Sharif Digital Repository

Scattered and little research in the field of Persian speech synthesis systems has been performed during the last ten years. Comprehensive framework that properly implements and adapts statistical speech synthesis methods for Persian has not been conducted yet. In this paper, recent statistical parametric speech synthesis methods including CLUSTERGEN, traditional HMM-based speech synthesis and its STRAIGHT version, are implemented and adapted for Persian language. CCR test is carried out to compare these methods with each other and with unit selection method. Listeners Score samples based on CMOS. The methods were ranked by averaging the CCR scores. The results show that STRAIGHT-based... 

In speech production systems, the vocal tract is modeled by a filter and glottal pulse by an excitation signal. In most traditional systems impulse train or noise is used as the excitation. In this paper the effects of different excitation signals on Mel-generalized cepstral filters (LPC, warped LPC, mel cepstral and ML cepstral) are studied. Excitation signals with different pulse shapes are used. Furthermore, based on voicing power, noise factor is added to the excitation signals. Totally 600 different experiments with different filter types, number of coefficients, pulse shapes, pulse widths and noise power are preformed. Synthesized speech is evaluated by PESQ measure. A number of pulse... 

One of the main elements of a spoken dialogue system is the Spoken Language Understanding (SLU) unit. Hidden Vector State (HVS) is one of the popular statistical methods applied to the SLU component. Extended Hidden Vector State (EHVS) is an enhanced version of the HVS. Although both parsers need only abstract data annotation, it is quiet time consuming and difficult to label the data. Thus, we present a novel active learning method for the EHVS parser to reduce the human labeling effort. The active learner makes use of pattern classification to select the informative data based on four different uncertainty measures. Experiments are done on a Persian dataset, the University Information... 

Excitation signal has essential role in speech synthesis filters to produce natural speech. In this study, a new method is proposed for modeling the glottal pulse shape of a speaker. A search is done on the glottal pulse shape space using simulated annealing method. The PESQ measure and Cepstral distance between the original signal and the synthesized signal are used as the cost function. An LPC filter with 10 coefficients is used as the synthesis filter. The PESQ value between the original and synthesized speech using traditional impulse is 2.402. Here, the glottal pulse for a certain speaker is modeled in three different experiments. In the first experiment, the negative PESQ measure is... 

Learning a foreign language encompasses a process of trial and error that can lead to the progression of the learner's interlanguage, meaning that the learners make use of their language knowledge that receives feedback and brings about progress through diagnosis of the problems and being corrected by the teacher. Therefore, there exists a need to analyze the underlying reasons causing the students to remain taciturn with a low willingness to communicate (WTC) during a language class. There has been much work done investigating the role of the learner's stable personal characteristics (traits) along with their situational changes at a specific time (states) that influence the student's... 

Using 3-D scanned data to analyze and extract pore network plays a vital role in investigation of porous media's characteristics. In this paper, a new simple method is developed to detect pores and throats for analyzing the connectivity and permeability of the network. This automated method utilizes some of the common and well-known image processing functions which are widely accessible by researchers and this has led to an easy algorithm implementation. In this method, after polishing and quality control of images, using city-block distance function and watershed segmentation algorithm, pores and throats are detected and 3-D network is produced. This method can also be applied on 2-D images... 

In this paper, we try to recognize and distinguish different groups of arrhythmia using novel features which have been obtained from the heart rate's phase space in recent years. For this purpose, we used Triangular Phase Space Mapping (TPSM) and Parabolic Phase Space Mapping (PPSM). For recognition, we used three groups of 15 subjects using the Physionet database (Arrhythmia, Congestive Heart Failure (CHF), and Atrial Fibrillation (AF)) with Normal Sinus Rhythm (NSR). The obtained features discriminate arrhythmia from NSR by p