Advances in Nonlinear Speech Processing

We present in this volume a collection of revised selected papers from the ISCA Tutorial and Research Workshop on Nonlinear Speech Processing (NO- LISP 2007)held in Paris,France, 2225May,2007.NOLISP 2007was organized by the University Pierre and Marie Curie (UPMC) with the generous support of ISCA (International Speech Communication Association), EURASIP and the IEEE. NOLISP 2007 was the ?rst follow-on workshop to a series of three e- lier events related to nonlinear speech processing, that were organized within the framework of the European COST action 277 Nonlinear speech processing (20012005). The ?nancial support of ISCA enabled the attendance of leading researchers from various parts of the world. Theexciting?eldofspeechprocessinghaswitnessedtremendousdevelopment over the past 20 years or so, thanks to both technological progress and to the growing focus of research on a number of key application areas. However, some speci?cities of the speech signal are still not well addressed by the currently availablemodels. Hence, new nonconventionalmodels and processing techniques needtobeinvestigatedinordertofosterand/oraccompanyfutureprogress,even if they do not match immediately the level of performance and understanding of the current state-of-the-art approaches. ThepurposeofNOLISPistopresentanddiscussnovelideas,workandresults related to such alternative techniques for speech processing, which depart from mainstream approaches. It declared its intent to be an interdisciplinary forum, intertwining research in di?erent ?elds of speech processing with its growing applications in everyday practice.

Klappentext

This book constitutes the thoroughly refereed postproceedings of the International Conference on Non-Linear Speech Processing, NOLISP 2007, held in Paris, France, in May 2007.

The 24 revised full papers presented were carefully reviewed and selected from numerous submissions for inclusion in the book. The papers are organized in topical sections on nonlinear and nonconventional techniques, speech synthesis, speaker recognition, speech recognition, speech analysis, and exploitation of non-linear techniques.

Inhalt
Non-Linear and Non-Conventional Techniques.- Phase-Based Methods for Voice Source Analysis.- Some Experiments in Audio-Visual Speech Processing.- Exploiting Nonlinearity in Adaptive Signal Processing.- Speech Synthesis.- Mixing HMM-Based Spanish Speech Synthesis with a CBR for Prosody Estimation.- Objective and Subjective Evaluation of an Expressive Speech Corpus.- Speaker Recognition.- On the Usefulness of Linear and Nonlinear Prediction Residual Signals for Speaker Recognition.- Multi Filter Bank Approach for Speaker Verification Based on Genetic Algorithm.- Speaker Recognition Via Nonlinear Phonetic- and Speaker-Discriminative Features.- Perceptron-Based Class Verification.- Speech Recognition.- Manifold Learning-Based Feature Transformation for Phone Classification.- Word Recognition with a Hierarchical Neural Network.- Hybrid Models for Automatic Speech Recognition: A Comparison of Classical ANN and Kernel Based Methods.- Towards Phonetically-Driven Hidden Markov Models: Can We Incorporate Phonetic Landmarks in HMM-Based ASR?.- A Hybrid Genetic-Neural Front-End Extension for Robust Speech Recognition over Telephone Lines.- Efficient Viterbi Algorithms for Lexical Tree Based Models.- Speech Analysis.- Non-stationary Self-consistent Acoustic Objects as Atoms of Voiced Speech.- The Hartley Phase Cepstrum as a Tool for Signal Analysis.- Voiced Speech Analysis by Empirical Mode Decomposition.- Estimation of Glottal Closure Instances from Speech Signals by Weighted Nonlinear Prediction.- Quantitative Perceptual Separation of Two Kinds of Degradation in Speech Denoising Applications.- Exploitation of non-linear techniques.- An Efficient VAD Based on a Generalized Gaussian PDF.- Estimating the Dispersion of the Biometric Glottal Signature in Continuous Speech.- TrajectoryMixture Density Networks with Multiple Mixtures for Acoustic-Articulatory Inversion.- Application of Feature Subset Selection Based on Evolutionary Algorithms for Automatic Emotion Recognition in Speech.