Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection

Konstantinos Koumpis; Steve Renals; Mahesan Niranjan

Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection

Konstantinos Koumpis, Steve Renals, Mahesan Niranjan

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

Abstract

This paper presents a novel data-driven approach to summarizing spoken audio transcripts utilizing lexical and prosodic features. The former are obtained from a speech recognizer and the latter are extracted automatically from speech waveforms. We employ a feature subset selection algorithm, based on ROC curves, which examines different combinations of features at different target operating conditions. The approach is evaluated on the IBM Voicemail corpus, demonstrating that it is possible and desirable to avoid complete commitment to a single best classifier or feature set.

Original language	English
Title of host publication	7th European Conference on Speech Communication and Technology
Subtitle of host publication	Eurospeech 2001 Scandinavia
Editors	Paul Dalsgaard, Børge Lindberg, Henrik Benner, Zheng-Hua Tan
Place of Publication	Aalborg, Denmark
Publisher	Kommunik Grafiske Løsninger
Pages	2377-2380
ISBN (Print)	87-90834-10-0
Publication status	Published - 2001
Event	7th European Conference on Speech Communication and Technology (Eurospeech 2001 Scandinavia) - Aalborg Congress and Culture Centre, Aalborg, Denmark Duration: 3 Sep 2001 → 7 Sep 2001

Conference

Conference	7th European Conference on Speech Communication and Technology (Eurospeech 2001 Scandinavia)
Country/Territory	Denmark
City	Aalborg
Period	3/09/01 → 7/09/01

Access to Document

eurospeech01Submitted manuscript, 61 KB

http://www.isca-speech.org/archive/eurospeech_2001/e01_2377.html

Cite this

Koumpis, K., Renals, S., & Niranjan, M. (2001). Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection. In P. Dalsgaard, B. Lindberg, H. Benner, & Z-H. Tan (Eds.), 7th European Conference on Speech Communication and Technology: Eurospeech 2001 Scandinavia (pp. 2377-2380). Kommunik Grafiske Løsninger . http://www.isca-speech.org/archive/eurospeech_2001/e01_2377.html

Koumpis, Konstantinos ; Renals, Steve ; Niranjan, Mahesan. / Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection. 7th European Conference on Speech Communication and Technology: Eurospeech 2001 Scandinavia. editor / Paul Dalsgaard ; Børge Lindberg ; Henrik Benner ; Zheng-Hua Tan. Aalborg, Denmark : Kommunik Grafiske Løsninger , 2001. pp. 2377-2380

@inproceedings{8696ed5c7ff143878f131ac71a17b4ff,

title = "Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection",

abstract = "This paper presents a novel data-driven approach to summarizing spoken audio transcripts utilizing lexical and prosodic features. The former are obtained from a speech recognizer and the latter are extracted automatically from speech waveforms. We employ a feature subset selection algorithm, based on ROC curves, which examines different combinations of features at different target operating conditions. The approach is evaluated on the IBM Voicemail corpus, demonstrating that it is possible and desirable to avoid complete commitment to a single best classifier or feature set.",

author = "Konstantinos Koumpis and Steve Renals and Mahesan Niranjan",

year = "2001",

language = "English",

isbn = "87-90834-10-0",

pages = "2377--2380",

editor = "Paul Dalsgaard and B{\o}rge Lindberg and Henrik Benner and Zheng-Hua Tan",

booktitle = "7th European Conference on Speech Communication and Technology",

publisher = " Kommunik Grafiske L{\o}sninger ",

note = "7th European Conference on Speech Communication and Technology (Eurospeech 2001 Scandinavia) ; Conference date: 03-09-2001 Through 07-09-2001",

}

Koumpis, K, Renals, S & Niranjan, M 2001, Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection. in P Dalsgaard, B Lindberg, H Benner & Z-H Tan (eds), 7th European Conference on Speech Communication and Technology: Eurospeech 2001 Scandinavia. Kommunik Grafiske Løsninger , Aalborg, Denmark, pp. 2377-2380, 7th European Conference on Speech Communication and Technology (Eurospeech 2001 Scandinavia), Aalborg, Denmark, 3/09/01. <http://www.isca-speech.org/archive/eurospeech_2001/e01_2377.html>

Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection. / Koumpis, Konstantinos; Renals, Steve; Niranjan, Mahesan.

7th European Conference on Speech Communication and Technology: Eurospeech 2001 Scandinavia. ed. / Paul Dalsgaard; Børge Lindberg; Henrik Benner; Zheng-Hua Tan. Aalborg, Denmark : Kommunik Grafiske Løsninger , 2001. p. 2377-2380.

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

TY - GEN

T1 - Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection

AU - Koumpis, Konstantinos

AU - Renals, Steve

AU - Niranjan, Mahesan

PY - 2001

Y1 - 2001

N2 - This paper presents a novel data-driven approach to summarizing spoken audio transcripts utilizing lexical and prosodic features. The former are obtained from a speech recognizer and the latter are extracted automatically from speech waveforms. We employ a feature subset selection algorithm, based on ROC curves, which examines different combinations of features at different target operating conditions. The approach is evaluated on the IBM Voicemail corpus, demonstrating that it is possible and desirable to avoid complete commitment to a single best classifier or feature set.

AB - This paper presents a novel data-driven approach to summarizing spoken audio transcripts utilizing lexical and prosodic features. The former are obtained from a speech recognizer and the latter are extracted automatically from speech waveforms. We employ a feature subset selection algorithm, based on ROC curves, which examines different combinations of features at different target operating conditions. The approach is evaluated on the IBM Voicemail corpus, demonstrating that it is possible and desirable to avoid complete commitment to a single best classifier or feature set.

M3 - Conference contribution

SN - 87-90834-10-0

SP - 2377

EP - 2380

BT - 7th European Conference on Speech Communication and Technology

A2 - Dalsgaard, Paul

A2 - Lindberg, Børge

A2 - Benner, Henrik

A2 - Tan, Zheng-Hua

PB - Kommunik Grafiske Løsninger

CY - Aalborg, Denmark

T2 - 7th European Conference on Speech Communication and Technology (Eurospeech 2001 Scandinavia)

Y2 - 3 September 2001 through 7 September 2001

ER -

Extractive Summarization of Voicemail using Lexical and Prosodic Feature Subset Selection

Abstract

Conference

Access to Document

Fingerprint

Cite this