Diagnosing Stuttering using Support Vector Machine and LDA Classification Method in Persian Language

چکیده مقاله

Stuttering, is one of the most common speech disorder, which human encounter with it. Beside usual methods which is used by speechthropist, several artificial intelligent based methods have been used to identify and classify stuttering. Lack of sufficient real number of cases for training and testing, prevent researchers to use algorithms such as artificial neural network (ANN), hidden Markov model (HMM) and etc. Here we have used support vector machine (SVM), which is much suitable algorithm for classifying sparse date with a appropriate accuracy. Our proposed system consists of five steps include: 1. Receiving sample signal, 2. Preprocessing sample signal, 3. compute the required features, 4. Feature extraction, and 5. Categorizing each sample to the appropriate class. For a better investigation, we have used different features, such as Mel frequency Cepstrum coefficient (MFCC), Max FFT, Kurtosis, Skewness and etc. In addition, LDA is used as method for reducing mass of features’ data and getting the most out of it. Also, in experiment section, 20 labeled samples from 10 people with fluent speech and 10 patients who referred to speech therapy centers, were used. Results show, the ability of proposed schema for resolution each sample of data. The best result belongs to Max FFT with the 100% of accuracy.

دریافت لینک دانلود مقاله