Class n-gram models for very large vocabulary speech recognition of Finnish and Estonian

Matti Varjokallio*, Mikko Kurimo, Sami Virpioja

*Tämän työn vastaava kirjoittaja

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussaConference article in proceedingsScientificvertaisarvioitu

5 Sitaatiot (Scopus)

Abstrakti

We study class n-gram models for very large vocabulary speech recognition of Finnish and Estonian. The models are trained with vocabulary sizes of several millions of words using automatically derived classes. To evaluate the models on Finnish and an Estonian broadcast news speech recognition task, we modify Aalto University’s LVCSR decoder to operate with the class n-grams and very large vocabularies. Linear interpolation of a standard n-gram model and a class n-gram model provides relative perplexity improvements of 21.3% for Finnish and 12.8% for Estonian over the n-gram model. The relative improvements in word error rates are 5.5% for Finnish and 7.4% for Estonian. We also compare our word-based models to a state-of-the-art unlimited vocabulary recognizer utilizing subword n-gram models, and show that the very large vocabulary word-based models can perform equally well or better.

AlkuperäiskieliEnglanti
OtsikkoProceedings of the 4th International Conference on Statistical Language and Speech Processing, SLSP 2016
KustantajaSpringer
Sivut133-144
Sivumäärä12
Vuosikerta9918 LNCS
ISBN (painettu)9783319459240, 9783319459257
DOI - pysyväislinkit
TilaJulkaistu - 2016
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisussa
TapahtumaInternational Conference on Statistical Language and Speech Processing - Pilsen, Tshekki
Kesto: 11 lokak. 201612 lokak. 2016
Konferenssinumero: 4
http://grammars.grlmc.com/SLSP2016/

Julkaisusarja

NimiLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Vuosikerta9918 LNCS
ISSN (painettu)03029743
ISSN (elektroninen)16113349

Conference

ConferenceInternational Conference on Statistical Language and Speech Processing
LyhennettäSLSP
Maa/AlueTshekki
KaupunkiPilsen
Ajanjakso11/10/201612/10/2016
www-osoite

Sormenjälki

Sukella tutkimusaiheisiin 'Class n-gram models for very large vocabulary speech recognition of Finnish and Estonian'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

Siteeraa tätä