الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 125 |  |  |
| | | from | 125 | | |
Abstract
Sound recognition is a useful technique focused on ideas of conventional pattern recognition and techniques of audio signal analysis. Sound recognition techniques include preliminary data analysis, extraction of features and classification algorithms. Identification of sound can identify function vectors, which generated by preliminary processing and linear prediction. The value of simulation research has been shown in recent years to differentiate various sounds. Spoken voice recognition systems have been developed for a broad range of application, from limited vocabulary keyword recognition to interactive communication and control systems for vocabulary on personal computers. There are two main factors causing the current problems in sound recognition: reach and noisy conditions, as well as sound recognition needs for more languages and specific topics, as sound recognition systems need a lot of knowledge to operate properly, and some of this system hasn’t been collected its languages and topics enoughly. At recent years, singing voice separation and music jinss identification have attracted considerable attention and interest in many real world applications as part of the sound recognition. The goal of singing voice separation approaches is to separate singing voice from music signal, which is an important technology for getting music information retrieval (MIR). To manage large music collections, tools able to extract useful information about musical pieces directly from audio signals are needed. Melody has been extracted by using Robust Principle Component Analysis (RPCA) firstly, then Arabic musical jins has been identified. Mel frequency cepstral Coefficients (MFCC) and Short Time Energy (STE) techniques are used to extract feature for the music signal. Support Vector Machine (SVM) classifier is used for the purpose of jins classification process. Results show that this method achieves important success for arabic music analysis. The best effect of (MFCC) and (STE) for music features extraction has been result, and high accuracy classification by using SVM arrive to 98 %.