Photochemical stability of cellulose textile surfaces painted with some reactive azo-triazine dyes

The influence of ultraviolet irradiation of different doses (λ > 300 nm) on the structural and color modifications of cotton fabrics painted with four different azo-triazine dyes (Reactive Yellow 143, Reactive Orange 13, Reactive Red 183, and Reactive Red 2) was studied. High irradiation doses up to 3500 J cm^?2 led to changes in the dyes structures. Structural changes before and after the complete irradiation were compared by applying FTIR, UV–Vis, and near infrared chemical imaging techniques. Color modifications were also investigated. Color differences significantly increased with the irradiation dose for all the studied samples.     

Pitch Estimation by Block and Instantaneous Methods

Choosing a pitch estimation algorithm is not a simple task. One must balance between the accuracy and the reliability of the estimates. Two classes of methods are available. The first one, known as the block methods class, gives noise robust solutions and has an intrinsic averaging property, but is not very accurate, especially for the transition regions. The second one, known as the instantaneous (or event-based) methods class, gives very accurate estimates, but is considered to be inadequate in the presence of noise.In this paper, we present potential enhancements of the performance in pitch estimation, based on both block and instantaneous methods. In this respect we discuss mainly two algorithms: a nonlinear cepstral algorithm and a wavelet-based one. The first algorithm, due to the proposed nonlinear model, enhances the classical linear model performance related to the accuracy of the estimated pitch for the transition regions and to the robustness in the presence of noise. Concerning the second algorithm, to the inherent accuracy of the estimated pitch, we add robust estimates even in the presence of noise, based on the multiresolution properties of an improved wavelet transform. The obtained enhancements were evaluated on a hand-labeled speech database, and the improved algorithms are now being applied in our research concerning speech compression and prosody.     

Statistical and Hybrid Methods for Speech Recognition in Romanian

The present paper describes the evolution of our work concerning the problem of speech recognition. Beginning with a classical hidden Markov model (HMM), we have investigated two ways to improve the performance of this basic structure. The first way was to realize a neuro-statistical hybrid by integrating a multilayer perceptron (MLP) as a posteriori probability estimator. The system was further refined by adding supplementary discriminative training (DT) based on the minimum classification error (MCE). Tests performed on a 15,000 isolated spoken-word database, showed an increase in the recognition rate from 92.2% for the HMM-based recognition system, to 94.7% for the HMM-MLP system, and then to 98.1% for the refined HMM-MLP-DT system. The second way to improve the classical HMM was to build a fuzzy-statistical hybrid, FHMM, based on a fuzzy similarity measure instead of the probabilistic measure specific to the usual statistical model. The benefits of the fuzzy measure introduction were evaluated on a vowel recognition task, and a decrease of approximately 3% in the error rate is reported.