Tailoring Commercial Off-the-Shelf (COTS) Voice Recognition Software for Handling Multiple Languages: An Experimental Study
In this experimental study, we propose to investigate the role played by language typology in adapting commercial-off-the-shelf [COTS] speech-to-text software [STS] developed for one language belonging to a particular language group is capable of recognizing speech input provided by a language belonging to a different language group. Typical language groups include Indo-European, Dravidian, Semitic, Sino-Tibetan, alphabet-based, syllabary-based, etc. Studies by the author [1-3] and other researchers [4-6] have shown that the basic technique works for languages drawn from within the same language group; however, similar cross linguistic-group studies have not been conducted.
Marshall, Roger (2004). Tailoring Commercial Off-the-Shelf (COTS) Voice Recognition Software for Handling Multiple Languages: An Experimental Study. CARS Summer Grants. Item 122.
This document is currently not available here.