Unsupervised speaker change detection for mobile device recorded speech

Olli Vuorinen; Johannes Peltola; Satu-Marja Mäkelä

doi:10.1109/ICASSP.2007.366346

Unsupervised speaker change detection for mobile device recorded speech

Olli Vuorinen, Johannes Peltola, Satu-Marja Mäkelä

BA1619 Data-Driven Life

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

5 Citations (Scopus)

Abstract

In this paper we propose an unsupervised speaker change detection (SCD) system developed for mobile device applications. We use Bayesian information criterion (BIC) to find initial speaker changes, which are then verified or discarded in the second phase by utilizing modified BIC and silence detector information. Silence information usage after initial BIC in decision making is useful to separate real changes from noise peaks. Enhanced peak detector adjusts BIC penalty parameter automatically, which improve the robustness and feasibility. Improved BIC based false alarm compensation (FAC) merges effectively consecutive segments belonging to same speaker. Our experiments have shown the robustness of the algorithm and it produces very satisfactory results for difficult mobile phone recorded speech data.

Original language	English
Title of host publication	IEEE International Conference on Acoustics, Speech, and Signal Processing. Honolulu, HI, USA, 15-20 April 2007. IEEE Cat. No. 07CH378
Publisher	IEEE Institute of Electrical and Electronic Engineers
ISBN (Electronic)	1-4244-0728-1
ISBN (Print)	1-4244-0727-3
DOIs	https://doi.org/10.1109/ICASSP.2007.366346
Publication status	Published - 2007
MoE publication type	A4 Article in a conference publication
Event	IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '07 - Honolulu, United States Duration: 15 Apr 2007 → 20 Apr 2007

Conference

Conference	IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '07
Abbreviated title	ICASSP '07
Country	United States
City	Honolulu
Period	15/04/07 → 20/04/07

Keywords

Metadata
Mobile audio segmentation
Multimedia database
Speaker change detection
Speaker segmentation

Access to Document

10.1109/ICASSP.2007.366346

Fingerprint Dive into the research topics of 'Unsupervised speaker change detection for mobile device recorded speech'. Together they form a unique fingerprint.

Cite this

@inproceedings{a09f7772bb414373a144017453fb11bc,

title = "Unsupervised speaker change detection for mobile device recorded speech",

abstract = "In this paper we propose an unsupervised speaker change detection (SCD) system developed for mobile device applications. We use Bayesian information criterion (BIC) to find initial speaker changes, which are then verified or discarded in the second phase by utilizing modified BIC and silence detector information. Silence information usage after initial BIC in decision making is useful to separate real changes from noise peaks. Enhanced peak detector adjusts BIC penalty parameter automatically, which improve the robustness and feasibility. Improved BIC based false alarm compensation (FAC) merges effectively consecutive segments belonging to same speaker. Our experiments have shown the robustness of the algorithm and it produces very satisfactory results for difficult mobile phone recorded speech data.",

keywords = "Metadata, Mobile audio segmentation, Multimedia database, Speaker change detection, Speaker segmentation",

author = "Olli Vuorinen and Johannes Peltola and Satu-Marja M{\"a}kel{\"a}",

year = "2007",

doi = "10.1109/ICASSP.2007.366346",

language = "English",

isbn = "1-4244-0727-3",

booktitle = "IEEE International Conference on Acoustics, Speech, and Signal Processing. Honolulu, HI, USA, 15-20 April 2007. IEEE Cat. No. 07CH378",

publisher = "IEEE Institute of Electrical and Electronic Engineers",

address = "United States",

note = "IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '07, ICASSP '07 ; Conference date: 15-04-2007 Through 20-04-2007",

}

Vuorinen, O, Peltola, J & Mäkelä, S-M 2007, Unsupervised speaker change detection for mobile device recorded speech. in IEEE International Conference on Acoustics, Speech, and Signal Processing. Honolulu, HI, USA, 15-20 April 2007. IEEE Cat. No. 07CH378. IEEE Institute of Electrical and Electronic Engineers, IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '07, Honolulu, United States, 15/04/07. https://doi.org/10.1109/ICASSP.2007.366346

Unsupervised speaker change detection for mobile device recorded speech. / Vuorinen, Olli; Peltola, Johannes ; Mäkelä, Satu-Marja.

IEEE International Conference on Acoustics, Speech, and Signal Processing. Honolulu, HI, USA, 15-20 April 2007. IEEE Cat. No. 07CH378. IEEE Institute of Electrical and Electronic Engineers, 2007.

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Unsupervised speaker change detection for mobile device recorded speech

AU - Vuorinen, Olli

AU - Peltola, Johannes

AU - Mäkelä, Satu-Marja

PY - 2007

Y1 - 2007

N2 - In this paper we propose an unsupervised speaker change detection (SCD) system developed for mobile device applications. We use Bayesian information criterion (BIC) to find initial speaker changes, which are then verified or discarded in the second phase by utilizing modified BIC and silence detector information. Silence information usage after initial BIC in decision making is useful to separate real changes from noise peaks. Enhanced peak detector adjusts BIC penalty parameter automatically, which improve the robustness and feasibility. Improved BIC based false alarm compensation (FAC) merges effectively consecutive segments belonging to same speaker. Our experiments have shown the robustness of the algorithm and it produces very satisfactory results for difficult mobile phone recorded speech data.

AB - In this paper we propose an unsupervised speaker change detection (SCD) system developed for mobile device applications. We use Bayesian information criterion (BIC) to find initial speaker changes, which are then verified or discarded in the second phase by utilizing modified BIC and silence detector information. Silence information usage after initial BIC in decision making is useful to separate real changes from noise peaks. Enhanced peak detector adjusts BIC penalty parameter automatically, which improve the robustness and feasibility. Improved BIC based false alarm compensation (FAC) merges effectively consecutive segments belonging to same speaker. Our experiments have shown the robustness of the algorithm and it produces very satisfactory results for difficult mobile phone recorded speech data.

KW - Metadata

KW - Mobile audio segmentation

KW - Multimedia database

KW - Speaker change detection

KW - Speaker segmentation

U2 - 10.1109/ICASSP.2007.366346

DO - 10.1109/ICASSP.2007.366346

M3 - Conference article in proceedings

SN - 1-4244-0727-3

BT - IEEE International Conference on Acoustics, Speech, and Signal Processing. Honolulu, HI, USA, 15-20 April 2007. IEEE Cat. No. 07CH378

PB - IEEE Institute of Electrical and Electronic Engineers

T2 - IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '07

Y2 - 15 April 2007 through 20 April 2007

ER -