The authors present a study of a speaker verification system for telephone data based on large-vocabulary speech recognition. After describing the recognition engine, they give details of the verification algorithm and draw comparisons with other systems. The system has been tested on a test set taken from the Switchboard corpus of conversational telephone speech, and they present results showing how performance varies with length of test utterance, and whether or not the training data has been transcribed. The dominant factor in performance appears to be channel or handset mismatch between training and testing data.
Speaker verification through large vocabulary continuous speech recognition
Michael Newman,L. Gillick,Y. Ito,Don McAllaster,B. Peskin
Published 1996 in Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP '96
ABSTRACT
PUBLICATION RECORD
- Publication year
1996
- Venue
Proceeding of Fourth International Conference on Spoken Language Processing. ICSLP '96
- Publication date
1996-10-03
- Fields of study
Computer Science
- Identifiers
- External record
- Source metadata
Semantic Scholar
CITATION MAP
EXTRACTION MAP
CLAIMS
CONCEPTS
- channel mismatch
Differences in recording channel conditions between training and testing speech.
Aliases: channel difference
- conversational telephone speech
Spontaneous spoken dialogue recorded over telephone channels.
Aliases: telephone speech
- handset mismatch
Differences in telephone handset characteristics between training and testing speech.
Aliases: microphone mismatch
- large-vocabulary continuous speech recognition
A speech recognition approach that decodes unrestricted continuous speech using a large vocabulary.
Aliases: LVCSR
- speaker verification system
A system that decides whether a speech sample matches a claimed speaker identity.
Aliases: speaker verifier
- switchboard corpus
A corpus of conversational telephone speech used here as the test set for evaluation.
Aliases: SWBD
- test utterance length
The duration or amount of speech available in the test sample used for verification.
Aliases: utterance length
- training and testing data
The speech data used to fit the system and the separate speech data used for evaluation.
Aliases: train-test data
- transcribed training data
Training speech whose audio is paired with manual transcript labels.
Aliases: transcribed data
- verification performance
The measured effectiveness of the speaker verification system on the evaluation set.
Aliases: system performance
REFERENCES
Showing 1-8 of 8 references · Page 1 of 1
CITED BY
Showing 1-40 of 40 citing papers · Page 1 of 1