preview

Semi Automatic Segmentation System For Syllables Extraction From Continuous Arabic Audio Signal

Better Essays

Semi-automatic Segmentation System for Syllables Extraction from Continuous Arabic Audio Signal Mohamed S. Abdo*, 1 , Ahmed H. Kandil 2. 1 Biomedical Engineering Department, Faculty of Engineering, Minia University, Minia, Egypt. 2 Systems and Biomedical Engineering Department, Faculty of Engineering, Cairo University, Giza, Egypt. Bioengmsa@yahoo.com , ahkandil_1@yahoo.com ABSTRACT This paper introduces a double stage speaker independent segmentation system for the breaking-up of Arabic spoken sentences into its isolated syllables. The main goal is to implement an accurate system for the construction of acoustical Arabic syllables database. Syllable-Based Arabic Speech Verification/Recognition is the prospective goal for this work. The study experimented the employing of the template matching technique with a selected acoustical features for the allocation of syllables with sharp boundaries. The proposed methods can manipulate the explored features in two stages of decomposition to segment 2544 syllables from a sample size of 276 utterances achieving segmentation consistency rate of about 91.5 %. KEYWORDS Arabic syllables, Automatic Segmentation, Boundaries detection, Delta-MFCC feature. 1. Introduction S ignal and information processing areas concerns the representation, analysis, and communication of the information in various types of signals including audio, video, and biological or sensory data. Speech and natural language processing (SNLP) is spreaded across

Get Access