Title | ||
---|---|---|
Large Vocabulary Speech Recognition of Slovenian Language Using Data-Driven Morphological Models |
Abstract | ||
---|---|---|
A system for large vocabulary continuous speech recognition of the Slovenian language is described. Two types of modelling units are examined: words and sub-words. A data-driven algorithm is used to automatically obtain word decompositions. The performances of one-pass and two-pass decoding strategies were compared. The new models gave promising results. Recognition accuracy was improved by 3.41% absolute at approx. the same recognition time. On the other hand we achieved 30% increase in real time performance at the same recognition error. |
Year | DOI | Venue |
---|---|---|
2002 | 10.1007/3-540-46154-X_46 | TSD |
Keywords | Field | DocType |
recognition time,slovenian language,recognition accuracy,data-driven morphological models,large vocabulary speech recognition,new model,recognition error,real time performance,large vocabulary continuous speech,data-driven algorithm,promising result,modelling unit,real time,speech recognition | Slavic languages,Data modeling,Vocabulary speech recognition,Data-driven,Approx,Computer science,Speech recognition,Natural language processing,Artificial intelligence,Decoding methods,Vocabulary,Acoustic model | Conference |
ISBN | Citations | PageRank |
3-540-44129-8 | 1 | 0.39 |
References | Authors | |
3 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Tomas Rotovnik | 1 | 1 | 0.72 |
Mirjam Sepesy Maučec | 2 | 506 | 26.34 |
Bogomir Horvat | 3 | 138 | 20.01 |
Zdravko Kacic | 4 | 240 | 36.22 |