computing, PIC 2015",

publisher = "IEEE",

address = "United States",

note = "3rd IEEE International Conference on Progress in Informatics and Computing, PIC 2015 ; Conference date: 18-12-2015 Through 20-12-2015",

}

Almgotir Kadhim, H, Woo, L & Dlay, S 2016, Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features. in L Xiao & Y Wang (eds), Proceedings of 2015 IEEE International Conference on Progress in Informatics and Computing, PIC 2015. IEEE, pp. 286-291, 3rd IEEE International Conference on Progress in Informatics and Computing, PIC 2015, Nanjing, China, 18/12/15. https://doi.org/10.1109/PIC.2015.7489855

Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features. / Almgotir Kadhim, Hasan; Woo, Lok; Dlay, Satnam.
Proceedings of 2015 IEEE International Conference on Progress in Informatics and Computing, PIC 2015. ed. / Liang Xiao; Yinglin Wang. IEEE, 2016. p. 286-291.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features

AU - Almgotir Kadhim, Hasan

AU - Woo, Lok

AU - Dlay, Satnam

PY - 2016/6/13

Y1 - 2016/6/13

N2 - To simplify the jobs of speaker diarization and speech separation, at first, speech signal should be segregated to two speech formats, dialog and mixture. This paper describes a new algorithm which achieves that first step efficiently. The algorithm is based on Perceptual Linear Predictive feature extraction, optimized k-means and both top-down & bottom-up scenarios. After extracting features of the observation signal, k-means clusters the statistical properties such as variances of the PDF (histogram) of clustered extracted features. k-means is optimized by discounting the worst pattern of clustering step through doing the k-means procedure twice. The feedback loop is necessary for the guiding of the optimized k-means by exploiting the attributes of ordinary k-means. The results of segregation are excellent. The calculated diarization error rate of outputs is very limited.

AB - To simplify the jobs of speaker diarization and speech separation, at first, speech signal should be segregated to two speech formats, dialog and mixture. This paper describes a new algorithm which achieves that first step efficiently. The algorithm is based on Perceptual Linear Predictive feature extraction, optimized k-means and both top-down & bottom-up scenarios. After extracting features of the observation signal, k-means clusters the statistical properties such as variances of the PDF (histogram) of clustered extracted features. k-means is optimized by discounting the worst pattern of clustering step through doing the k-means procedure twice. The feedback loop is necessary for the guiding of the optimized k-means by exploiting the attributes of ordinary k-means. The results of segregation are excellent. The calculated diarization error rate of outputs is very limited.

KW - bottom-up scenario

KW - clustering

KW - dierization error rate

KW - k-means

KW - perceptual linear predictive

KW - segmentation

KW - speaker diarization

KW - speech segregation

KW - speech separation

KW - top-down scenario

U2 - 10.1109/PIC.2015.7489855

DO - 10.1109/PIC.2015.7489855

M3 - Conference contribution

AN - SCOPUS:84979681573

SN - 9781467380867

SP - 286

EP - 291

BT - Proceedings of 2015 IEEE International Conference on Progress in Informatics and Computing, PIC 2015

A2 - Xiao, Liang

A2 - Wang, Yinglin

PB - IEEE

T2 - 3rd IEEE International Conference on Progress in Informatics and Computing, PIC 2015

Y2 - 18 December 2015 through 20 December 2015

ER -

Novel algorithm for speech segregation by optimized k-means of statistical properties of clustered features

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this