Hasil Pencarian  ::  Simpan CSV :: Kembali

"This book provides various speech enhancement algorithms for digital hearing aids. It covers information on noise signals extracted from silences of speech signal. The description of the algorithm used for this purpose is also provided. Different types of adaptive filters such as Least Mean Squares (LMS), Normalized LMS (NLMS) and Recursive Lease Squares (RLS) are described for noise reduction in the speech signals. Different types of noises are taken to generate noisy speech signals, and therefore information on various noises signals is provided. The comparative performance of various adaptive filters for noise reduction in speech signals is also described. In addition, the book provides a speech enhancement technique using adaptive filtering and necessary frequency strength enhancement using wavelet transform as per the requirement of audiogram for digital hearing aids.- Presents speech enhancement techniques for improving performance of digital hearing aids;- Covers various types of adaptive filters and their advantages and limitations;- Provides a hybrid speech enhancement technique using wavelet transform and adaptive filters."

"Ada tiga tahap utama dalam pengenalan ucapan yaitu tahap segmentasi untuk menemukan batas-batas fonem, tahap ekstraksi untuk mendapatkan nilai atas segmen-segmen sinyal dan tahap klasifikasi untuk pembandingan nilai segmen dengan nilai acuan. Di antara ketiganya, segmentasi merupakan tahap yang paling menentukan, karena jika keluarannya salah maka dua tahap beriktnya menjadi sia-sia.Berbagai metode dipergunakan dalam teknik segmentasi. Salah satunya Adapted Local Trigonometric Transforms (ALIT) yang mendekomposisi sinyal suara menjadi partisi-partisi trigonometris dan orthonormal dengan pemanfaatan DCT (Discrete Cosine Transform) yang memang memiliki keunggulan dalam hal kompresi sinyal [7]. Meski demikian, teknik ini juga memiliki kekurangan yaitu sensitif terhadap pergeseran sinyal ucapan.Teknik lain yang lebih tahan terhadap pergeseran sinyal ucapan dan bahkan derau adalah Parametric Filtering (PF) [8]. Teknik ini menggabungkan parametric filter bank dengan sebuah analisa otokorelasi selisih-satu (lag-one autocorrelation) untuk menghasilkan fungsi karakterisasi yang baru sama sekali dan tahan terhadap perubahan frekuensi dan derau. Otokorelasi selisih-satu merupakan korelasi sinyal terhadap sinyal itu sendiri setelah mengalami penundaan sebesar satu sekuensi. Fungsi ini mampu menampilkan dan menangkap perubahan spektral dan energi walau dalam domain waktu.Sinyal di-filter dalam sebuah filter bank yang berupa Parametric HR filter all-pole orde satu, lalu dikolerasikan dan didemodulasi untuk diambil yang diperlukan. Sambil dilakukan pengukuran distorsi untuk mencari nilai batas segmen, sinyal dapat ditampilkan sehingga mempermudah analisa. Tesis ini dilakukan dengan mensegmentasikan suku-kata bahasa Indonesia secara hampir real time dengan penundaan maksimum 10 detik.Analisa dilakukan terhadap performansi sistem dari sisi akurasi, waktu proses dan stabilitas. Diharapkan ini akan menjadi langkah awal yang tepat bagi tahap-tahap pengenalan ucapan selanjutnya dalam merintis sistem Voice To Text yang sesuai untuk suku-kata bahasa Indonesia.

---

There are three major steps in speech recognition which are accordingly segmentation for end-point detection, extraction to get the values upon the segmented signals and classification that yielding the corresponding references. Among the three, segmentation is the most important step since once it's output is incorrect then the next two will be nothing.There are methods used in this step. Among others are the so-called Adapted Local Trigonometric Transforms (ALIT} which decomposite signals into orthonormal-trigonometrical parts using Discrete Cosine Transform (DC?) that is good in signal compressing [7]. Anyway, this method lacks of sensitivity over unstable signals.Another method that is robust to such sigwls as well as noise is Parametric Filtering (PF) [8]. This method combines a parametric filter bank and a lag-one autocorrelation analysis to produce a new characterisation function that overcome the frequency shift and noise. Lag-one autocorrelation means correlating the signal toward itself after one-sequence delayed. It also displays and cacthes spectral and energy changes even in time domain.Signal is filtered in a filter banks which are all-pole first order parametric IIR filter, after which it is correlated by lag-one autocorrelation. Next, signal is demodulated to yield what is needed. While being processed by distortion measures to find end-points, the signal can be displayed to make the analysis visible. Research done by segmenting Indonesia syllabels in almost real time with ten-second maximum delay.There are analysis upon system performances in accusation, time consuming and stability point of view. Hopefully, this will become fisrt step for the next speech recognition steps and lead to the suitable VoiceToText system for Indonesia syllabels."

"This book is designed to serve as a core text for courses in advanced engineering mathematics required by many engineering departments. The style of presentation is such that the student, with a minimum of assistance, can follow the step-by-step derivations. Liberal use of examples and homework prob­lems aid the student in the study of the topics presented. Ordinary differential equations, including a number of physical applica­tions, are reviewed in Chapter One. The use of series methods are presented in Chapter Two, Subsequent chapters present Laplace transforms, matrix theory and applications, vector analysis, Fourier series and transforms, partial differential equations, numerical methods using finite differences, complex vari­ables, and wavelets. The material is presented so that four or five subjects can be covered in a single course, depending on the topics chosen and the completeness of coverage.Incorporated in this textbook is the use of certain computer software packages. Short tutorials on Maple, demonstrating how problems in engineering mathematics can be solved with a computer algebra system, are included in most sections of the text. Problems have been identified at the end of sections to be solved specifically with Maple, and there are computer laboratory activities, which are more difficult problems designed for Maple. In addition, MATLAB and Excel have been included in the solution of problems in several of the chapters.There is a solutions manual available for those who select the text for their course. This text can be used in two semesters of engineering mathematics. The many helpful features make the text relatively easy to use in the classroom."

"This monograph presents teaching material in the field of differential equations while addressing applications and topics in electrical and biomedical engineering primarily. The book contains problems with varying levels of difficulty, including Matlab simulations. The target audience comprises advanced undergraduate and graduate students as well as lecturers, but the book may also be beneficial for practicing engineers alike."

"This book introduces medical imaging, its security requirements, and various security mechanisms using data hiding approaches. The book in particular provides medical data hiding techniques using various advanced image transforms and encryption methods. The book focuses on two types of data hiding techniques: steganography and watermarking for medical images. The authors show how these techniques are used for security and integrity verification of medical images and designed for various types of medical images such as grayscale image and color image. The implementation of techniques are done using discrete cosine transform (DCT), discrete wavelet transform (DWT), singular value decomposition (SVD), redundant DWT (RDWT), fast discrete curvelet transform (FDCuT), finite ridgelet transform (FRT) and non-subsampled contourlet transform (NSCT). The results of these techniques are also demonstrated after description of each technique. Finally, some future research directions are provided for security of medical images in telemedicine application."