A new quality measure for topic segmentation of text and speech

Mehryar Mohri, Pedro Moreno, Eugene Weinstein

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

Abstract

The recent proliferation of large multimedia collections has gathered immense attention from the speech research community, because speech recognition enables the transcription and indexing of such collections. Topicality information can be used to improve transcription quality and enable content navigation. In this paper, we give a novel quality measure for topic segmentation algorithms that improves over previously used measures. Our measure takes into account not only the presence or absence of topic boundaries but also the content of the text or speech segments labeled as topic-coherent. Additionally, we demonstrate that topic segmentation quality of spoken language can be improved using speech recognition lattices. Using lattices, improvements over the baseline one-best topic model are observed when measured with the previously existing topic segmentation quality measure, as well as the new measure proposed in this paper (9.4% and 7.0% relative error reduction, respectively).

Original languageEnglish (US)
Title of host publicationProceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH
Pages2743-2746
Number of pages4
StatePublished - 2009
Event10th Annual Conference of the International Speech Communication Association, INTERSPEECH 2009 - Brighton, United Kingdom
Duration: Sep 6 2009Sep 10 2009

Other

Other10th Annual Conference of the International Speech Communication Association, INTERSPEECH 2009
CountryUnited Kingdom
CityBrighton
Period9/6/099/10/09

Fingerprint

Transcription
Speech recognition
Navigation
Multimedia
Language
Research

Keywords

  • Speech processing
  • Speech recognition lattices
  • Text similarity
  • Topic segmentation

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Signal Processing
  • Software
  • Sensory Systems

Cite this

Mohri, M., Moreno, P., & Weinstein, E. (2009). A new quality measure for topic segmentation of text and speech. In Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH (pp. 2743-2746)

A new quality measure for topic segmentation of text and speech. / Mohri, Mehryar; Moreno, Pedro; Weinstein, Eugene.

Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. 2009. p. 2743-2746.

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

Mohri, M, Moreno, P & Weinstein, E 2009, A new quality measure for topic segmentation of text and speech. in Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. pp. 2743-2746, 10th Annual Conference of the International Speech Communication Association, INTERSPEECH 2009, Brighton, United Kingdom, 9/6/09.
Mohri M, Moreno P, Weinstein E. A new quality measure for topic segmentation of text and speech. In Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. 2009. p. 2743-2746
Mohri, Mehryar ; Moreno, Pedro ; Weinstein, Eugene. / A new quality measure for topic segmentation of text and speech. Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH. 2009. pp. 2743-2746
@inproceedings{207ce13fc1464547a48d6faaed1b8e38,
title = "A new quality measure for topic segmentation of text and speech",
abstract = "The recent proliferation of large multimedia collections has gathered immense attention from the speech research community, because speech recognition enables the transcription and indexing of such collections. Topicality information can be used to improve transcription quality and enable content navigation. In this paper, we give a novel quality measure for topic segmentation algorithms that improves over previously used measures. Our measure takes into account not only the presence or absence of topic boundaries but also the content of the text or speech segments labeled as topic-coherent. Additionally, we demonstrate that topic segmentation quality of spoken language can be improved using speech recognition lattices. Using lattices, improvements over the baseline one-best topic model are observed when measured with the previously existing topic segmentation quality measure, as well as the new measure proposed in this paper (9.4{\%} and 7.0{\%} relative error reduction, respectively).",
keywords = "Speech processing, Speech recognition lattices, Text similarity, Topic segmentation",
author = "Mehryar Mohri and Pedro Moreno and Eugene Weinstein",
year = "2009",
language = "English (US)",
pages = "2743--2746",
booktitle = "Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH",

}

TY - GEN

T1 - A new quality measure for topic segmentation of text and speech

AU - Mohri, Mehryar

AU - Moreno, Pedro

AU - Weinstein, Eugene

PY - 2009

Y1 - 2009

N2 - The recent proliferation of large multimedia collections has gathered immense attention from the speech research community, because speech recognition enables the transcription and indexing of such collections. Topicality information can be used to improve transcription quality and enable content navigation. In this paper, we give a novel quality measure for topic segmentation algorithms that improves over previously used measures. Our measure takes into account not only the presence or absence of topic boundaries but also the content of the text or speech segments labeled as topic-coherent. Additionally, we demonstrate that topic segmentation quality of spoken language can be improved using speech recognition lattices. Using lattices, improvements over the baseline one-best topic model are observed when measured with the previously existing topic segmentation quality measure, as well as the new measure proposed in this paper (9.4% and 7.0% relative error reduction, respectively).

AB - The recent proliferation of large multimedia collections has gathered immense attention from the speech research community, because speech recognition enables the transcription and indexing of such collections. Topicality information can be used to improve transcription quality and enable content navigation. In this paper, we give a novel quality measure for topic segmentation algorithms that improves over previously used measures. Our measure takes into account not only the presence or absence of topic boundaries but also the content of the text or speech segments labeled as topic-coherent. Additionally, we demonstrate that topic segmentation quality of spoken language can be improved using speech recognition lattices. Using lattices, improvements over the baseline one-best topic model are observed when measured with the previously existing topic segmentation quality measure, as well as the new measure proposed in this paper (9.4% and 7.0% relative error reduction, respectively).

KW - Speech processing

KW - Speech recognition lattices

KW - Text similarity

KW - Topic segmentation

UR - http://www.scopus.com/inward/record.url?scp=70450210023&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70450210023&partnerID=8YFLogxK

M3 - Conference contribution

SP - 2743

EP - 2746

BT - Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

ER -