Dynamic topic-based adaptation of language models: a comparison between different approaches

Echeverry Correa, Julian David and Martínez González, Beatriz and San Segundo Hernández, Rubén and Cordoba Herralde, Ricardo de and Ferreiros López, Javier (2014). Dynamic topic-based adaptation of language models: a comparison between different approaches. In: "VIII Jornadas en Tecnologías del Habla and IV Iberian SLTech Workshop (IberSPEECH 2014)", 19/10/2014 - 21/10/2014, Las Palmas de Gran Canaria, Spain. pp. 139-148.

Description

Title: Dynamic topic-based adaptation of language models: a comparison between different approaches
Author/s:
  • Echeverry Correa, Julian David
  • Martínez González, Beatriz
  • San Segundo Hernández, Rubén
  • Cordoba Herralde, Ricardo de
  • Ferreiros López, Javier
Item Type: Presentation at Congress or Conference (Article)
Event Title: VIII Jornadas en Tecnologías del Habla and IV Iberian SLTech Workshop (IberSPEECH 2014)
Event Dates: 19/10/2014 - 21/10/2014
Event Location: Las Palmas de Gran Canaria, Spain
Title of Book: VIII Jornadas en Tecnologías del Habla and IV Iberian SLTech Workshop (IberSPEECH 2014)
Date: 2014
Subjects:
Freetext Keywords: Language model adaptation, topic identification, automatic speech recognition, information retrieval
Faculty: E.T.S.I. Telecomunicación (UPM)
Department: Ingeniería Electrónica
Creative Commons Licenses: Recognition - No derivative works - Non commercial

Full text

[img]
Preview
PDF - Requires a PDF viewer, such as GSview, Xpdf or Adobe Acrobat Reader
Download (2MB) | Preview

Abstract

This paper presents a dynamic LM adaptation based on the topic that has been identified on a speech segment. We use LSA and the given topic labels in the training dataset to obtain and use the topic models. We propose a dynamic language model adaptation to improve the recognition performance in "a two stages" AST system. The final stage makes use of the topic identification with two variants: the first on uses just the most probable topic and the other one depends on the relative distances of the topics that have been identified. We perform the adaptation of the LM as a linear interpolation between a background model and topic-based LM. The interpolation weight id dynamically adapted according to different parameters. The proposed method is evaluated on the Spanish partition of the EPPS speech database. We achieved a relative reduction in WER of 11.13% over the baseline system which uses a single blackground LM.

Funding Projects

TypeCodeAcronymLeaderTitle
Government of SpainTIN2011-28169-C05-03UnspecifiedUnspecifiedUnspecified
Madrid Regional GovernmentS2009/TIC-1542UnspecifiedUnspecifiedUnspecified

More information

Item ID: 37537
DC Identifier: http://oa.upm.es/37537/
OAI Identifier: oai:oa.upm.es:37537
Deposited by: Memoria Investigacion
Deposited on: 14 Oct 2015 17:36
Last Modified: 06 Jun 2016 17:36
  • Logo InvestigaM (UPM)
  • Logo GEOUP4
  • Logo Open Access
  • Open Access
  • Logo Sherpa/Romeo
    Check whether the anglo-saxon journal in which you have published an article allows you to also publish it under open access.
  • Logo Dulcinea
    Check whether the spanish journal in which you have published an article allows you to also publish it under open access.
  • Logo de Recolecta
  • Logo del Observatorio I+D+i UPM
  • Logo de OpenCourseWare UPM