Publication:
An Illustrated Methodology for Evaluating ASR Systems

Thumbnail Image
Identifiers
URI: https://hdl.handle.net/10016/20354
ISSN: 0302-9743
ISBN: 978-3-642-37424-1 (Print)
ISBN: 978-3-642-37425-8 (Online)
DOI: 10.1007/978-3-642-37425-8_3
UXXI: CC/0000023019
Files
illustrated_LNCS_2013..pdf (615.73 KB)
Publication date
2011-07
Defense date
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Springer
Impact
Google Scholar
Export
Research Projects
Organizational Units
Journal Issue
Abstract
Automatic speech recognition technology can be integrated in an information retrieval process to allow searching on multimedia contents. But, in order to assure an adequate retrieval performance is necessary to state the quality of the recognition phase, especially in speaker-independent and domainindependent environments. This paper introduces a methodology to accomplish the evaluation of different speech recognition systems in several scenarios considering also the creation of new corpora of different types (broadcast news, interviews, etc.), especially in other languages apart from English that are not widely addressed in speech community.
Description
Proceeding of: 9th International Workshop on Adaptive Multimedia Retrieval (AMR 2011) Took place 2011, July, 18-19, in Barcelona, Spain. The event Web site is http://stel.ub.edu/amr2011/
Keywords
Automatic Speech Recognition (ASR), ASR Evaluation, Audio Transcription
Bibliographic citation
Adaptive Multimedia Retrieval. Large-Scale Multimedia Retrieval and Evaluation. LNCS 7836 (2013) pp. 33–42
Collections
DI - LABDA - Comunicaciones en Congresos y otros eventos
DI - LABDA - Capítulos de Monografías