Cross-Word Modeling for Arabic Speech Recognition utilizes phonological rulesin order to model the cross-word problem, a merging of adjacent words in speech caused by continuous speech, to enhance the performance of continuous speech recognition systems. The author aims to provide an understanding of the cross-word problem and how it can be avoided, specifically focusing on Arabic phonology using an HHM-based classifier. Contents and results will be presented with help of an up-and-running ASR system. Examines Arabic phonology, using an HHM-based classifier. Uses phonological rules to overcome what is known as the cross-word problem INDICE: 1. An Overview of Speech Recognition Systems. 2. Arabic Speech Recognition Systems. 3. The Baseline System. 4. Cross-Word Pronunciation Variations. 5. Modeling of Arabic Cross-Word Prononciation Variations. 6. Performance and Evaluation.
- ISBN: 978-1-4614-1212-0
- Editorial: Springer New York
- Encuadernacion: Cartoné
- Páginas: 79
- Fecha Publicación: 28/01/2012
- Nº Volúmenes: 1
- Idioma: Inglés