Publication Details
Automatic Language Identification System
Matějka Pavel, Ing., Ph.D. (DCGM)
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Schwarz Petr, Ing., Ph.D. (DCGM)
speech processing, automatic language identification
This paper presents the language identification (LID) systemdeveloped in Speech@FIT. The system consists of twoparts: Acoustic LID determines the language directly on thebasis of features derived from the speech signal. We haveimproved existing approaches by adding discriminative trainingof acoustic models. In phonotactic LID, speech is firsttranscribed by phoneme recognizer into strings or graphs (lattices)of phonemes. On these, language models are trainedto capture statistics of sequences of phonemes. We have pioneeredthe use of so called îanti-modelsî for this task. All experimentalresults are reported on standard NIST 2003 data;comparison with other published results is favorable to oursystem.
This paper presents the language identification (LID) system developed in Speech@FIT. The system consists of two parts: Acoustic LID determines the language directly on the basis of features derived from the speech signal. We have improved existing approaches by adding discriminative training of acoustic models. In phonotactic LID, speech is first transcribed by phoneme recognizer into strings or graphs (lattices) of phonemes. On these, language models are trained to capture statistics of sequences of phonemes. We have pioneered the use of so called îanti-modelsî for this task. All experimental results are reported on standard NIST 2003 data; comparison with other published results is favorable to our system.
@inproceedings{BUT22285,
author="Jan {Černocký} and Pavel {Matějka} and Lukáš {Burget} and Petr {Schwarz}",
title="Automatic Language Identification System",
booktitle="Sborník příspěvků z odborného semináře {"}Nové technologie v radiokomunikacích{"}",
year="2006",
pages="1--6",
publisher="University of Defence in Brno",
address="Brno",
url="http://www.fit.vutbr.cz/~cernocky/publi/2006/acr2006.pdf"
}