Although the research community pays little attention to (Inverse) Text Normalization (TN and ITN), this is an essential module in Text-to-Speech (TTS) and Speech Recognition (SR) systems. It has a significant development timeline and requires deep linguistic expertise. One of the main issues is ambiguity resolution, which is particularly problematic when handling numerals in different languages, especially those with gender or case variation. In this paper, we present a framework that can deal simultaneously with TN and ITN and which was applied to twelve different languages. The rules were tested and subsequently refined. The overall performance of the system is presented and discussed.
A Multi-lingual TN/ITN Framework for Speech Technology
CHESI C;
2010-01-01
Abstract
Although the research community pays little attention to (Inverse) Text Normalization (TN and ITN), this is an essential module in Text-to-Speech (TTS) and Speech Recognition (SR) systems. It has a significant development timeline and requires deep linguistic expertise. One of the main issues is ambiguity resolution, which is particularly problematic when handling numerals in different languages, especially those with gender or case variation. In this paper, we present a framework that can deal simultaneously with TN and ITN and which was applied to twelve different languages. The rules were tested and subsequently refined. The overall performance of the system is presented and discussed.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
