ACM TechNews

"..."

Bringing Down the Language Barrier... Automatically
ICT Results (05/02/08)

The European Union funded TC-STAR project is developing automatic speech-to-speech translation technology. "For humans, translation is difficult. We have to master both the source language and the target language, and machine translation is significantly more difficult than that," says FBK-irst researcher Marcello Federico. "To our knowledge, TC-STAR has been the first project in the world addressing unrestricted speech-to-speech translation." The TC-STAR project included the development of three technologies. Automatic Speech Recognition (ASR) is used to transcribe the spoken word, Spoken Language Translation (SLT) translates the source language to the target language, and Text to Speech (TTS) generates the spoken output. All of these technologies still need to be perfected. A key innovation of the project was to combine the output of several ASR and SLT systems to make the transcription and translations phases more accurate. Based on the Bilingual Evaluation Understudy method used to compare machine and human translations, the quality of the translations improved by between 40 percent and 60 percent over the course of the project, and 70 percent of the words were translated correctly, though they were sometimes misplaced in the sentence. Components developed by the TC-STAR project have been made available under an open source license.

http://cordis.europa.eu/ictresults/index.cfm/section/news/tpl/
article/BrowsingType/Features/ID/89699

---

© Copyright 2008 Information, Inc. This service may be reproduced for internal distribution.