|
| Speaker: | Ying Cheng |
| Motorola Australian Research Centre, | |
| Botany, NSW, Australia | |
| Date: | Monday 16th November 1998 |
| Time: | 11:30 - 12:30 |
| Place: | Seminar Room 357, Building E6A, Macquarie University |
Abstract:
The presentation will first introduce an overview of the spoken language processing (SLP) and then give some examples of its applications in the real world.
In general, four topics are involved in the SLP domain including speech recognition, natural language understanding, discourse management and language generation. However, the presentation will mainly focus on the aspects of speech recognition and natural language understanding. A speech recognition component aims to obtain correct transcriptions of spoken utterances. Acoustic and language modeling are used to represent acoustic characteristics of speech units (e.g. phoneme or syllables) and linguistic regularities of a spoken language. A natural language understanding component (or more precisely in this context, spoken language understanding) is used to generate meaning representation from a sequence of words or a word lattice (output from speech recognizer). In a spoken language processing system, the natural language understanding has to deal with phenomena of spoken language such as ungrammatical errors, false starts, repititions and in particular speech recognition errors.Enquiries: swilliam@mri.mq.edu.au
| Last modified: November, 1998 |