Sharif Digital Repository

We present our work on improving the numerator graph for discriminative training using the lattice-free maximum mutual information (MMI) criterion. Specifically, we propose a scheme for creating unconstrained numerator graphs by removing time constraints from the baseline numerator graphs. This leads to much smaller graphs and therefore faster preparation of training supervisions. By testing the proposed un-constrained supervisions using factorized time-delay neural network (TDNN) models, we observe 0.5% to 2.6% relative improvement over the state-of-the-art word error rates on various large-vocabulary speech recognition databases. © 2018 IEEE  

In an attempt to reduce the oxidation and Cr evaporation rates, various protective coating layers with a nominal composition of NiCo 2-xFexO4 (x = 0, 0.5, 1) were deposited on the SUS 430 ferritic stainless steel substrate, as interconnect for solid oxide fuel cell application, by sol-gel dip coating method. Then, the coated samples were soaked at 750 C for 2.5 h in N2 and subsequently for 2.5 h in air. Phase composition and microstructure of the coatings were characterized by X-ray diffraction (XRD) and scanning electron microscopy (SEM). Moreover, long-term isothermal oxidation experiment and area specific resistance (ASR) measurement were also carried out on the coated samples. Results... 

Hybrid deep neural network hidden Markov models (DNN-HMM) have achieved impressive results on large vocabulary continuous speech recognition (LVCSR) tasks. However, the recent approaches using DNN-HMM models are not explored much for text recognition. Inspired by the current work in automatic speech recognition (ASR) and machine translation, we present an open vocabulary sub-word text recognition system. The sub-word lexicon and sub-word language model (LM) helps in overcoming the challenge of recognizing out of vocabulary (OOV) words, and a time delay neural network (TDNN) and convolution neural network (CNN) based DNN-HMM optical model (OM) efficiently models the sequence dependency in the... 

Prior studies were conducted in an isolated manner for the use of glass powder (GP) either as ordinary Portland cement (OPC) replacement material in blended cement or as a precursor in AAM material. There was no reported literature that made a direct comparison on the properties of blended cement and AAM containing GP with an equivalent mix composition. The present study was aimed to evaluate and compare the effects of GP as a partial precursor on properties of both OPC and alkali-activated slag (AAS) mortars. For this purpose, cement and ground granulated blast furnace slag (GGBFS) was partially replaced with 10%, 20%, 30%, and 40% (by weight) GP in OPC and AAS mortars, respectively. The... 

Hamoon-Jazmoorian plain is located in southeast of Iran. Overexploitation of groundwater in this plain has led to water level decline and caused serious problems such as land subsidence, aquifer destruction and water quality degradation. The increasing population and agricultural development along with drought and climate change, have further increased the pressure on water resources in this region over the last years. In order to overcome such crisis, introduction of surface water into an aquifer at particular locations can be a suitable solution. A wide variety of methods have been developed to recharge groundwater, one of which is aquifer storage and recovery (ASR). One of the fundamental...